fix(ingestion): an unlistable brain root is a scan error, not clean

PR #159 round 2: with the brain root at mode 0100 (enterable, not
listable) WalkDir failed at the root, Scan discarded the error and
reported clean, and startup passed while search, which goes straight
into wiki/, served a planted symlink. Scan now returns the root's walk
error, so startup refuses on anything but a missing root. Also pins
that a cleared set is reported again when it reappears.

Refs #157

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

ingestion/internal/api/handler.go

@@ -13,6 +13,7 @@ import (
 
 	"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/claim"
+	"github.com/mathiasbq/hyperguild/ingestion/internal/extract"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/pipeline"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/search"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/vectorstore"
@@ -480,7 +481,7 @@ func (h *Handler) IngestPath(w http.ResponseWriter, r *http.Request) {
 				allWarnings = append(allWarnings, fmt.Sprintf("skipped %s: %v", path, rerr))
 				return nil
 			}
-			content, readErr := ReadIngestFile(h.brainDir, path)
+			content, readErr := extract.Text(path)
 			if readErr != nil {
 				allWarnings = append(allWarnings, fmt.Sprintf("extract %s: %v", path, readErr))
 				return nil
@@ -509,7 +510,7 @@ func (h *Handler) IngestPath(w http.ResponseWriter, r *http.Request) {
 			writeError(w, http.StatusBadRequest, fmt.Sprintf("unsupported file extension: %s", ext))
 			return
 		}
-		content, readErr := ReadIngestFile(h.brainDir, req.Path)
+		content, readErr := extract.Text(req.Path)
 		if readErr != nil {
 			writeError(w, http.StatusInternalServerError, fmt.Sprintf("extract text: %v", readErr))
 			return

ingestion/internal/api/ingestpath_root.go

@@ -2,9 +2,6 @@ package api
 
 import (
 	"errors"
-	"os"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/extract"
 	"path/filepath"
 	"strings"
 )
@@ -22,11 +19,6 @@ func ResolveIngestPath(brainDir, p string) (string, error) {
 	if strings.TrimSpace(p) == "" {
 		return "", ErrOutsideIngestRoot
 	}
-	// raw/ itself must be a real directory: if it were a symlink, containment
-	// inside it would mean nothing (hyperguild#153 item 4).
-	if fi, lerr := os.Lstat(filepath.Join(brainDir, "raw")); lerr != nil || fi.Mode()&os.ModeSymlink != 0 || !fi.IsDir() {
-		return "", ErrOutsideIngestRoot
-	}
 	root, err := filepath.EvalSymlinks(filepath.Join(brainDir, "raw"))
 	if err != nil {
 		return "", ErrOutsideIngestRoot
@@ -44,24 +36,3 @@ func ResolveIngestPath(brainDir, p string) (string, error) {
 	}
 	return real, nil
 }
-
-// ReadIngestFile extracts text from an ingest path already confined by
-// ResolveIngestPath, reading it through the raw/ root rather than reopening
-// the path by name (hyperguild#153 item 3): if the file is swapped for a
-// symlink after the check, the read still cannot leave raw/.
-func ReadIngestFile(brainDir, resolved string) (string, error) {
-	root, rawPath, err := OpenRawRoot(brainDir)
-	if err != nil {
-		return "", ErrOutsideIngestRoot
-	}
-	defer func() { _ = root.Close() }()
-	realRoot, err := filepath.EvalSymlinks(rawPath)
-	if err != nil {
-		return "", ErrOutsideIngestRoot
-	}
-	rel, err := filepath.Rel(realRoot, resolved)
-	if err != nil || rel == ".." || strings.HasPrefix(rel, ".."+string(filepath.Separator)) || filepath.IsAbs(rel) {
-		return "", ErrOutsideIngestRoot
-	}
-	return extract.TextIn(root, rel)
-}

ingestion/internal/api/promote.go

@@ -1,9 +1,7 @@
 package api
 
 import (
-	"errors"
 	"fmt"
-	"io/fs"
 	"os"
 	"path/filepath"
 	"regexp"
@@ -30,25 +28,25 @@ var datePrefix = regexp.MustCompile(`^\d{4}-\d{2}-\d{2}-`)
 // (natural review order). An absent raw/ dir yields an empty slice, not an
 // error. Only .md files are listed; tunnel-candidate files are skipped.
 func ListPending(brainDir string) ([]PendingNote, error) {
-	root, _, err := OpenRawRoot(brainDir)
+	dir := filepath.Join(brainDir, "raw")
+	entries, err := os.ReadDir(dir)
 	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
+		if os.IsNotExist(err) {
 			return []PendingNote{}, nil
 		}
 		return nil, fmt.Errorf("read raw dir: %w", err)
 	}
-	defer func() { _ = root.Close() }()
-	entries, err := fs.ReadDir(root.FS(), ".")
-	if err != nil {
-		return nil, fmt.Errorf("read raw dir: %w", err)
-	}
 
 	out := make([]PendingNote, 0, len(entries))
 	for _, e := range entries {
-		if !e.Type().IsRegular() || !strings.HasSuffix(e.Name(), ".md") || strings.HasPrefix(e.Name(), "tunnel-candidates-") {
-			continue // directories, and symlinks whose target may be anywhere (#153)
+		if e.IsDir() || !strings.HasSuffix(e.Name(), ".md") || strings.HasPrefix(e.Name(), "tunnel-candidates-") {
+			continue
 		}
-		raw, info, readErr := readRawNote(root, e.Name())
+		info, statErr := e.Info()
+		if statErr != nil {
+			continue
+		}
+		raw, readErr := os.ReadFile(filepath.Join(dir, e.Name()))
 		if readErr != nil {
 			continue
 		}
@@ -107,9 +105,12 @@ func PromoteNote(brainDir string, opts PromoteOptions) (string, error) {
 	}
 
 	src := filepath.Join(brainDir, "raw", base)
-	raw, err := readRawSource(brainDir, base)
+	raw, err := os.ReadFile(src)
 	if err != nil {
-		return "", err
+		if os.IsNotExist(err) {
+			return "", fmt.Errorf("pending note %q does not exist in raw/", base)
+		}
+		return "", fmt.Errorf("read source: %w", err)
 	}
 
 	// Collision: never silently overwrite an existing promoted note.
@@ -179,9 +180,12 @@ func RejectNote(brainDir string, opts RejectOptions) (string, error) {
 	}
 
 	src := filepath.Join(brainDir, "raw", base)
-	raw, err := readRawSource(brainDir, base)
+	raw, err := os.ReadFile(src)
 	if err != nil {
-		return "", err
+		if os.IsNotExist(err) {
+			return "", fmt.Errorf("pending note %q does not exist in raw/", base)
+		}
+		return "", fmt.Errorf("read source: %w", err)
 	}
 
 	dest := filepath.Join(brainDir, "raw", "rejected", base)
@@ -220,24 +224,3 @@ func excerpt(s string, n int) string {
 	}
 	return strings.TrimSpace(string(r))
 }
-
-// readRawSource reads a pending note for promote/reject through the raw root,
-// refusing symlinks and a symlinked raw/ (hyperguild#153).
-func readRawSource(brainDir, base string) ([]byte, error) {
-	root, _, err := OpenRawRoot(brainDir)
-	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
-			return nil, fmt.Errorf("pending note %q does not exist in raw/", base)
-		}
-		return nil, fmt.Errorf("read source: %w", err)
-	}
-	defer func() { _ = root.Close() }()
-	raw, _, err := readRawNote(root, base)
-	if err != nil {
-		if errors.Is(err, os.ErrNotExist) {
-			return nil, fmt.Errorf("pending note %q does not exist in raw/", base)
-		}
-		return nil, fmt.Errorf("read source: %w", err)
-	}
-	return raw, nil
-}

ingestion/internal/api/raw_race_test.go

@@ -1,178 +0,0 @@
-package api_test
-
-import (
-	"encoding/base64"
-	"encoding/json"
-	"net/http"
-	"net/http/httptest"
-	"os"
-	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/api"
-)
-
-// hyperguild#153 / PR #156: behavioural race tests, so the control does not
-// rest on lints that list forbidden APIs (round 3: an os.ReadFile or
-// Root.ReadFile rewrite passed the lint and leaked 19k times in 4s). A
-// swapper keeps renaming a regular note, a symlink to an outside file, or a
-// FIFO into raw/r.md while the public readers loop. Any leaked byte fails;
-// any hang fails.
-
-const raceSentinel = "SENTINEL-153-RACE"
-
-// startSwapper runs fn until the test ends, and waits for it to exit before
-// t.TempDir's cleanup removes the directory it writes into (hyperguild#162:
-// an unawaited swapper raced the cleanup, "directory not empty").
-func startSwapper(t *testing.T, fn func(stop *atomic.Bool)) {
-	t.Helper()
-	var stop atomic.Bool
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() { defer wg.Done(); fn(&stop) }()
-	t.Cleanup(func() { stop.Store(true); wg.Wait() })
-}
-
-func swap(raw, outside string, fifo bool, stop *atomic.Bool) {
-	tgt := filepath.Join(raw, "r.md")
-	for i := 0; !stop.Load(); i++ {
-		tmp := filepath.Join(raw, ".swap")
-		_ = os.Remove(tmp)
-		switch {
-		case i%2 == 0:
-			_ = os.WriteFile(tmp, []byte("# r\n\ninside"), 0o644)
-		case fifo:
-			_ = syscall.Mkfifo(tmp, 0o644)
-		default:
-			_ = os.Symlink(outside, tmp)
-		}
-		_ = os.Rename(tmp, tgt)
-	}
-}
-
-func raceBrain(t *testing.T) (brain, raw, outside string) {
-	t.Helper()
-	brain = t.TempDir()
-	raw = filepath.Join(brain, "raw")
-	if err := os.MkdirAll(raw, 0o755); err != nil {
-		t.Fatal(err)
-	}
-	outside = filepath.Join(t.TempDir(), "secret.md")
-	if err := os.WriteFile(outside, []byte("# s\n\n"+raceSentinel), 0o644); err != nil {
-		t.Fatal(err)
-	}
-	if err := os.WriteFile(filepath.Join(raw, "r.md"), []byte("# r\n\ninside"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-	return brain, raw, outside
-}
-
-// readers runs every public raw/ reader once and returns all text they
-// produced (excerpts, extracted text).
-func readers(brain, raw string) string {
-	var b strings.Builder
-	if ns, err := api.ListPending(brain); err == nil {
-		for _, n := range ns {
-			b.WriteString(n.Excerpt)
-		}
-	}
-	if s, err := api.ReadIngestFile(brain, filepath.Join(raw, "r.md")); err == nil {
-		b.WriteString(s)
-	}
-	return b.String()
-}
-
-func TestRawReadersLeakNothingUnderASymlinkSwapRace(t *testing.T) {
-	brain, raw, outside := raceBrain(t)
-	startSwapper(t, func(stop *atomic.Bool) { swap(raw, outside, false, stop) })
-	leaks, iters := 0, 0
-	for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); iters++ {
-		leaks += strings.Count(readers(brain, raw), raceSentinel)
-	}
-	if leaks > 0 {
-		t.Fatalf("outside file leaked %d times in %d iterations", leaks, iters)
-	}
-	t.Logf("%d iterations, 0 leaks", iters)
-}
-
-func TestRawReadersDoNotHangUnderAFIFOSwapRace(t *testing.T) {
-	brain, raw, outside := raceBrain(t)
-	startSwapper(t, func(stop *atomic.Bool) { swap(raw, outside, true, stop) })
-	var iters atomic.Int64
-	done := make(chan struct{})
-	go func() {
-		for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
-			_ = readers(brain, raw)
-			iters.Add(1)
-		}
-		close(done)
-	}()
-	select {
-	case <-done:
-	case <-time.After(6 * time.Second):
-		// Unblock the stuck open so the test process can exit.
-		if f, err := os.OpenFile(filepath.Join(raw, "r.md"), os.O_WRONLY|syscall.O_NONBLOCK, 0); err == nil {
-			_ = f.Close()
-		}
-		t.Fatalf("a raw/ reader hung on a swapped-in FIFO after %d iterations", iters.Load())
-	}
-}
-
-// PR #156 round 4: the pdf/docmark branch of TextIn copies the file through
-// the root to a temp file; pointing the extractor back at the raw/ path by
-// name stayed green because the race tests only used r.md (a stub docmark
-// then received the outside file's bytes 34k times in 4s). This races a
-// symlink into raw/r.docx and has a fake docmark count the sentinel in every
-// document it is sent.
-func TestDocmarkExtractionLeaksNothingUnderASymlinkSwapRace(t *testing.T) {
-	var seen atomic.Int64
-	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
-		var req struct {
-			Params struct {
-				Arguments struct {
-					ContentBase64 string `json:"content_base64"`
-				} `json:"arguments"`
-			} `json:"params"`
-		}
-		_ = json.NewDecoder(r.Body).Decode(&req)
-		if b, err := base64.StdEncoding.DecodeString(req.Params.Arguments.ContentBase64); err == nil &&
-			strings.Contains(string(b), raceSentinel) {
-			seen.Add(1)
-		}
-		w.Header().Set("Content-Type", "application/json")
-		_, _ = w.Write([]byte(`{"jsonrpc":"2.0","id":1,"result":{"content":[{"type":"text","text":"converted"}]}}`))
-	}))
-	defer srv.Close()
-	t.Setenv("DOCMARK_URL", srv.URL)
-
-	brain, raw, outside := raceBrain(t)
-	tgt := filepath.Join(raw, "r.docx")
-	if err := os.WriteFile(tgt, []byte("inside docx bytes"), 0o644); err != nil {
-		t.Fatal(err)
-	}
-	startSwapper(t, func(stop *atomic.Bool) {
-		for i := 0; !stop.Load(); i++ {
-			tmp := filepath.Join(raw, ".swapdocx")
-			_ = os.Remove(tmp)
-			if i%2 == 0 {
-				_ = os.WriteFile(tmp, []byte("inside docx bytes"), 0o644)
-			} else {
-				_ = os.Symlink(outside, tmp)
-			}
-			_ = os.Rename(tmp, tgt)
-		}
-	})
-	iters := 0
-	for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); iters++ {
-		_, _ = api.ReadIngestFile(brain, tgt)
-	}
-	if n := seen.Load(); n > 0 {
-		t.Fatalf("docmark received the outside file %d times in %d iterations", n, iters)
-	}
-	t.Logf("%d iterations, 0 leaks to docmark", iters)
-}

