fix(llm): send generous max_tokens on every request

The copied OpenAI-compatible client sent no max_tokens. Thinking models
(qwen3, deepseek-r1) spend their budget on the reasoning trace and return
EMPTY content when max_tokens is unset, which the summarizer treats as an
error. ADR-004 says change Tapir's copy rather than the hyperguild upstream,
so set a generous default (8192) leaving room for both reasoning and output.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

internal/adapters/llm/client.go

@@ -17,11 +17,20 @@ import (
 	"time"
 )
 
+// defaultMaxTokens is sent on every request. Tapir CHANGES this from the
+// hyperguild copy (ADR-004 says change the copy, not the upstream): thinking
+// models (qwen3, deepseek-r1) spend their budget on reasoning and return EMPTY
+// content when max_tokens is unset or too low. A generous ceiling leaves room
+// for both the reasoning trace and the actual summary. See
+// docs/homelab-integration.md.
+const defaultMaxTokens = 8192
+
 // Client calls an OpenAI-compatible chat completions endpoint.
 type Client struct {
 	baseURL    string
 	apiKey     string
 	model      string
+	maxTokens  int
 	httpClient *http.Client
 }
 
@@ -31,6 +40,7 @@ func New(baseURL, apiKey, model string, timeout time.Duration) *Client {
 		baseURL:    strings.TrimRight(baseURL, "/"),
 		apiKey:     apiKey,
 		model:      model,
+		maxTokens:  defaultMaxTokens,
 		httpClient: &http.Client{Timeout: timeout},
 	}
 }
@@ -39,6 +49,7 @@ type chatRequest struct {
 	Model       string    `json:"model"`
 	Messages    []message `json:"messages"`
 	Temperature float64   `json:"temperature"`
+	MaxTokens   int       `json:"max_tokens,omitempty"`
 }
 
 type message struct {
@@ -62,6 +73,7 @@ func (c *Client) Complete(ctx context.Context, system, user string) (string, err
 			{Role: "user", Content: user},
 		},
 		Temperature: 0.2,
+		MaxTokens:   c.maxTokens,
 	}
 	b, err := json.Marshal(body)
 	if err != nil {