ADR-027 ships chat-only, trust-the-model for v1. The decision records a known
limitation: because some summaries were weak-model output, the user has reason
not to fully trust a chat answer's fidelity to the transcript, and v1 gives no
in-UI way to check. The model-switcher partially compensates (two models
disagreeing is itself a signal), but there is no ground-truth view.
Proposal (the natural v2 ADR-027 anticipates)
A "show source / view transcript" affordance on the chat panel that reveals the
stored transcript (ADR-021 already persists it — this is cheap).
Optional: highlight/anchor the passage an answer leans on (stretch).
Why deferred, not now
v1 is deliberately minimal. ADR-021's stored transcript makes this cheap whenever
demand is confirmed; recorded so it's a known next step, not a rediscovery.
Constraints
Still stored-transcript-only — viewing the transcript is a read of stored content,
never a fetch. No new persisted state required.
Source: ADR-027 §5 (chat-only, trust-the-model — recorded limitation).
## Context
ADR-027 ships chat-only, trust-the-model for v1. The decision records a known
limitation: because some summaries were weak-model output, the user has reason
not to fully trust a chat answer's fidelity to the transcript, and v1 gives no
in-UI way to check. The model-switcher partially compensates (two models
disagreeing is itself a signal), but there is no ground-truth view.
## Proposal (the natural v2 ADR-027 anticipates)
- A "show source / view transcript" affordance on the chat panel that reveals the
stored transcript (ADR-021 already persists it — this is cheap).
- Optional: highlight/anchor the passage an answer leans on (stretch).
## Why deferred, not now
v1 is deliberately minimal. ADR-021's stored transcript makes this cheap whenever
demand is confirmed; recorded so it's a known next step, not a rediscovery.
## Constraints
Still stored-transcript-only — viewing the transcript is a read of stored content,
never a fetch. No new persisted state required.
Source: ADR-027 §5 (chat-only, trust-the-model — recorded limitation).
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Context
ADR-027 ships chat-only, trust-the-model for v1. The decision records a known
limitation: because some summaries were weak-model output, the user has reason
not to fully trust a chat answer's fidelity to the transcript, and v1 gives no
in-UI way to check. The model-switcher partially compensates (two models
disagreeing is itself a signal), but there is no ground-truth view.
Proposal (the natural v2 ADR-027 anticipates)
stored transcript (ADR-021 already persists it — this is cheap).
Why deferred, not now
v1 is deliberately minimal. ADR-021's stored transcript makes this cheap whenever
demand is confirmed; recorded so it's a known next step, not a rediscovery.
Constraints
Still stored-transcript-only — viewing the transcript is a read of stored content,
never a fetch. No new persisted state required.
Source: ADR-027 §5 (chat-only, trust-the-model — recorded limitation).