ingestion/internal/api/raw_symlink_test.go

@@ -1,80 +0,0 @@
-package api_test
-
-import (
-	"os"
-	"path/filepath"
-	"strings"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/api"
-)
-
-// hyperguild#153: readers of brain/raw/ followed symlinks. /pending showed the
-// target's contents and promote would copy it into wiki/ (and brain-sync to
-// Gitea). raw/ is writable via Syncthing, so a symlink there is reachable
-// input. Every raw/ reader now treats a symlink as not-a-note.
-const sentinel153 = "SENTINEL-153-OUTSIDE"
-
-func rawWithSymlink(t *testing.T) (brain string) {
-	t.Helper()
-	brain = t.TempDir()
-	raw := filepath.Join(brain, "raw")
-	require.NoError(t, os.MkdirAll(raw, 0o755))
-	outside := filepath.Join(t.TempDir(), "secret.md")
-	require.NoError(t, os.WriteFile(outside, []byte("---\ncreated_at: 2026-01-01T00:00:00Z\n---\n# s\n\n"+sentinel153), 0o644))
-	require.NoError(t, os.Symlink(outside, filepath.Join(raw, "2026-01-01-leak.md")))
-	require.NoError(t, os.WriteFile(filepath.Join(raw, "2026-01-02-real.md"), []byte("# real\n\nreal body"), 0o644))
-	return brain
-}
-
-func TestListPendingSkipsSymlinkedNotes(t *testing.T) {
-	notes, err := api.ListPending(rawWithSymlink(t))
-	require.NoError(t, err)
-	var names []string
-	for _, n := range notes {
-		names = append(names, n.Filename)
-		assert.NotContains(t, n.Excerpt, sentinel153)
-	}
-	assert.Equal(t, []string{"2026-01-02-real.md"}, names)
-}
-
-func TestPromoteNoteRefusesASymlinkedSource(t *testing.T) {
-	brain := rawWithSymlink(t)
-	_, err := api.PromoteNote(brain, api.PromoteOptions{Filename: "2026-01-01-leak.md", Wing: "testing", Hall: "facts"})
-	require.Error(t, err)
-	_, statErr := os.Stat(filepath.Join(brain, "wiki", "testing", "facts", "leak.md"))
-	assert.True(t, os.IsNotExist(statErr), "nothing may be promoted from a symlink")
-}
-
-func TestRejectNoteRefusesASymlinkedSource(t *testing.T) {
-	brain := rawWithSymlink(t)
-	_, err := api.RejectNote(brain, api.RejectOptions{Filename: "2026-01-01-leak.md"})
-	require.Error(t, err)
-	b, _ := os.ReadFile(filepath.Join(brain, "raw", "rejected", "2026-01-01-leak.md"))
-	assert.False(t, strings.Contains(string(b), sentinel153))
-}
-
-// Item 4: if raw/ itself is a symlink, containment inside it means nothing.
-func TestResolvePendingPromoteRejectAndReadIngestRefuseASymlinkedRawRoot(t *testing.T) {
-	brain := t.TempDir()
-	elsewhere := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(elsewhere, "x.md"), []byte("# x\n\n"+sentinel153), 0o644))
-	require.NoError(t, os.Symlink(elsewhere, filepath.Join(brain, "raw")))
-
-	_, err := api.ResolveIngestPath(brain, "x.md")
-	assert.Error(t, err, "ingest must refuse a symlinked raw/")
-	notes, err := api.ListPending(brain)
-	assert.Error(t, err, "pending must refuse a symlinked raw/")
-	assert.Empty(t, notes)
-	_, err = api.PromoteNote(brain, api.PromoteOptions{Filename: "x.md", Wing: "testing", Hall: "facts"})
-	assert.Error(t, err, "promote must refuse a symlinked raw/")
-	_, err = api.RejectNote(brain, api.RejectOptions{Filename: "x.md"})
-	assert.Error(t, err, "reject must refuse a symlinked raw/")
-	text, err := api.ReadIngestFile(brain, filepath.Join(elsewhere, "x.md"))
-	assert.Error(t, err, "ReadIngestFile must refuse a symlinked raw/")
-	assert.NotContains(t, text, sentinel153)
-	// The watcher's refusal is tested in internal/watcher.
-}

ingestion/internal/api/rawroot.go

@@ -1,67 +0,0 @@
-package api
-
-import (
-	"errors"
-	"fmt"
-	"io"
-	"os"
-	"path/filepath"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/extract"
-)
-
-// errNotANote is returned for a raw/ entry that is not a regular file — in
-// particular a symlink, whose target may be anywhere (hyperguild#153).
-var errNotANote = errors.New("not a regular note file in raw/ (symlinks are refused)")
-
-// errRawNotDir means brain/raw is not a real directory (e.g. a symlink), so
-// containment inside it would mean nothing (hyperguild#153 item 4).
-var errRawNotDir = errors.New("brain/raw is not a real directory")
-
-// OpenRawRoot opens brain/raw as an os.Root, refusing a raw/ that is itself a
-// symlink. Reads through the root cannot leave it, even if an entry is
-// swapped for a symlink between a check and the read. It returns the root
-// and raw/'s path. A missing raw/ returns an error wrapping os.ErrNotExist.
-func OpenRawRoot(brainDir string) (*os.Root, string, error) {
-	p := filepath.Join(brainDir, "raw")
-	fi, err := os.Lstat(p)
-	if err != nil {
-		return nil, "", err
-	}
-	if fi.Mode()&os.ModeSymlink != 0 || !fi.IsDir() {
-		return nil, "", errRawNotDir
-	}
-	r, err := os.OpenRoot(p)
-	if err != nil {
-		return nil, "", fmt.Errorf("open raw root: %w", err)
-	}
-	return r, p, nil
-}
-
-// readRawNote reads one raw/ entry by basename through the root, refusing
-// anything that is not a regular file.
-//
-// Accepted residual: a HARD link to an outside file is a regular file and is
-// read. Syncthing syncs contents, not links, so a peer cannot create one;
-// fs.protected_hardlinks=1 and the pod's secrets live on a different
-// filesystem from the brain volume, so a local caller cannot link them in.
-// Refusing Nlink > 1 is the next step if either assumption changes.
-func readRawNote(r *os.Root, base string) ([]byte, os.FileInfo, error) {
-	fi, err := r.Lstat(base)
-	if err != nil {
-		return nil, nil, err
-	}
-	if !fi.Mode().IsRegular() {
-		return nil, nil, errNotANote
-	}
-	f, _, err := extract.OpenRegular(r, base)
-	if err != nil {
-		return nil, nil, err
-	}
-	defer func() { _ = f.Close() }()
-	b, err := io.ReadAll(f)
-	if err != nil {
-		return nil, nil, err
-	}
-	return b, fi, nil
-}

ingestion/internal/api/readnote_root_test.go

@@ -1,43 +0,0 @@
-package api_test
-
-import (
-	"os"
-	"path/filepath"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/api"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
-)
-
-// hyperguild#157 slice 2: ReadNote (brain_get, read-after-write) read by
-// path; it now reads through brainfs.
-func TestReadNoteRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
-	brain := t.TempDir()
-	facts := filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(facts, "leak.md")))
-	_, body, _, err := api.ReadNote(brain, "wiki/w/facts/leak.md")
-	assert.Error(t, err)
-	assert.NotContains(t, body, brainfstest.Sentinel)
-
-	p := filepath.Join(facts, "r.md")
-	require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644))
-	fm, body, _, err := api.ReadNote(brain, "wiki/w/facts/r.md")
-	require.NoError(t, err)
-	assert.Equal(t, "y", fm["x"])
-	assert.Contains(t, body, "inside")
-	read := func() string {
-		fm, body, _, _ := api.ReadNote(brain, "wiki/w/facts/r.md")
-		s := body
-		for k, v := range fm {
-			s += k + v
-		}
-		return s
-	}
-	brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, read)
-	brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() { _ = read() })
-}

ingestion/internal/api/update.go

@@ -4,7 +4,6 @@ import (
 	"crypto/sha256"
 	"encoding/hex"
 	"fmt"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
 	"os"
 	"path/filepath"
 	"strings"
@@ -175,17 +174,11 @@ func UpdateNote(brainDir string, opts UpdateNoteOptions) (relPath, contentHash,
 // write primitive behind brain_get: the hash it returns equals the hash
 // brain_write / brain_update returned for the same bytes.
 func ReadNote(brainDir, relPath string) (fm map[string]string, body, contentHash string, err error) {
-	rel, _, err := resolveWithin(brainDir, relPath)
+	_, abs, err := resolveWithin(brainDir, relPath)
 	if err != nil {
 		return nil, "", "", err
 	}
-	// Through the brain root, regular files only (hyperguild#157).
-	bfs, err := brainfs.Open(brainDir)
-	if err != nil {
-		return nil, "", "", fmt.Errorf("read note: %w", err)
-	}
-	defer func() { _ = bfs.Close() }()
-	raw, err := bfs.ReadRegular(rel)
+	raw, err := os.ReadFile(abs)
 	if err != nil {
 		if os.IsNotExist(err) {
 			return nil, "", "", fmt.Errorf("note %q does not exist", relPath)

ingestion/internal/audit/ntfy.go

@@ -49,19 +49,7 @@ func (n *NtfyNotifier) Notify(ctx context.Context, msg string) error {
 	}
 	defer func() { _ = resp.Body.Close() }()
 	if resp.StatusCode < 200 || resp.StatusCode >= 300 {
-		return &RejectedError{Status: resp.StatusCode}
+		return fmt.Errorf("ntfy notify: status %d", resp.StatusCode)
 	}
 	return nil
 }
-
-// RejectedError is a non-2xx ntfy response.
-type RejectedError struct{ Status int }
-
-func (e *RejectedError) Error() string { return fmt.Sprintf("ntfy notify: status %d", e.Status) }
-
-// Permanent reports whether resending the same message cannot succeed: a 4xx
-// other than 429 (rate limited). Callers may drop such a message; anything
-// else (5xx, 429, transport errors) is an outage and must be retried.
-func (e *RejectedError) Permanent() bool {
-	return e.Status >= 400 && e.Status < 500 && e.Status != http.StatusTooManyRequests
-}

ingestion/internal/audit/ntfy_rejected_test.go

@@ -1,27 +0,0 @@
-package audit
-
-import (
-	"context"
-	"errors"
-	"net/http"
-	"net/http/httptest"
-	"testing"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// A non-2xx ntfy response is a *RejectedError, and only a client-side refusal
-// other than rate limiting is permanent: callers may drop a message on a
-// permanent rejection and must retry the rest (hyperguild#157, PR #160).
-func TestNotifyClassifiesRejectionsAsPermanentOrNot(t *testing.T) {
-	for status, permanent := range map[int]bool{413: true, 400: true, 403: true, 429: false, 500: false, 503: false} {
-		srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { w.WriteHeader(status) }))
-		err := NewNtfyNotifier(srv.URL, "").Notify(context.Background(), "m")
-		srv.Close()
-		var rej *RejectedError
-		require.True(t, errors.As(err, &rej), "status %d: %v", status, err)
-		assert.Equal(t, status, rej.Status)
-		assert.Equal(t, permanent, rej.Permanent(), "status %d", status)
-	}
-}

