Repository navigation
Refactor: reading a thread returns a Transcript with its note - #23
Conversation
ChatStore.read(chat, ...) -> Transcript(messages, note) replaces read_thread. The staleness note is computed with the read, so no caller can print a thread and forget to say it may be incomplete. Transcript.render(text, oldest_first) owns JSON/text output; --text and --oldest-first were untested before. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VCgWWsJbTMzsnA2WJkmsjC
|
Review summary (all four priority invariants checked):
🤖 Generated with Claude Code |
|
Correction: the inline comments I referenced did not post, because those lines are not in the diff. Details: Blocker:
The README and SKILL.md say this is 'not editable by a program', but that is only a prose rule for everything except the
Either way, add a test for non-interactive plus Nit: README.md:149, SKILL.md:17 and the skill description. 🤖 Generated with Claude Code |
Stacked on #22.
Summary
The staleness note now comes with the read, so a caller can't print a thread without saying it may be incomplete.
Evidence
test_a_transcript_carries_its_own_staleness_note: chat.db mtime is backdated, and.notereports it (red → green).test_a_transcript_renders_as_json_or_text_newest_or_oldest_first(red → green).--textand--oldest-firsthad no tests before.Merge Danger
Door: two-way
Blast Radius: library callers.
ChatStore.read_threadis gone; use.read(...).messages. CLI output is unchanged.🤖 Generated with Claude Code
https://claude.ai/code/session_01VCgWWsJbTMzsnA2WJkmsjC