[bniladridas]

Human overview

A scoped agents claude-channel run now refuses a --self that disagrees with the scope frame's agent_person_id. The mismatch exits 2 and is rejected before the scope is bound or any subsequent message event is handled, so a wrong id cannot drop another member's posts or push the agent's own. Unscoped runs retain their existing --self and is_you behavior. Fixes #254, item 3 of 4 follow-ups from #122.

AI reviewer

Details for agents: what the diff does, invariants, verification, checklist

What it does

• ChannelWorker._accept_scope now compares a supplied --self with seen.agent_person_id using _same_id (UUID case-insensitive) before setting self.bound or handling events.
• A mismatch raises _Ended(2) with a message naming both ids. The channel emits its standard ended control event, but no message is fetched or pushed after the mismatch.
• The discovery path is unchanged: when --self is absent, self_id still comes from the scope frame, and unscoped streams still use --self or the is_you fallback.

Invariants

• The scope frame remains the source of truth for a scoped run; --self can only confirm it.
• The check is before self.bound = seen, so a mismatch does not bind the channel to the scope.
• No API, cursor-file, or contract changes.

Where to look hardest

• The check is inside the first-scope branch only. Reconnects still compare the whole scope against self.bound, so a changed agent_person_id still ends the channel.
• _same_id canonicalizes UUIDs, so case differences are not treated as a mismatch.

Verification

• uv run --extra dev pytest tests/test_claude_channel.py -q: 135 passed in 21.16s.
• uv run --extra dev pytest -q: 3344 passed, 2 skipped in 578.78s.
• uv run --extra dev ruff format --check: 150 files already formatted.
• uv run --extra dev ruff check: All checks passed.
• uv run --extra dev ty check src/pepper_cli/: All checks passed.
• Coverage was not run locally; CI measures it.

Checklist (repo-specific)

• uv run ruff format --check and uv run ruff check clean (no new lint ignores)
• uv run ty check src/pepper_cli/ clean
• uv run pytest passes; coverage floor in .claude/maintainerd.json not lowered (pytest passed; coverage not run locally, CI measures it)
• Server-contract changes: N/A, no /api/v1 or # MIRROR: changes
• No token or credential is logged, printed, or put in a fixture
• README and the bundled SKILL.md updated if commands or output changed (N/A: no command or output change)
• Help: help/cli.md updated for a new or changed command, or not user-facing because N/A, no new or changed command and the flag syntax is unchanged

Refs #254.

[vycari-review]

Vycari Review

Round 1 on 832befa. The verdict is the Vycari Review check run, and each open finding is an inline thread. Reply on every P0 and P1 thread: Fixed in <sha>, or Disputed: with a reason. P2s are the author's call: fix one, or reply Deferred: #<issue> (the PR's one follow-up issue) or Declined: with a reason, and VR stops re-raising it. From round 2 on, VR raises no new P2s: it checks the fixes and looks for P0s and P1s.

Open: P0 0 · P1 0 · P2 0 · disputed 0 · confirmed fixed 0 · withdrawn 0

Hand-off: Greptile requested on 832befa.

No open findings this round.

P2s, judge's notes and closed findings

Judge's notes

No candidate findings were raised.

[chatgpt-codex-connector]

Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-14T06:23:01.601091Z	832befa	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[vycari-review]

@greptileai review