ingestion/internal/brainfs/brainfs.go

@@ -1,153 +0,0 @@
-// Package brainfs reads the brain directory through an os.Root (hyperguild#157).
-// The brain dir is a Syncthing sendreceive folder and Syncthing syncs
-// symlinks; a reader that walks and reads by path follows a planted symlink
-// out of the brain, and blocks forever on a planted FIFO. Readers that go
-// through a Brain only ever read regular files inside the brain: os.Root
-// refuses to leave the root even if an entry is swapped mid-read, and every
-// open is non-blocking with the type checked on the opened handle.
-package brainfs
-
-import (
-	"errors"
-	"fmt"
-	"io"
-	"io/fs"
-	"os"
-	"path/filepath"
-	"syscall"
-)
-
-// ErrNotRegular is returned for an entry that is not a regular file.
-var ErrNotRegular = errors.New("not a regular file")
-
-// ErrTooLarge is returned by ReadRegularMax for a file over the limit.
-var ErrTooLarge = errors.New("file too large")
-
-// Brain is an open brain directory.
-type Brain struct {
-	root *os.Root
-	dir  string
-}
-
-// Open opens brainDir (which may itself be a mount symlink) as a root.
-func Open(brainDir string) (*Brain, error) {
-	r, err := os.OpenRoot(brainDir)
-	if err != nil {
-		return nil, fmt.Errorf("open brain root: %w", err)
-	}
-	return &Brain{root: r, dir: brainDir}, nil
-}
-
-// Close releases the root.
-func (b *Brain) Close() error { return b.root.Close() }
-
-// Rel converts a path under the brain dir to a root-relative path.
-func (b *Brain) Rel(path string) (string, error) {
-	rel, err := filepath.Rel(b.dir, path)
-	if err != nil {
-		return "", err
-	}
-	if rel == ".." || len(rel) > 2 && rel[:3] == ".."+string(filepath.Separator) {
-		return "", fmt.Errorf("%s is outside the brain dir", path)
-	}
-	return filepath.ToSlash(rel), nil
-}
-
-// ReadRegular reads a brain-relative path, refusing anything that is not a
-// regular file, both before the open and on the opened handle.
-func (b *Brain) ReadRegular(rel string) ([]byte, error) {
-	fi, err := b.root.Lstat(rel)
-	if err != nil {
-		return nil, err
-	}
-	if !fi.Mode().IsRegular() {
-		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
-	}
-	return b.readOpened(rel)
-}
-
-// readOpened opens rel non-blocking and reads it only if the OPENED handle is
-// a regular file. This is the check that matters under a race: an entry
-// swapped for a FIFO after the Lstat, with a writer holding it open, would
-// otherwise park io.ReadAll forever (PR #161 review: removing it hung search
-// after 16 iterations).
-//
-// Accepted residual, as in api/rawroot.go: a HARD link to an outside file is
-// a regular file and is read. Syncthing syncs contents, not links, so a peer
-// cannot create one; fs.protected_hardlinks=1 and the pod's secrets are on a
-// different filesystem from the brain volume, so a local caller cannot link
-// them in. Refusing Nlink > 1 is the next step if either changes.
-func (b *Brain) readOpened(rel string) ([]byte, error) { return b.readOpenedMax(rel, -1) }
-
-// ReadRegularMax is ReadRegular with a size limit, checked on the opened
-// handle before reading (max < 0 means no limit).
-func (b *Brain) ReadRegularMax(rel string, max int64) ([]byte, error) {
-	fi, err := b.root.Lstat(rel)
-	if err != nil {
-		return nil, err
-	}
-	if !fi.Mode().IsRegular() {
-		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
-	}
-	return b.readOpenedMax(rel, max)
-}
-
-func (b *Brain) readOpenedMax(rel string, max int64) ([]byte, error) {
-	f, err := b.root.OpenFile(rel, os.O_RDONLY|syscall.O_NONBLOCK, 0)
-	if err != nil {
-		return nil, err
-	}
-	defer func() { _ = f.Close() }()
-	hfi, err := f.Stat()
-	if err != nil {
-		return nil, err
-	}
-	if !hfi.Mode().IsRegular() {
-		return nil, fmt.Errorf("%s: %w", rel, ErrNotRegular)
-	}
-	if max >= 0 && hfi.Size() > max {
-		return nil, fmt.Errorf("%s is %d bytes (limit %d): %w", rel, hfi.Size(), max, ErrTooLarge)
-	}
-	var r io.Reader = f
-	if wrapAfterStat != nil {
-		r = wrapAfterStat(r)
-	}
-	if max < 0 {
-		return io.ReadAll(r)
-	}
-	// The handle's size can grow after the check; bound the read itself.
-	data, err := io.ReadAll(io.LimitReader(r, max+1))
-	if err != nil {
-		return nil, err
-	}
-	if int64(len(data)) > max {
-		return nil, fmt.Errorf("%s grew past %d bytes while read: %w", rel, max, ErrTooLarge)
-	}
-	return data, nil
-}
-
-// wrapAfterStat is a test seam: it runs between the size check and the read,
-// and wraps the reader the read consumes.
-var wrapAfterStat func(io.Reader) io.Reader
-
-// WalkRegular walks a brain-relative directory through the root and calls fn
-// for each regular file (directories are descended; symlinks, FIFOs and
-// devices are skipped, never followed). An entry error is passed to onErr and
-// the walk continues; a nil onErr ignores it.
-func (b *Brain) WalkRegular(relDir string, fn func(rel string) error, onErr func(rel string, err error)) error {
-	return fs.WalkDir(b.root.FS(), filepath.ToSlash(relDir), func(p string, d fs.DirEntry, err error) error {
-		if err != nil {
-			if onErr != nil {
-				onErr(p, err)
-			}
-			if d != nil && d.IsDir() {
-				return fs.SkipDir
-			}
-			return nil
-		}
-		if d.IsDir() || !d.Type().IsRegular() {
-			return nil
-		}
-		return fn(p)
-	})
-}

ingestion/internal/brainfs/brainfs_test.go

@@ -1,141 +0,0 @@
-package brainfs
-
-import (
-	"errors"
-	"io"
-	"os"
-	"path/filepath"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// The handle check is what stops a FIFO swapped in after the Lstat. A FIFO
-// with no writer reads EOF at once, so a race test could not see the check
-// missing; here a writer holds the FIFO open, as any local process can, and
-// readOpened must refuse it promptly (PR #161 review).
-func TestReadOpenedRefusesAFIFOHeldOpenByAWriter(t *testing.T) {
-	dir := t.TempDir()
-	fifo := filepath.Join(dir, "p.md")
-	require.NoError(t, syscall.Mkfifo(fifo, 0o644))
-	w, err := os.OpenFile(fifo, os.O_RDWR, 0)
-	require.NoError(t, err)
-	defer func() { _ = w.Close() }()
-
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-
-	done := make(chan error, 1)
-	go func() { _, err := b.readOpened("p.md"); done <- err }()
-	select {
-	case err := <-done:
-		assert.True(t, errors.Is(err, ErrNotRegular), "got %v", err)
-	case <-time.After(2 * time.Second):
-		t.Fatal("readOpened blocked on a FIFO with a writer")
-	}
-}
-
-func TestReadRegularRefusesSymlinksAndReadsRegularFiles(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(dir, "ok.md"), []byte("ok"), 0o644))
-	outside := filepath.Join(t.TempDir(), "s.md")
-	require.NoError(t, os.WriteFile(outside, []byte("secret"), 0o644))
-	require.NoError(t, os.Symlink(outside, filepath.Join(dir, "link.md")))
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	got, err := b.ReadRegular("ok.md")
-	require.NoError(t, err)
-	assert.Equal(t, "ok", string(got))
-	_, err = b.ReadRegular("link.md")
-	assert.True(t, errors.Is(err, ErrNotRegular), "got %v", err)
-	_, err = b.ReadRegular("../" + filepath.Base(outside))
-	assert.Error(t, err)
-}
-
-func TestRelAcceptsPathsInsideAndRefusesEscapes(t *testing.T) {
-	dir := t.TempDir()
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	for in, want := range map[string]string{
-		filepath.Join(dir, "wiki", "x.md"): "wiki/x.md",
-		filepath.Join(dir, "..foo", "x"):   "..foo/x",
-		dir:                                ".",
-	} {
-		got, err := b.Rel(in)
-		require.NoError(t, err, in)
-		assert.Equal(t, want, got, in)
-	}
-	for _, bad := range []string{filepath.Join(dir, "..", "x"), dir + "-evil/x", "/etc/passwd", filepath.Join(dir, "wiki", "..", "..", "x")} {
-		_, err := b.Rel(bad)
-		assert.Error(t, err, bad)
-	}
-}
-
-func TestWalkRegularYieldsOnlyRegularFiles(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, os.MkdirAll(filepath.Join(dir, "a"), 0o755))
-	require.NoError(t, os.WriteFile(filepath.Join(dir, "a", "n.md"), []byte("n"), 0o644))
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(dir, "a", "dirlink")))
-	require.NoError(t, os.Symlink(filepath.Join(dir, "a", "n.md"), filepath.Join(dir, "a", "filelink.md")))
-	require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "a", "p.md"), 0o644))
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	var got []string
-	require.NoError(t, b.WalkRegular(".", func(rel string) error { got = append(got, rel); return nil }, nil))
-	assert.Equal(t, []string{"a/n.md"}, got)
-}
-
-func TestReadRegularMaxRefusesAFileOverTheLimit(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(dir, "big.md"), make([]byte, 100), 0o644))
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	_, err = b.ReadRegularMax("big.md", 99)
-	assert.True(t, errors.Is(err, ErrTooLarge), "got %v", err)
-	got, err := b.ReadRegularMax("big.md", 100)
-	require.NoError(t, err)
-	assert.Len(t, got, 100)
-}
-
-func TestReadRegularMaxRefusesAndStopsReadingAFileThatGrowsPastTheLimit(t *testing.T) {
-	dir := t.TempDir()
-	p := filepath.Join(dir, "grow.md")
-	require.NoError(t, os.WriteFile(p, make([]byte, 10), 0o644))
-	consumed := &countingReader{}
-	wrapAfterStat = func(r io.Reader) io.Reader {
-		f, err := os.OpenFile(p, os.O_WRONLY|os.O_APPEND, 0)
-		require.NoError(t, err)
-		_, err = f.Write(make([]byte, 1000))
-		require.NoError(t, err)
-		require.NoError(t, f.Close())
-		consumed.r = r
-		return consumed
-	}
-	t.Cleanup(func() { wrapAfterStat = nil })
-	b, err := Open(dir)
-	require.NoError(t, err)
-	defer func() { _ = b.Close() }()
-	got, err := b.ReadRegularMax("grow.md", 100)
-	assert.True(t, errors.Is(err, ErrTooLarge), "got %v", err)
-	assert.Nil(t, got)
-	assert.LessOrEqual(t, consumed.n, 101, "read past max+1 bytes")
-}
-
-type countingReader struct {
-	r io.Reader
-	n int
-}
-
-func (c *countingReader) Read(p []byte) (int, error) {
-	n, err := c.r.Read(p)
-	c.n += n
-	return n, err
-}

