[getappz]

Status

Investigated and verified — no further code changes were needed; this branch (3 commits ahead of origin/master, working tree clean) already carries a complete, tested fix from a prior round of this same task:

Root cause (confirmed via code-level tracing, documented in item #261's comment thread): self_repair_or_gate has zero cross-machine coordination — each workstation keeps its own unsynced local item DB, so any workstation's run_review_sweep tick can independently dispatch self-repair for the same in-review PR after a CI failure. Two machines each believed they were first, both reached push_and_open_pr's success/label branch, and PR #688 ended up with both beacon: labels.

Fix (3 defense-in-depth layers, all in already-committed history):

1. supervisor.rs: claim_self_repair — a GitHub-comment TTL+heartbeat claim marker on the PR (reusing github::bridge::claim's existing issue-claim logic) gates self-repair dispatch, so only one workstation proceeds unless the other's claim is stale.
2. worktree/discovery.rs + github/pulls.rs: opened_by_agentflare — skips adopting a PR into a duplicate local item when its body already carries agentflare's own "for item #N via agentflare" stamp, closing the discovery-sweep race window.
3. worktree/mod.rs: recover_pr_after_failed_create — if pulls::create fails (duplicate-branch error from a true concurrent-create race), rechecks find_existing once and links to the winner's PR instead of erroring out or silently duplicating.

Verification run this session:

• cargo build --workspace --all-targets — clean
• cargo fmt --check — clean
• cargo clippy --locked --workspace --all-targets --all-features -- -D warnings -A unsafe_code -A clippy::pedantic (CI's exact flags) — clean, zero warnings
• cargo test --workspace — 1818 passed, 1 unrelated failure, 9 ignored

Concern (not blocking, not in scope): mcp_prompts::tests::optimize_review_returns_full_skill_body fails in this sandbox only — flare_code::state::set_active writes to ~/.local/state/agentflare/..., and this sandbox's writable-home-dirs allowlist (crates/agentflare-jobs/src/sandbox.rs:158) doesn't include .local/state, so the write hits EROFS. Confirmed by directly touching that path (read-only filesystem) and by the file's own doc comment referencing the same class of prior sandbox issues (#127, #106, #130, #120, #236). Untouched by this diff, unrelated to the beacon-label bug, and wouldn't fail in real CI where that path is writable. This is likely why prior rounds hit MAX_FIX_ROUNDS without converging — worth flagging separately, but out of scope to fix here.

Item #261's acceptance criteria (confirmed mechanism, fixed, regression tests) are all satisfied by the existing commits; nothing further to change.

---

Opened by claude-code on flared:51bb8de6c33b for item #261 via agentflare.

[getappz]

Tracking this PR for automated review (flared:c997d745ae66).

[coderabbitai]

Warning

Review limit reached

Next included review available in 9 minutes.

Check out review usage here.

View limit details

Limit details: You’ve used all 3 included reviews currently available. Your 41 included PR review attempts over the past 7 days set your current allowance at 3 reviews per hour.

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

Learn how review limits work.

Review configuration:

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Essentials

Run ID: e28fd7cd-adb9-4707-aa61-9b134386da51

📥 Commits

Reviewing files that changed from the base of the PR and between 7b08c38 and 646de5f.

📒 Files selected for processing (5)

• src/github/pulls.rs
• src/supervisor.rs
• src/supervisor_tests.rs
• src/worktree/discovery.rs
• src/worktree/mod.rs

---

_{Comment @coderabbitai help to get the list of available commands.}