ingestion/internal/brainfs/brainfstest/brainfstest.go

@@ -1,103 +0,0 @@
-// Package brainfstest is the shared harness for hyperguild#157's swap-race
-// tests: it keeps swapping a regular note, a symlink to an outside file, or a
-// FIFO (with a writer held open) into a path while a reader under test loops.
-// It observes two things: no sentinel byte from the outside file reaches the
-// reader's output, and the reader returns within a timeout. It does not cover
-// hard links, other paths into the reader, or races it does not generate.
-package brainfstest
-
-import (
-	"os"
-	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"testing"
-	"time"
-)
-
-// Sentinel marks the outside file's contents.
-const Sentinel = "zqbrainfssentinelonefiftyseven"
-
-// Outside writes a file outside any brain dir containing Sentinel.
-func Outside(t *testing.T) string {
-	t.Helper()
-	p := filepath.Join(t.TempDir(), "secret.md")
-	if err := os.WriteFile(p, []byte("# secret\n\n"+Sentinel), 0o644); err != nil {
-		t.Fatal(err)
-	}
-	return p
-}
-
-// StartSwap renames, in a loop, a regular note (content inside) and either
-// a symlink to outside or a FIFO into path, until the test ends. FIFOs get a
-// writer held open until the swapper exits, so a reader that skips the
-// handle check parks forever instead of reading EOF. The swapper is stopped
-// and awaited before t.TempDir cleanup.
-func StartSwap(t *testing.T, path, inside, outside string, fifo bool) {
-	t.Helper()
-	var stop atomic.Bool
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() {
-		defer wg.Done()
-		tmp := path + ".swap"
-		var writers []*os.File
-		defer func() {
-			for _, w := range writers {
-				_ = w.Close()
-			}
-		}()
-		for i := 0; !stop.Load(); i++ {
-			_ = os.Remove(tmp)
-			switch {
-			case i%2 == 0:
-				_ = os.WriteFile(tmp, []byte(inside), 0o644)
-			case fifo:
-				if err := syscall.Mkfifo(tmp, 0o644); err == nil {
-					if w, err := os.OpenFile(tmp, os.O_RDWR, 0); err == nil {
-						writers = append(writers, w)
-					}
-				}
-			default:
-				_ = os.Symlink(outside, tmp)
-			}
-			_ = os.Rename(tmp, path)
-		}
-	}()
-	t.Cleanup(func() { stop.Store(true); wg.Wait() })
-}
-
-// AssertNoLeak loops read for d while a symlink swapper runs on path, and
-// fails if any output contains Sentinel.
-func AssertNoLeak(t *testing.T, path, inside string, d time.Duration, read func() string) {
-	t.Helper()
-	StartSwap(t, path, inside, Outside(t), false)
-	leaks, iters := 0, 0
-	for deadline := time.Now().Add(d); time.Now().Before(deadline); iters++ {
-		leaks += strings.Count(read(), Sentinel)
-	}
-	if leaks > 0 {
-		t.Fatalf("outside file leaked %d times in %d iterations", leaks, iters)
-	}
-}
-
-// AssertNoHang loops read for d while a FIFO swapper (held writers) runs on
-// path, and fails if a read does not return within timeout.
-func AssertNoHang(t *testing.T, path, inside string, d, timeout time.Duration, read func()) {
-	t.Helper()
-	StartSwap(t, path, inside, Outside(t), true)
-	done := make(chan struct{})
-	go func() {
-		for deadline := time.Now().Add(d); time.Now().Before(deadline); {
-			read()
-		}
-		close(done)
-	}()
-	select {
-	case <-done:
-	case <-time.After(timeout):
-		t.Fatal("reader hung on a swapped-in FIFO with a writer held open")
-	}
-}

ingestion/internal/extract/callsites_test.go

@@ -1,130 +0,0 @@
-package extract
-
-import (
-	"go/ast"
-	"go/parser"
-	"go/token"
-	"os"
-	"path/filepath"
-	"strconv"
-	"strings"
-	"testing"
-)
-
-const extractPkg = "github.com/mathiasbq/hyperguild/ingestion/internal/extract"
-
-// parseNonTest parses every non-test Go file under the module.
-func parseNonTest(t *testing.T) map[string]*ast.File {
-	t.Helper()
-	fset := token.NewFileSet()
-	out := map[string]*ast.File{}
-	err := filepath.WalkDir(filepath.Join("..", ".."), func(p string, d os.DirEntry, err error) error {
-		if err != nil {
-			return err
-		}
-		if d.IsDir() && (d.Name() == "vendor" || d.Name() == ".git") {
-			return filepath.SkipDir
-		}
-		if d.IsDir() || !strings.HasSuffix(p, ".go") || strings.HasSuffix(p, "_test.go") {
-			return nil
-		}
-		f, perr := parser.ParseFile(fset, p, nil, 0)
-		if perr != nil {
-			return perr
-		}
-		out[filepath.ToSlash(p)] = f
-		return nil
-	})
-	if err != nil {
-		t.Fatal(err)
-	}
-	return out
-}
-
-// PR #156 review: reading a caller-chosen path by name reopens it after the
-// containment check (hyperguild#153 item 3). Outside this package, nothing
-// may reference extract.Text — called OR taken as a value — under whatever
-// name the package is imported (round 2: an alias and a func value slipped
-// past a plain text search).
-func TestNoPackageReferencesExtractTextOutsideExtract(t *testing.T) {
-	for path, f := range parseNonTest(t) {
-		if strings.Contains(path, "internal/extract/") {
-			continue
-		}
-		names := map[string]bool{}
-		for _, imp := range f.Imports {
-			if p, _ := strconv.Unquote(imp.Path.Value); p == extractPkg {
-				name := "extract"
-				if imp.Name != nil {
-					name = imp.Name.Name
-				}
-				if name == "." {
-					t.Errorf("%s dot-imports extract; import it by name so its use stays checkable", path)
-				}
-				names[name] = true
-			}
-		}
-		if len(names) == 0 {
-			continue
-		}
-		ast.Inspect(f, func(n ast.Node) bool {
-			if sel, ok := n.(*ast.SelectorExpr); ok {
-				if id, ok := sel.X.(*ast.Ident); ok && names[id.Name] && sel.Sel.Name == "Text" {
-					t.Errorf("%s references %s.Text; use extract.TextIn through an os.Root", path, id.Name)
-				}
-			}
-			return true
-		})
-	}
-}
-
-// PR #156 review round 2: the FIFO fix (open non-blocking, check the handle)
-// lives in OpenRegular, and reverting any caller to a plain Root.Open stayed
-// green. In the three raw/ reader FILES parsed below (and only those: a
-// wrapper in another file is not seen), no Open/OpenFile/ReadFile selector
-// may appear outside OpenRegular, called or taken as a method value. This is
-// a second layer; the behavioural race tests in internal/api and
-// internal/watcher are what catch a rewrite through any API (round 3).
-// osOpenAllowed names the functions in guarded files that may call os.Open*
-// on a path that is not a raw/ read.
-var osOpenAllowed = map[string]bool{
-	"internal/watcher/watcher.go:appendWatcherLog": true, // ingest log append, not a raw/ read
-	// Text reads by path; outside this package only TextIn's temp copy reaches
-	// it (TestNoPackageReferencesExtractTextOutsideExtract enforces that).
-	"internal/extract/extract.go:Text": true,
-}
-
-func TestRawReaderFilesUseNoOpenOrReadFileOutsideOpenRegular(t *testing.T) {
-	guarded := []string{"internal/api/rawroot.go", "internal/extract/extract.go", "internal/watcher/watcher.go"}
-	files := parseNonTest(t)
-	for _, g := range guarded {
-		var f *ast.File
-		for p, af := range files {
-			if strings.HasSuffix(p, g) {
-				f = af
-			}
-		}
-		if f == nil {
-			t.Fatalf("guarded file %s not found", g)
-		}
-		for _, decl := range f.Decls {
-			fn, ok := decl.(*ast.FuncDecl)
-			if !ok || fn.Name.Name == "OpenRegular" {
-				continue
-			}
-			ast.Inspect(fn, func(n ast.Node) bool {
-				sel, ok := n.(*ast.SelectorExpr)
-				if !ok {
-					return true
-				}
-				if sel.Sel.Name == "Open" || sel.Sel.Name == "OpenFile" || sel.Sel.Name == "ReadFile" {
-					if id, ok := sel.X.(*ast.Ident); ok && id.Name == "os" && osOpenAllowed[g+":"+fn.Name.Name] {
-						return true // a named non-raw write (the ingest log append)
-					}
-					t.Errorf("%s: %s calls .%s; raw/ reads must go through OpenRegular", g, fn.Name.Name, sel.Sel.Name)
-				}
-				return true
-			})
-		}
-	}
-}

ingestion/internal/extract/extract.go

@@ -3,10 +3,8 @@ package extract
 
 import (
 	"fmt"
-	"io"
 	"os"
 	"strings"
-	"syscall"
 )
 
 // Text reads the file at path and returns its plain-text content.
@@ -43,70 +41,3 @@ func fileExt(path string) string {
 	}
 	return ""
 }
-
-// TextIn is Text for a file read through an os.Root (hyperguild#153): the
-// read cannot leave the root, so a path checked earlier cannot be swapped
-// for a symlink pointing elsewhere before it is read. Formats that need an
-// external tool (pdf, docmark) are copied to a temp file from the handle
-// opened through the root, then extracted from the copy.
-func TextIn(r *os.Root, rel string) (string, error) {
-	fi, err := r.Lstat(rel)
-	if err != nil {
-		return "", fmt.Errorf("stat %s: %w", rel, err)
-	}
-	if !fi.Mode().IsRegular() {
-		return "", fmt.Errorf("%s: not a regular file", rel)
-	}
-	f, _, err := OpenRegular(r, rel)
-	if err != nil {
-		return "", fmt.Errorf("open %s: %w", rel, err)
-	}
-	defer func() { _ = f.Close() }()
-	ext := strings.ToLower(fileExt(rel))
-	switch ext {
-	case ".md", ".txt":
-		b, err := io.ReadAll(f)
-		if err != nil {
-			return "", fmt.Errorf("read %s: %w", rel, err)
-		}
-		return string(b), nil
-	case ".pdf", ".docx", ".xlsx", ".pptx", ".png", ".jpg", ".jpeg":
-		tmp, err := os.CreateTemp("", "ingest-*"+ext)
-		if err != nil {
-			return "", fmt.Errorf("temp copy: %w", err)
-		}
-		defer func() { _ = os.Remove(tmp.Name()) }()
-		if _, err := io.Copy(tmp, f); err != nil {
-			_ = tmp.Close()
-			return "", fmt.Errorf("copy %s: %w", rel, err)
-		}
-		if err := tmp.Close(); err != nil {
-			return "", fmt.Errorf("copy %s: %w", rel, err)
-		}
-		return Text(tmp.Name())
-	default:
-		return "", fmt.Errorf("unsupported file extension: %s", ext)
-	}
-}
-
-// OpenRegular opens rel inside r and returns it only if the OPENED handle is
-// a regular file. The open is non-blocking, so a FIFO swapped in after an
-// Lstat (PR #156 review: it hung open(2) forever) is refused promptly rather
-// than stalling the reader. Callers keep their Lstat check, which refuses
-// symlinks outright; this closes the race between that check and the open.
-func OpenRegular(r *os.Root, rel string) (*os.File, os.FileInfo, error) {
-	f, err := r.OpenFile(rel, os.O_RDONLY|syscall.O_NONBLOCK, 0)
-	if err != nil {
-		return nil, nil, err
-	}
-	fi, err := f.Stat()
-	if err != nil {
-		_ = f.Close()
-		return nil, nil, err
-	}
-	if !fi.Mode().IsRegular() {
-		_ = f.Close()
-		return nil, nil, fmt.Errorf("%s: not a regular file", rel)
-	}
-	return f, fi, nil
-}

ingestion/internal/extract/root_test.go

@@ -1,64 +0,0 @@
-package extract
-
-import (
-	"os"
-	"path/filepath"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// hyperguild#153 item 3: ingest checked a path, then extract.Text reopened it
-// by name, so a file swapped for a symlink in between was followed out of
-// raw/. TextIn reads through an os.Root, which refuses to leave the root.
-func TestTextInReadsInsideTheRootAndRefusesEscapes(t *testing.T) {
-	root := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(root, "ok.md"), []byte("inside"), 0o644))
-	outside := filepath.Join(t.TempDir(), "secret.md")
-	require.NoError(t, os.WriteFile(outside, []byte("SENTINEL-153-TEXTIN"), 0o644))
-	require.NoError(t, os.Symlink(outside, filepath.Join(root, "leak.md")))
-
-	r, err := os.OpenRoot(root)
-	require.NoError(t, err)
-	defer func() { _ = r.Close() }()
-
-	got, err := TextIn(r, "ok.md")
-	require.NoError(t, err)
-	assert.Equal(t, "inside", got)
-
-	got, err = TextIn(r, "leak.md")
-	assert.Error(t, err)
-	assert.NotContains(t, got, "SENTINEL-153-TEXTIN")
-
-	_, err = TextIn(r, "../"+filepath.Base(outside))
-	assert.Error(t, err)
-}
-
-// PR #156 review: a regular file swapped for a FIFO between Lstat and Open
-// blocked open(2) forever. OpenRegular opens non-blocking and checks the type
-// on the opened handle, so a FIFO is refused promptly instead of hanging.
-func TestOpenRegularRefusesAFIFOWithoutBlocking(t *testing.T) {
-	dir := t.TempDir()
-	require.NoError(t, syscall.Mkfifo(filepath.Join(dir, "pipe.md"), 0o644))
-	r, err := os.OpenRoot(dir)
-	require.NoError(t, err)
-	defer func() { _ = r.Close() }()
-
-	done := make(chan error, 1)
-	go func() {
-		f, _, err := OpenRegular(r, "pipe.md")
-		if f != nil {
-			_ = f.Close()
-		}
-		done <- err
-	}()
-	select {
-	case err := <-done:
-		assert.Error(t, err)
-	case <-time.After(2 * time.Second):
-		t.Fatal("OpenRegular blocked on a FIFO")
-	}
-}

ingestion/internal/mcp/handlers.go

@@ -12,6 +12,7 @@ import (
 	"github.com/mathiasbq/hyperguild/ingestion/internal/api"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/brain"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/capture"
+	"github.com/mathiasbq/hyperguild/ingestion/internal/extract"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/graphsync"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/notes"
 	"github.com/mathiasbq/hyperguild/ingestion/internal/pipeline"
@@ -548,7 +549,7 @@ func (s *Server) brainIngest(ctx context.Context, args json.RawMessage) (json.Ra
 			return nil, err
 		}
 		a.Path = resolved
-		text, err := api.ReadIngestFile(s.brainDir, a.Path)
+		text, err := extract.Text(a.Path)
 		if err != nil {
 			return nil, fmt.Errorf("extract: %w", err)
 		}

ingestion/internal/mcp/readexcerpt_root_test.go

@@ -1,34 +0,0 @@
-package mcp
-
-import (
-	"os"
-	"path/filepath"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
-)
-
-// hyperguild#157 slice 2: readExcerpt (brain_context) read by path, so a
-// symlink in the Syncthing-shared brain put an outside file in context, and
-// a FIFO hung it. It now reads through brainfs.
-func TestReadExcerptRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
-	brain := t.TempDir()
-	facts := filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	require.NoError(t, os.Symlink(brainfstest.Outside(t), filepath.Join(facts, "leak.md")))
-	assert.NotContains(t, readExcerpt(brain, "wiki/w/facts/leak.md", 500), brainfstest.Sentinel)
-
-	p := filepath.Join(facts, "r.md")
-	require.NoError(t, os.WriteFile(p, []byte("---\nx: y\n---\ninside"), 0o644))
-	assert.Equal(t, "inside", readExcerpt(brain, "wiki/w/facts/r.md", 500))
-	brainfstest.AssertNoLeak(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, func() string {
-		return readExcerpt(brain, "wiki/w/facts/r.md", 500)
-	})
-	brainfstest.AssertNoHang(t, p, "---\nx: y\n---\ninside", 1500*time.Millisecond, 6*time.Second, func() {
-		_ = readExcerpt(brain, "wiki/w/facts/r.md", 500)
-	})
-}

ingestion/internal/mcp/tools_context.go

@@ -4,7 +4,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
+	"os"
 	"path/filepath"
 	"sort"
 	"strings"
@@ -181,13 +181,8 @@ func readExcerpt(brainDir, relPath string, maxLen int) string {
 	if relPath == "" {
 		return ""
 	}
-	// Through the brain root, regular files only (hyperguild#157).
-	bfs, err := brainfs.Open(brainDir)
-	if err != nil {
-		return ""
-	}
-	defer func() { _ = bfs.Close() }()
-	content, err := bfs.ReadRegular(filepath.ToSlash(relPath))
+	full := filepath.Join(brainDir, filepath.FromSlash(relPath))
+	content, err := os.ReadFile(full)
 	if err != nil {
 		return ""
 	}

ingestion/internal/notes/edit.go

@@ -5,7 +5,6 @@ import (
 	"encoding/hex"
 	"errors"
 	"fmt"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
 	"os"
 	"path/filepath"
 	"strings"
@@ -72,18 +71,14 @@ func ReadScoped(brainDir, path string, scope Scope) ([]byte, error) {
 }
 
 func readContained(t Target) ([]byte, error) {
-	// Through the brain root, regular files only, size checked on the opened
-	// handle (hyperguild#157): Resolve's Lstat could be raced by a swap, and
-	// os.ReadFile blocked on a FIFO.
-	bfs, err := brainfs.Open(t.Brain)
+	fi, err := os.Stat(t.Abs)
 	if err != nil {
-		return nil, fmt.Errorf("notes: %w", err)
+		return nil, fmt.Errorf("notes: stat %s: %w", t.Rel, err)
 	}
-	defer func() { _ = bfs.Close() }()
-	b, err := bfs.ReadRegularMax(t.Rel, maxNoteBytes)
-	if errors.Is(err, brainfs.ErrTooLarge) {
-		return nil, fmt.Errorf("%w: %s (limit %d)", ErrTooLarge, t.Rel, maxNoteBytes)
+	if fi.Size() > maxNoteBytes {
+		return nil, fmt.Errorf("%w: %s is %d bytes (limit %d)", ErrTooLarge, t.Rel, fi.Size(), maxNoteBytes)
 	}
+	b, err := os.ReadFile(t.Abs)
 	if err != nil {
 		return nil, fmt.Errorf("notes: read %s: %w", t.Rel, err)
 	}

ingestion/internal/notes/path.go

@@ -58,9 +58,8 @@ var (
 
 // Target is a resolved, contained path.
 type Target struct {
-	Rel   string // brain-relative, cleaned
-	Abs   string // absolute, symlink-checked
-	Brain string // the brain root Rel is relative to
+	Rel string // brain-relative, cleaned
+	Abs string // absolute, symlink-checked
 }
 
 // Resolve canonicalises path, then checks it against the scope's allowlist,
@@ -110,7 +109,7 @@ func Resolve(brainDir, path string, scope Scope) (Target, error) {
 		return Target{}, err
 	}
 
-	return Target{Rel: clean, Abs: abs, Brain: brainDir}, nil
+	return Target{Rel: clean, Abs: abs}, nil
 }
 
 // allow enforces the scope's allowlist. The two scopes are DISJOINT, not

ingestion/internal/notes/read_root_test.go

@@ -1,40 +0,0 @@
-package notes
-
-import (
-	"os"
-	"path/filepath"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs/brainfstest"
-)
-
-// hyperguild#157 slice 2: ReadScoped (brain_read) Lstat-checked the final
-// component, then reopened by name with os.ReadFile: a swap after the check
-// was followed, and a FIFO blocked the read. It now reads through brainfs.
-func TestReadScopedRefusesSymlinksAndSurvivesSwapRaces(t *testing.T) {
-	brain := t.TempDir()
-	facts := filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	p := filepath.Join(facts, "r.md")
-	read := func() string {
-		b, _ := ReadScoped(brain, "wiki/w/facts/r.md", ScopeNotes)
-		return string(b)
-	}
-	require.NoError(t, os.WriteFile(p, []byte("inside"), 0o644))
-	assert.Equal(t, "inside", read())
-	brainfstest.AssertNoLeak(t, p, "inside", 1500*time.Millisecond, read)
-	brainfstest.AssertNoHang(t, p, "inside", 1500*time.Millisecond, 6*time.Second, func() { _ = read() })
-}
-
-func TestReadScopedRefusesANoteOverMaxNoteBytes(t *testing.T) {
-	brain := t.TempDir()
-	facts := filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	require.NoError(t, os.WriteFile(filepath.Join(facts, "big.md"), make([]byte, maxNoteBytes+1), 0o644))
-	_, err := ReadScoped(brain, "wiki/w/facts/big.md", ScopeNotes)
-	assert.ErrorIs(t, err, ErrTooLarge)
-}

ingestion/internal/search/brainroot_test.go

@@ -1,149 +0,0 @@
-package search
-
-import (
-	"os"
-	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"syscall"
-	"testing"
-	"time"
-
-	"github.com/stretchr/testify/assert"
-	"github.com/stretchr/testify/require"
-)
-
-// hyperguild#157 option 1: search walked and read the brain by path, so a
-// symlink anywhere in the Syncthing-shared brain dir (wiki/<w>/<h>/leak.md ->
-// /any/file) put the target's contents in brain_query results (reproduced in
-// the PR #156 review). Search now walks and reads through an os.Root on the
-// brain dir and only reads regular files.
-
-const sentinel157 = "zqsentinelonefiftyseven"
-
-func searchBrain(t *testing.T) (brain, facts, outside string) {
-	t.Helper()
-	brain = t.TempDir()
-	facts = filepath.Join(brain, "wiki", "w", "facts")
-	require.NoError(t, os.MkdirAll(facts, 0o755))
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "knowledge"), 0o755))
-	outside = filepath.Join(t.TempDir(), "secret.md")
-	require.NoError(t, os.WriteFile(outside, []byte("# secret\n\n"+sentinel157), 0o644))
-	return brain, facts, outside
-}
-
-func leaked(rs []Result) int {
-	n := 0
-	for _, r := range rs {
-		n += strings.Count(r.Excerpt+r.Title, sentinel157)
-	}
-	return n
-}
-
-func TestQueryDoesNotFollowASymlinkInTheWiki(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(outside, filepath.Join(facts, "leak.md")))
-	require.NoError(t, os.WriteFile(filepath.Join(facts, "real.md"), []byte("# real\n\nzqrealnote"), 0o644))
-	rs, err := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-	require.NoError(t, err)
-	assert.Zero(t, leaked(rs))
-	rs, err = Query(brain, QueryOptions{Query: "zqrealnote", Limit: 10})
-	require.NoError(t, err)
-	assert.Len(t, rs, 1, "regular notes are still found")
-}
-
-func TestQueryDoesNotFollowASymlinkedDirectory(t *testing.T) {
-	brain, _, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(filepath.Dir(outside), filepath.Join(brain, "wiki", "linked")))
-	rs, err := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-	require.NoError(t, err)
-	assert.Zero(t, leaked(rs))
-}
-
-// swapInto keeps renaming a regular note, and either a symlink to outside or
-// a FIFO, into path. Each FIFO has a writer held open (as any local process
-// can): a FIFO with no writer reads EOF at once, which would let a reader
-// that skips the handle check pass (PR #161 round 2: with a held writer, a
-// bypass of readOpened hung Query after 5 iterations).
-// startSwap runs swapInto until the test ends and waits for it before
-// t.TempDir's cleanup (hyperguild#162).
-func startSwap(t *testing.T, path, outside string, fifo bool) {
-	t.Helper()
-	var stop atomic.Bool
-	var wg sync.WaitGroup
-	wg.Add(1)
-	go func() { defer wg.Done(); swapInto(path, outside, fifo, &stop) }()
-	t.Cleanup(func() { stop.Store(true); wg.Wait() })
-}
-
-func swapInto(path, outside string, fifo bool, stop *atomic.Bool) {
-	tmp := path + ".swap"
-	// Writers stay open until the swapper exits, so a reader parked on a
-	// FIFO stays parked (closing one would hand the reader EOF and hide the
-	// hang). Past the fd limit, new FIFOs simply have no writer.
-	var writers []*os.File
-	defer func() {
-		for _, w := range writers {
-			_ = w.Close()
-		}
-	}()
-	for i := 0; !stop.Load(); i++ {
-		_ = os.Remove(tmp)
-		switch {
-		case i%2 == 0:
-			_ = os.WriteFile(tmp, []byte("# r\n\nzqinside "+"zqracenote"), 0o644)
-		case fifo:
-			if err := syscall.Mkfifo(tmp, 0o644); err == nil {
-				if w, err := os.OpenFile(tmp, os.O_RDWR, 0); err == nil {
-					writers = append(writers, w)
-				}
-			}
-		default:
-			_ = os.Symlink(outside, tmp)
-		}
-		_ = os.Rename(tmp, path)
-	}
-}
-
-func TestQueryLeaksNothingUnderASymlinkSwapRace(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	startSwap(t, filepath.Join(facts, "r.md"), outside, false)
-	leaks, iters := 0, 0
-	for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); iters++ {
-		rs, _ := Query(brain, QueryOptions{Query: sentinel157, Limit: 10})
-		leaks += leaked(rs)
-	}
-	if leaks > 0 {
-		t.Fatalf("outside file leaked %d times in %d iterations", leaks, iters)
-	}
-}
-
-func TestQueryDoesNotHangUnderAFIFOSwapRace(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	startSwap(t, filepath.Join(facts, "r.md"), outside, true)
-	done := make(chan struct{})
-	go func() {
-		for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
-			_, _ = Query(brain, QueryOptions{Query: "zqracenote", Limit: 10})
-		}
-		close(done)
-	}()
-	select {
-	case <-done:
-	case <-time.After(6 * time.Second):
-		if f, err := os.OpenFile(filepath.Join(facts, "r.md"), os.O_WRONLY|syscall.O_NONBLOCK, 0); err == nil {
-			_ = f.Close()
-		}
-		t.Fatal("search hung on a swapped-in FIFO")
-	}
-}
-
-// hydrate (vector-only hits) reads by relative path; it must refuse a symlink.
-func TestHydrateRefusesASymlink(t *testing.T) {
-	brain, facts, outside := searchBrain(t)
-	require.NoError(t, os.Symlink(outside, filepath.Join(facts, "leak.md")))
-	r, err := hydrate(brain, "wiki/w/facts/leak.md")
-	assert.Error(t, err)
-	assert.NotContains(t, r.Excerpt, sentinel157)
-}

ingestion/internal/search/search.go

@@ -5,7 +5,6 @@ import (
 	"bufio"
 	"context"
 	"fmt"
-	"github.com/mathiasbq/hyperguild/ingestion/internal/brainfs"
 	sharedfm "github.com/mathiasbq/hyperguild/ingestion/internal/frontmatter"
 	"log/slog"
 	"os"
@@ -120,30 +119,20 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 		return nil, err
 	}
 
-	// Walk and read through an os.Root on the brain dir, regular files only
-	// (hyperguild#157): a symlink planted in the Syncthing-shared brain must
-	// not put an outside file in results, nor a FIFO hang the query.
-	bfs, err := brainfs.Open(brainDir)
-	if err != nil {
-		return nil, fmt.Errorf("search: %w", err)
-	}
-	defer func() { _ = bfs.Close() }()
-
 	var results []Result
 	for _, dir := range roots {
 		if _, statErr := os.Stat(dir); os.IsNotExist(statErr) {
 			continue
 		}
-		relDir, relErr := bfs.Rel(dir)
-		if relErr != nil {
-			return nil, fmt.Errorf("search: %w", relErr)
-		}
-		err := bfs.WalkRegular(relDir, func(relPath string) error {
-			if !strings.HasSuffix(relPath, ".md") {
+		err := filepath.WalkDir(dir, func(path string, d os.DirEntry, err error) error {
+			if err != nil {
+				slog.Warn("search: skipping path", "path", path, "err", err)
 				return nil
 			}
-			path := filepath.Join(brainDir, filepath.FromSlash(relPath))
-			content, err := bfs.ReadRegular(relPath)
+			if d.IsDir() || !strings.HasSuffix(path, ".md") {
+				return nil
+			}
+			content, err := os.ReadFile(path)
 			if err != nil {
 				slog.Warn("search: skipping unreadable file", "path", path, "err", err)
 				return nil
@@ -166,7 +155,7 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 			origin, writtenBy := extractProvenance(string(content))
 			results = append(results, Result{
 				Path:    rel,
-				Title:   extractTitle(string(content), filepath.Base(relPath)),
+				Title:   extractTitle(string(content), d.Name()),
 				Excerpt: excerpt(string(content), 300),
 				Score:   score,
 				Wing:    wing,
@@ -177,8 +166,6 @@ func QueryContext(ctx context.Context, brainDir string, opts QueryOptions) ([]Re
 				WrittenBy: writtenBy,
 			})
 			return nil
-		}, func(p string, err error) {
-			slog.Warn("search: skipping path", "path", p, "err", err)
 		})
 		if err != nil {
 			return nil, err
@@ -288,13 +275,8 @@ func pathInScope(relPath, wing, hall string) bool {
 // excerpt, wing, and hall populated. Used for paths that surface only
 // via vector search.
 func hydrate(brainDir, relPath string) (Result, error) {
-	// Through the brain root, regular files only (hyperguild#157).
-	bfs, err := brainfs.Open(brainDir)
-	if err != nil {
-		return Result{}, err
-	}
-	defer func() { _ = bfs.Close() }()
-	content, err := bfs.ReadRegular(filepath.ToSlash(relPath))
+	full := filepath.Join(brainDir, filepath.FromSlash(relPath))
+	content, err := os.ReadFile(full)
 	if err != nil {
 		return Result{}, err
 	}

ingestion/internal/symlinkguard/symlinkguard.go

@@ -9,7 +9,6 @@ package symlinkguard
 
 import (
 	"context"
-	"errors"
 	"fmt"
 	"io/fs"
 	"log/slog"
@@ -22,14 +21,7 @@ import (
 )
 
 // DefaultInterval is used when Watch is given a non-positive interval.
-const DefaultInterval = 30 * time.Second
-
-// Alert bounds (PR #160 review): the planter chooses the paths, so neither an
-// alert's size nor the queue may grow with what was planted.
-const (
-	maxPathsPerAlert = 10 // further paths are counted, not listed
-	maxPending       = 5  // queued alerts; the rest become one summary
-)
+const DefaultInterval = 5 * time.Minute
 
 // Notifier is the out-of-band alert sink (audit.Notifier satisfies it).
 type Notifier interface {
@@ -49,7 +41,13 @@ func (r Result) Clean() bool { return len(r.Symlinks) == 0 && len(r.Unreadable)
 // String describes the findings for a log line or alert. Paths are quoted, so
 // a name containing a newline cannot blur two findings together.
 func (r Result) String() string {
-	q := r.quoted
+	q := func(xs []string) string {
+		out := make([]string, len(xs))
+		for i, x := range xs {
+			out[i] = strconv.Quote(x)
+		}
+		return strings.Join(out, ", ")
+	}
 	var parts []string
 	if len(r.Symlinks) > 0 {
 		parts = append(parts, fmt.Sprintf("%d symlink(s): %s", len(r.Symlinks), q(r.Symlinks)))
@@ -60,24 +58,6 @@ func (r Result) String() string {
 	return strings.Join(parts, "; ")
 }
 
-// quoted joins at most maxPathsPerAlert paths, each strconv-quoted, and
-// counts the rest.
-func (Result) quoted(xs []string) string {
-	shown := xs
-	if len(shown) > maxPathsPerAlert {
-		shown = shown[:maxPathsPerAlert]
-	}
-	out := make([]string, len(shown))
-	for i, x := range shown {
-		out[i] = strconv.Quote(x)
-	}
-	s := strings.Join(out, ", ")
-	if extra := len(xs) - len(shown); extra > 0 {
-		s += fmt.Sprintf(" and %d more", extra)
-	}
-	return s
-}
-
 // Scan walks the brain dir and reports every symlink in it and every
 // directory it could not read, continuing past both. Only the ROOT .git is
 // skipped (it is brain-sync's, and no reader walks it); a nested .git is
@@ -126,20 +106,15 @@ func Scan(brainDir string) (Result, error) {
 }
 
 // Watch rescans every interval (DefaultInterval if non-positive) until ctx
-// ends. Every symlink found is removed (the link, never its target:
-// hyperguild#157 decision), so one is readable for at most one interval.
-// Each removal is an alert; a persistent finding (a link it could not remove,
-// an unreadable dir, a scan error) alerts once per distinct state. Alerts
-// queue until Notify succeeds (see alertQueue): an outage delays them, and
-// any that do not fit the bounded queue, or are undeliverable, are counted in
-// a summary that is never dropped, so every incident is delivered or counted.
+// ends, and notifies once per distinct non-clean result. A result counts as
+// reported only after Notify succeeds, so an alert lost to a notifier outage
+// is retried on the next tick.
 func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notifier) {
 	if interval <= 0 {
 		interval = DefaultInterval
 	}
 	t := time.NewTicker(interval)
 	defer t.Stop()
-	q := &alertQueue{}
 	reported := ""
 	for {
 		select {
@@ -148,117 +123,31 @@ func Watch(ctx context.Context, brainDir string, interval time.Duration, n Notif
 		case <-t.C:
 		}
 		r, err := Scan(brainDir)
-		key := ""
-		if err != nil {
-			key = "cannot scan the brain dir: " + err.Error()
-		} else {
-			removed, failed := removeSymlinks(brainDir, r.Symlinks)
-			if len(removed) > 0 {
-				slog.Error("brain symlink tripwire removed symlinks", "count", len(removed))
-				q.push(fmt.Sprintf("brain symlink tripwire (hyperguild#157): removed %d symlink(s) from the brain dir: %s",
-					len(removed), Result{}.quoted(removed)))
-			}
-			key = Result{Symlinks: failed, Unreadable: r.Unreadable}.String()
-		}
-		if key != reported {
-			reported = key
-			if key != "" {
-				slog.Error("brain symlink tripwire", "finding", key)
-				q.push("brain symlink tripwire (hyperguild#157): " + key +
-					". Readers may follow symlinks out of the brain; fix them.")
-			}
-		}
-		q.flush(ctx, n)
-	}
-}
-
-// alertQueue holds alerts until Notify accepts them. It is bounded (at most
-// maxPending queued) and loses nothing from the record: every overflow and
-// every dropped message increments suppressed, sent as a summary that is
-// retried every tick and never dropped.
-//
-// A message is dropped only when the notifier rejects it permanently (an
-// error with Permanent() == true, e.g. ntfy 413). Every other failure is an
-// outage, and nothing is dropped however long it lasts (PR #160 round 2:
-// dropping after N raw failures lost the only incident of a long outage).
-// Each flush tries every queued message, so one undeliverable message cannot
-// block the others.
-type alertQueue struct {
-	items      []string
-	suppressed int
-}
-
-func (q *alertQueue) push(msg string) {
-	if len(q.items) >= maxPending {
-		q.suppressed++
-		return
-	}
-	q.items = append(q.items, msg)
-}
-
-func (q *alertQueue) flush(ctx context.Context, n Notifier) {
-	if n == nil {
-		q.items, q.suppressed = nil, 0
-		return
-	}
-	kept := q.items[:0]
-	for _, msg := range q.items {
-		err := n.Notify(ctx, msg)
+		var key, msg string
 		switch {
-		case err == nil:
-		case permanent(err):
-			slog.Error("symlink tripwire alert rejected permanently; counted in the summary", "err", err)
-			q.suppressed++
+		case err != nil:
+			key = "error: " + err.Error()
+			msg = "brain symlink tripwire (hyperguild#157): cannot scan the brain dir: " + err.Error()
+		case r.Clean():
+			reported = ""
+			continue
 		default:
+			key = r.String()
+			msg = "brain symlink tripwire (hyperguild#157): " + key +
+				". Readers may follow symlinks out of the brain; remove them."
+		}
+		if key == reported {
+			continue
+		}
+		slog.Error("brain symlink tripwire", "finding", key)
+		if n == nil {
+			reported = key
+			continue
+		}
+		if err := n.Notify(ctx, msg); err != nil {
 			slog.Warn("symlink tripwire notify failed; will retry", "err", err)
-			kept = append(kept, msg)
-		}
-	}
-	q.items = kept
-	if q.suppressed > 0 {
-		msg := fmt.Sprintf("brain symlink tripwire (hyperguild#157): %d further alert(s) were suppressed or undeliverable; check the ingestion logs", q.suppressed)
-		if err := n.Notify(ctx, msg); err == nil {
-			q.suppressed = 0
-		}
-	}
-}
-
-// permanent reports whether err says the message itself can never be sent.
-func permanent(err error) bool {
-	var p interface{ Permanent() bool }
-	return errors.As(err, &p) && p.Permanent()
-}
-
-// removeSymlinks unlinks each brain-relative symlink through an os.Root on
-// the brain dir, after re-checking it is still a symlink. Removing a symlink
-// never touches its target. It returns what was removed and what was not.
-func removeSymlinks(brainDir string, rels []string) (removed, failed []string) {
-	if len(rels) == 0 {
-		return nil, nil
-	}
-	resolved, err := filepath.EvalSymlinks(brainDir)
-	if err != nil {
-		return nil, rels
-	}
-	root, err := os.OpenRoot(resolved)
-	if err != nil {
-		return nil, rels
-	}
-	defer func() { _ = root.Close() }()
-	for _, rel := range rels {
-		fi, err := root.Lstat(rel)
-		if err != nil || fi.Mode()&os.ModeSymlink == 0 {
-			failed = append(failed, rel)
 			continue
 		}
-		// root.Remove, not os.RemoveAll: if the link is swapped for a
-		// non-empty directory after the Lstat, Remove refuses rather than
-		// deleting it recursively, and it cannot act outside the root.
-		if err := root.Remove(rel); err != nil {
-			failed = append(failed, rel)
-			continue
-		}
-		removed = append(removed, rel)
+		reported = key
 	}
-	return removed, failed
 }

ingestion/internal/symlinkguard/symlinkguard_test.go

@@ -3,13 +3,9 @@ package symlinkguard
 import (
 	"context"
 	"errors"
-	"fmt"
 	"os"
 	"path/filepath"
-	"strconv"
-	"strings"
 	"sync"
-	"sync/atomic"
 	"testing"
 	"time"
 
@@ -186,313 +182,8 @@ func TestWatchReportsASetAgainAfterItWasCleared(t *testing.T) {
 	go Watch(ctx, brain, 10*time.Millisecond, n)
 	require.NoError(t, os.Symlink(t.TempDir(), link))
 	require.Eventually(t, func() bool { return n.count() == 1 }, 2*time.Second, 10*time.Millisecond)
-	// The tripwire removed it; planting it again is a new event.
-	require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 10*time.Millisecond)
+	require.NoError(t, os.Remove(link))
+	time.Sleep(60 * time.Millisecond)
 	require.NoError(t, os.Symlink(t.TempDir(), link))
 	require.Eventually(t, func() bool { return n.count() == 2 }, 2*time.Second, 10*time.Millisecond)
 }
-
-// hyperguild#157 decision: the runtime tripwire removes a symlink it finds
-// (the link, never its target), so it is readable for at most one interval
-// instead of until a human acts, and the alert says what was removed.
-func TestWatchRemovesAnOutsideFileLinkButNotItsTargetOrRegularFiles(t *testing.T) {
-	brain := t.TempDir()
-	outside := filepath.Join(t.TempDir(), "target.md")
-	require.NoError(t, os.WriteFile(outside, []byte("keep me"), 0o644))
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
-	link := filepath.Join(brain, "wiki", "leak.md")
-	require.NoError(t, os.Symlink(outside, link))
-	regular := filepath.Join(brain, "wiki", "note.md")
-	require.NoError(t, os.WriteFile(regular, []byte("note"), 0o644))
-
-	n := &recNotifier{}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-
-	require.Eventually(t, func() bool {
-		_, err := os.Lstat(link)
-		return os.IsNotExist(err)
-	}, 2*time.Second, 10*time.Millisecond, "the symlink is removed")
-	require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 10*time.Millisecond)
-	b, err := os.ReadFile(outside)
-	require.NoError(t, err)
-	assert.Equal(t, "keep me", string(b), "the target is untouched")
-	b, err = os.ReadFile(regular)
-	require.NoError(t, err)
-	assert.Equal(t, "note", string(b), "regular files are untouched")
-	n.mu.Lock()
-	defer n.mu.Unlock()
-	assert.Contains(t, n.msgs[0], "removed")
-	assert.Contains(t, n.msgs[0], "wiki/leak.md")
-}
-
-func TestDefaultIntervalIsThirtySeconds(t *testing.T) {
-	assert.Equal(t, 30*time.Second, DefaultInterval)
-}
-
-// PR #160 review: the Lstat re-check and the os.Root confinement in
-// removeSymlinks had no test. Anything that is not a symlink inside the brain
-// is reported failed and nothing is deleted.
-func TestRemoveSymlinksRefusesNonSymlinksEscapesAndPathsThroughLinks(t *testing.T) {
-	brain := t.TempDir()
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki", "real"), 0o755))
-	regular := filepath.Join(brain, "wiki", "note.md")
-	require.NoError(t, os.WriteFile(regular, []byte("note"), 0o644))
-	inLinked := filepath.Join(brain, "wiki", "real", "x.md")
-	require.NoError(t, os.WriteFile(inLinked, []byte("x"), 0o644))
-	require.NoError(t, os.Symlink(filepath.Join(brain, "wiki", "real"), filepath.Join(brain, "wiki", "linkdir")))
-	outside := filepath.Join(filepath.Dir(brain), "outside-157.md")
-	require.NoError(t, os.WriteFile(outside, []byte("o"), 0o644))
-	t.Cleanup(func() { _ = os.Remove(outside) })
-
-	removed, failed := removeSymlinks(brain, []string{
-		"wiki/note.md", "wiki/linkdir/x.md", "../outside-157.md", "wiki/missing.md",
-	})
-	assert.Empty(t, removed)
-	assert.ElementsMatch(t, []string{"wiki/note.md", "wiki/linkdir/x.md", "../outside-157.md", "wiki/missing.md"}, failed)
-	for _, p := range []string{regular, inLinked, outside} {
-		_, err := os.Stat(p)
-		assert.NoError(t, err, "%s must not be deleted", p)
-	}
-}
-
-// Dir targets and in-brain targets survive too, not only an outside file.
-func TestWatchRemovesDirAndInBrainLinksButNotTheirTargets(t *testing.T) {
-	brain := t.TempDir()
-	outsideDir := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(outsideDir, "keep.md"), []byte("keep"), 0o644))
-	require.NoError(t, os.MkdirAll(filepath.Join(brain, "wiki"), 0o755))
-	inBrain := filepath.Join(brain, "wiki", "target.md")
-	require.NoError(t, os.WriteFile(inBrain, []byte("t"), 0o644))
-	require.NoError(t, os.Symlink(outsideDir, filepath.Join(brain, "wiki", "dirlink")))
-	require.NoError(t, os.Symlink(inBrain, filepath.Join(brain, "wiki", "inlink.md")))
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, nil)
-	require.Eventually(t, func() bool {
-		r, err := Scan(brain)
-		return err == nil && len(r.Symlinks) == 0
-	}, 2*time.Second, 10*time.Millisecond)
-	for _, p := range []string{filepath.Join(outsideDir, "keep.md"), inBrain} {
-		_, err := os.Stat(p)
-		assert.NoError(t, err, "%s must survive", p)
-	}
-}
-
-// PR #160 review: the planter chooses the paths. An alert names at most
-// maxPathsPerAlert of them and counts the rest, so its size stays bounded.
-func TestWatchAlertNamesBoundedPathsAndCountsTheRest(t *testing.T) {
-	brain := t.TempDir()
-	for i := 0; i < 50; i++ {
-		require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, strings.Repeat("n", 100)+strconv.Itoa(i))))
-	}
-	n := &recNotifier{}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-	require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 10*time.Millisecond)
-	n.mu.Lock()
-	msg := n.msgs[0]
-	n.mu.Unlock()
-	assert.Contains(t, msg, "removed 50 symlink(s)")
-	assert.Contains(t, msg, "and 40 more")
-	assert.Less(t, len(msg), 2500, "alert size is bounded regardless of how many links were planted")
-}
-
-// rejectNotifier refuses every message containing a marker, forever.
-type rejectNotifier struct {
-	recNotifier
-	reject string
-}
-
-// permanentErr is a rejection the notifier says can never succeed (like ntfy 413).
-type permanentErr struct{}
-
-func (permanentErr) Error() string   { return "413 payload too large" }
-func (permanentErr) Permanent() bool { return true }
-
-func (r *rejectNotifier) Notify(ctx context.Context, m string) error {
-	if strings.Contains(m, r.reject) {
-		r.mu.Lock()
-		r.msgs = append(r.msgs, "REJECTED")
-		r.mu.Unlock()
-		return permanentErr{}
-	}
-	return r.recNotifier.Notify(ctx, m)
-}
-
-// PR #160 review: one message the server will never accept blocked every
-// later alert forever (head-of-line). It is dropped after maxAttempts.
-func TestWatchDropsAMessageThatNeverSendsSoLaterAlertsFlow(t *testing.T) {
-	brain := t.TempDir()
-	n := &rejectNotifier{reject: "POISON"}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 10*time.Millisecond, n)
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "POISON")))
-	time.Sleep(30 * time.Millisecond)
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "later")))
-	require.Eventually(t, func() bool {
-		n.mu.Lock()
-		defer n.mu.Unlock()
-		for _, m := range n.msgs {
-			if strings.Contains(m, "later") {
-				return true
-			}
-		}
-		return false
-	}, 3*time.Second, 10*time.Millisecond, "a later alert is delivered despite a poison message")
-}
-
-// toggleNotifier fails while down is set.
-type toggleNotifier struct {
-	recNotifier
-	down atomic.Bool
-}
-
-func (r *toggleNotifier) Notify(ctx context.Context, m string) error {
-	if r.down.Load() {
-		return errors.New("ntfy down")
-	}
-	return r.recNotifier.Notify(ctx, m)
-}
-
-// PR #160 review: an outage queued one message per removal, unbounded. The
-// queue holds at most maxPending messages; the rest become one summary.
-func TestWatchQueueIsBoundedDuringAnOutage(t *testing.T) {
-	brain := t.TempDir()
-	n := &toggleNotifier{}
-	n.down.Store(true)
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 5*time.Millisecond, n)
-	for i := 0; i < 30; i++ {
-		link := filepath.Join(brain, "p"+strconv.Itoa(i))
-		require.NoError(t, os.Symlink(t.TempDir(), link))
-		require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond)
-	}
-	n.down.Store(false)
-	require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 5*time.Millisecond)
-	time.Sleep(100 * time.Millisecond)
-	assert.LessOrEqual(t, n.count(), maxPending+1, "queued alerts plus at most one suppression summary")
-	n.mu.Lock()
-	defer n.mu.Unlock()
-	found := false
-	for _, m := range n.msgs {
-		if strings.Contains(m, "suppressed") {
-			found = true
-		}
-	}
-	assert.True(t, found, "suppressed alerts are counted in a summary")
-}
-
-// Within the bound, every alert queued during an outage is delivered after
-// it, each naming its own removal (PR #160 review: truncating the queue after
-// the first send stayed green).
-func TestWatchDeliversEveryQueuedAlertAfterAShortOutage(t *testing.T) {
-	brain := t.TempDir()
-	n := &toggleNotifier{}
-	n.down.Store(true)
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 5*time.Millisecond, n)
-	for _, name := range []string{"qa", "qb", "qc"} {
-		link := filepath.Join(brain, name)
-		require.NoError(t, os.Symlink(t.TempDir(), link))
-		require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond)
-	}
-	n.down.Store(false)
-	require.Eventually(t, func() bool { return n.count() == 3 }, 2*time.Second, 5*time.Millisecond)
-	n.mu.Lock()
-	defer n.mu.Unlock()
-	for i, name := range []string{"qa", "qb", "qc"} {
-		assert.Contains(t, n.msgs[i], strconv.Quote(name))
-	}
-}
-
-// PR #160 round 2: dropping after N failures could not tell a poison message
-// from an outage, so a 15-minute outage silently lost the only incident. A
-// message is only dropped when it keeps failing while others send; through
-// an outage nothing is dropped, however long.
-func TestWatchLosesNoAlertThroughALongOutage(t *testing.T) {
-	brain := t.TempDir()
-	n := &toggleNotifier{}
-	n.down.Store(true)
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 2*time.Millisecond, n)
-	link := filepath.Join(brain, "only-incident")
-	require.NoError(t, os.Symlink(t.TempDir(), link))
-	require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 2*time.Millisecond)
-	time.Sleep(200 * time.Millisecond) // ~100 ticks of outage, far past any attempt limit
-	n.down.Store(false)
-	require.Eventually(t, func() bool { return n.count() >= 1 }, 2*time.Second, 5*time.Millisecond)
-	n.mu.Lock()
-	defer n.mu.Unlock()
-	assert.Contains(t, n.msgs[0], strconv.Quote("only-incident"))
-}
-
-// Every removal is either delivered or counted in the suppression summary:
-// none disappears from the record (round 2: 74 of 300 were in neither).
-func TestWatchAccountsForEveryRemovalDuringAnOutage(t *testing.T) {
-	brain := t.TempDir()
-	n := &toggleNotifier{}
-	n.down.Store(true)
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 2*time.Millisecond, n)
-	const plants = 25
-	for i := 0; i < plants; i++ {
-		link := filepath.Join(brain, "acct"+strconv.Itoa(i))
-		require.NoError(t, os.Symlink(t.TempDir(), link))
-		require.Eventually(t, func() bool { _, err := os.Lstat(link); return os.IsNotExist(err) }, 2*time.Second, 1*time.Millisecond)
-	}
-	n.down.Store(false)
-	require.Eventually(t, func() bool {
-		n.mu.Lock()
-		defer n.mu.Unlock()
-		for _, m := range n.msgs {
-			if strings.Contains(m, "suppressed") {
-				return true
-			}
-		}
-		return false
-	}, 2*time.Second, 5*time.Millisecond)
-	time.Sleep(50 * time.Millisecond)
-	n.mu.Lock()
-	defer n.mu.Unlock()
-	delivered, counted := 0, 0
-	for _, m := range n.msgs {
-		if strings.Contains(m, "removed 1 symlink(s)") {
-			delivered++
-		}
-		if i := strings.Index(m, "): "); i >= 0 && strings.Contains(m, "suppressed") {
-			_, _ = fmt.Sscanf(m[i+3:], "%d", &counted)
-		}
-	}
-	assert.Equal(t, plants, delivered+counted, "delivered %d + counted %d must equal %d removals", delivered, counted, plants)
-}
-
-// A poison message is still dropped (others send while it fails), and the
-// drop is counted in the summary.
-func TestWatchDropsAPoisonMessageAndCountsIt(t *testing.T) {
-	brain := t.TempDir()
-	n := &rejectNotifier{reject: "POISON"}
-	ctx, cancel := context.WithCancel(context.Background())
-	defer cancel()
-	go Watch(ctx, brain, 5*time.Millisecond, n)
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "POISON")))
-	time.Sleep(20 * time.Millisecond)
-	require.NoError(t, os.Symlink(t.TempDir(), filepath.Join(brain, "fine")))
-	require.Eventually(t, func() bool {
-		n.mu.Lock()
-		defer n.mu.Unlock()
-		for _, m := range n.msgs {
-			if strings.Contains(m, "1 further alert(s)") {
-				return true
-			}
-		}
-		return false
-	}, 3*time.Second, 5*time.Millisecond, "the dropped poison alert is counted in a summary")
-}

ingestion/internal/watcher/watcher.go

@@ -48,25 +48,8 @@ func Start(ctx context.Context, cfg Config) {
 func processDir(ctx context.Context, cfg Config, date string) []error {
 	rawDir := filepath.Join(cfg.BrainDir, "raw")
 
-	// Every read goes through an os.Root on raw/ (hyperguild#153): raw/ is
-	// writable via Syncthing, and a file swapped for a symlink between the walk
-	// and the read must not lead out of it. A symlinked raw/ is refused.
-	if fi, lerr := os.Lstat(rawDir); lerr != nil {
-		if os.IsNotExist(lerr) {
-			return nil
-		}
-		return []error{fmt.Errorf("stat raw dir: %w", lerr)}
-	} else if fi.Mode()&os.ModeSymlink != 0 || !fi.IsDir() {
-		return []error{fmt.Errorf("raw dir %s is not a real directory; refusing to ingest from it", rawDir)}
-	}
-	root, err := os.OpenRoot(rawDir)
-	if err != nil {
-		return []error{fmt.Errorf("open raw root: %w", err)}
-	}
-	defer func() { _ = root.Close() }()
-
 	var errs []error
-	err = filepath.WalkDir(rawDir, func(path string, d os.DirEntry, err error) error {
+	err := filepath.WalkDir(rawDir, func(path string, d os.DirEntry, err error) error {
 		if err != nil {
 			return err
 		}
@@ -103,10 +86,6 @@ func processDir(ctx context.Context, cfg Config, date string) []error {
 		if strings.HasPrefix(d.Name(), "tunnel-candidates-") {
 			return nil
 		}
-		// Only regular files are notes; a symlink's target may be anywhere.
-		if !d.Type().IsRegular() {
-			return nil
-		}
 
 		// Skip files that have already been processed or permanently failed.
 		if _, err := os.Stat(path + ".processed"); err == nil {
@@ -116,7 +95,7 @@ func processDir(ctx context.Context, cfg Config, date string) []error {
 			return nil
 		}
 
-		if err := processFile(ctx, cfg, root, rawDir, path, date); err != nil {
+		if err := processFile(ctx, cfg, path, date); err != nil {
 			errs = append(errs, fmt.Errorf("process %s: %w", filepath.Base(path), err))
 		}
 		return nil
@@ -130,15 +109,11 @@ func processDir(ctx context.Context, cfg Config, date string) []error {
 // processFile reads a file, calls pipeline.Run, copies it to processed/ or failed/,
 // and writes a marker file next to the original so the watcher skips it next poll.
 // The original file is never deleted, keeping Syncthing-connected vaults (e.g. Obsidian) intact.
-func processFile(ctx context.Context, cfg Config, root *os.Root, rawDir, path, date string) error {
+func processFile(ctx context.Context, cfg Config, path, date string) error {
 	filename := filepath.Base(path)
 	source := deriveSource(filename)
-	rel, err := filepath.Rel(rawDir, path)
-	if err != nil {
-		return fmt.Errorf("relative path: %w", err)
-	}
 
-	content, err := extract.TextIn(root, rel)
+	content, err := extract.Text(path)
 	if err != nil {
 		return fmt.Errorf("extract text: %w", err)
 	}
@@ -151,7 +126,7 @@ func processFile(ctx context.Context, cfg Config, root *os.Root, rawDir, path, d
 			return fmt.Errorf("mkdir failed dir: %w", mkErr)
 		}
 		dest := filepath.Join(failedDir, filename)
-		if cpErr := copyFromRoot(root, rel, dest); cpErr != nil {
+		if cpErr := copyFile(path, dest); cpErr != nil {
 			return fmt.Errorf("copy to failed: %w", cpErr)
 		}
 		if mkErr := os.WriteFile(path+".failed", []byte(runErr.Error()), 0o644); mkErr != nil {
@@ -179,7 +154,7 @@ func processFile(ctx context.Context, cfg Config, root *os.Root, rawDir, path, d
 		base := strings.TrimSuffix(filename, ext)
 		dest = filepath.Join(processedDir, base+"-"+time.Now().UTC().Format("150405")+ext)
 	}
-	if err := copyFromRoot(root, rel, dest); err != nil {
+	if err := copyFile(path, dest); err != nil {
 		return fmt.Errorf("copy to processed: %w", err)
 	}
 	if err := os.WriteFile(path+".processed", []byte(date), 0o644); err != nil {
@@ -190,6 +165,26 @@ func processFile(ctx context.Context, cfg Config, root *os.Root, rawDir, path, d
 	return nil
 }
 
+// copyFile copies src to dst, creating dst if it doesn't exist.
+func copyFile(src, dst string) error {
+	in, err := os.Open(src)
+	if err != nil {
+		return fmt.Errorf("open src: %w", err)
+	}
+	defer in.Close() //nolint:errcheck
+
+	out, err := os.Create(dst)
+	if err != nil {
+		return fmt.Errorf("create dst: %w", err)
+	}
+
+	if _, err := io.Copy(out, in); err != nil {
+		out.Close() //nolint:errcheck
+		return fmt.Errorf("copy: %w", err)
+	}
+	return out.Close()
+}
+
 // deriveSource turns a filename into a human-readable source name.
 // "shape-up-book.md" → "Shape Up Book"
 func deriveSource(filename string) string {
@@ -226,30 +221,3 @@ func appendWatcherLog(brainDir, filename string, runErr error, date string) erro
 	}
 	return f.Close()
 }
-
-// copyFromRoot archives a raw/ file read through the root, refusing anything
-// that is not a regular file, so a swapped-in symlink cannot copy an outside
-// file into raw/processed/ or raw/failed/ (hyperguild#153).
-func copyFromRoot(root *os.Root, rel, dst string) error {
-	fi, err := root.Lstat(rel)
-	if err != nil {
-		return err
-	}
-	if !fi.Mode().IsRegular() {
-		return fmt.Errorf("%s: not a regular file", rel)
-	}
-	in, _, err := extract.OpenRegular(root, rel)
-	if err != nil {
-		return err
-	}
-	defer func() { _ = in.Close() }()
-	out, err := os.Create(dst)
-	if err != nil {
-		return err
-	}
-	if _, err := io.Copy(out, in); err != nil {
-		_ = out.Close()
-		return err
-	}
-	return out.Close()
-}

ingestion/internal/watcher/watcher_test.go

@@ -7,10 +7,6 @@ import (
 	"fmt"
 	"os"
 	"path/filepath"
-	"strings"
-	"sync"
-	"sync/atomic"
-	"syscall"
 	"testing"
 	"time"
 
@@ -286,107 +282,3 @@ func TestProcessDir_SkipsTunnelCandidateFiles(t *testing.T) {
 	_, err = os.Stat(filepath.Join(brainDir, "raw", "processed", date, "tunnel-candidates-2026-07-19.md"))
 	assert.True(t, os.IsNotExist(err), "tunnel-candidates file should not be copied to processed/")
 }
-
-// hyperguild#153: raw/ is writable via Syncthing, and the watcher ingests
-// whatever lands there with no API call at all. WalkDir yields symlinked
-// files, and extract.Text followed them, so a symlink to any pod-readable
-// file would be extracted into the brain. The watcher now ingests only
-// regular files, read through an os.Root on raw/.
-func TestProcessDir_IngestsTheRealNoteAndSkipsTheSymlinkBesideIt(t *testing.T) {
-	brainDir := setupBrainDir(t)
-	outside := filepath.Join(t.TempDir(), "secret.md")
-	require.NoError(t, os.WriteFile(outside, []byte("SENTINEL-153-WATCHER"), 0o644))
-	// "a-" sorts before "real": a walk that stopped at the first symlink
-	// (filepath.SkipAll — one synced link halting all ingestion) would fail
-	// the call count below (PR #156 review).
-	require.NoError(t, os.Symlink(outside, filepath.Join(brainDir, "raw", "a-leak.md")))
-	require.NoError(t, os.WriteFile(filepath.Join(brainDir, "raw", "real.md"), []byte("real note"), 0o644))
-
-	var sawSentinel bool
-	calls := 0
-	cfg := Config{BrainDir: brainDir, Interval: time.Hour, Pipeline: pipeline.Config{
-		Complete: func(_ context.Context, _, user string) (string, error) {
-			calls++
-			if strings.Contains(user, "SENTINEL-153-WATCHER") {
-				sawSentinel = true
-			}
-			b, _ := json.Marshal([]pipeline.RawPage{{Title: "X", Type: "source", Subtype: "article", Content: "## S\n\nx\n"}})
-			return string(b), nil
-		},
-		Schema: "# Schema",
-	}}
-	errs := processDir(context.Background(), cfg, time.Now().UTC().Format("2006-01-02"))
-	assert.Empty(t, errs)
-	assert.False(t, sawSentinel, "the symlink target must never reach the pipeline")
-	assert.Equal(t, 1, calls, "exactly the real note is ingested; the symlink is not a note")
-}
-
-// A symlinked raw/ is refused outright: os.OpenRoot would otherwise follow
-// it and confine reads to wherever it points.
-func TestProcessDir_RefusesASymlinkedRawDir(t *testing.T) {
-	brainDir := t.TempDir()
-	elsewhere := t.TempDir()
-	require.NoError(t, os.WriteFile(filepath.Join(elsewhere, "x.md"), []byte("SENTINEL-153-RAWLINK"), 0o644))
-	require.NoError(t, os.Symlink(elsewhere, filepath.Join(brainDir, "raw")))
-	calls := 0
-	cfg := Config{BrainDir: brainDir, Interval: time.Hour, Pipeline: pipeline.Config{
-		Complete: func(context.Context, string, string) (string, error) { calls++; return "[]", nil },
-		Schema:   "# Schema",
-	}}
-	errs := processDir(context.Background(), cfg, time.Now().UTC().Format("2006-01-02"))
-	assert.NotEmpty(t, errs, "a symlinked raw/ must be reported, not silently ingested")
-	assert.Zero(t, calls)
-}
-
-// PR #156: a FIFO swapped into raw/ between the watcher's checks and its
-// read or archive copy must not hang the watcher, which runs unattended.
-func TestProcessDir_DoesNotHangUnderAFIFOSwapRace(t *testing.T) {
-	brainDir := setupBrainDir(t)
-	raw := filepath.Join(brainDir, "raw")
-	var stop atomic.Bool
-	var wg sync.WaitGroup
-	wg.Add(1)
-	// Awaited before t.TempDir's cleanup (hyperguild#162).
-	t.Cleanup(func() { stop.Store(true); wg.Wait() })
-	go func() {
-		defer wg.Done()
-		for i := 0; !stop.Load(); i++ {
-			tmp := filepath.Join(raw, ".swap")
-			_ = os.Remove(tmp)
-			if i%2 == 0 {
-				_ = os.WriteFile(tmp, []byte("x"), 0o644)
-			} else {
-				_ = syscall.Mkfifo(tmp, 0o644)
-			}
-			_ = os.Rename(tmp, filepath.Join(raw, "r.md"))
-		}
-	}()
-	cfg := Config{BrainDir: brainDir, Interval: time.Hour, Pipeline: pipeline.Config{
-		Complete: func(context.Context, string, string) (string, error) {
-			b, _ := json.Marshal([]pipeline.RawPage{{Title: "R", Type: "source", Subtype: "article", Content: "## S\n\nx\n"}})
-			return string(b), nil
-		},
-		Schema: "# Schema",
-	}}
-	var iters atomic.Int64
-	done := make(chan struct{})
-	go func() {
-		for deadline := time.Now().Add(1500 * time.Millisecond); time.Now().Before(deadline); {
-			_ = os.Remove(filepath.Join(raw, "r.md.processed"))
-			_ = os.Remove(filepath.Join(raw, "r.md.failed"))
-			_ = processDir(context.Background(), cfg, "2026-10-03")
-			iters.Add(1)
-		}
-		close(done)
-	}()
-	select {
-	case <-done:
-		stop.Store(true)
-	case <-time.After(8 * time.Second):
-		stop.Store(true)
-		if f, err := os.OpenFile(filepath.Join(raw, "r.md"), os.O_WRONLY|syscall.O_NONBLOCK, 0); err == nil {
-			_ = f.Close()
-		}
-		t.Fatalf("the watcher hung on a swapped-in FIFO after %d iterations", iters.Load())
-	}
-}