chore(deps): update experiments digest to 5def9f1

[renovate-fullsend]

This PR contains the following updates:

Package	Update	Change
experiments (changelog)	digest	cd6df4e → 5def9f1

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR has been generated by Mend Renovate.

[github-actions]

Site preview

Preview: https://ed34bdba-site.fullsend-ai.workers.dev

Commit: 6831c18bedae24a85c7099b90cf4997bc36440f7

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[fullsend-ai-retro]

🤖 Finished Retro · ✅ Success · Started 6:25 AM UTC · Completed 6:45 AM UTC
Commit: 6831c18 · View workflow run →

[fullsend-ai-retro]

Retro: PR #5228 — chore(deps): update experiments digest to 5def9f1

What happened

PR #5228 was a trivial 1-line submodule pointer update created by renovate-fullsend[bot], bumping the experiments git submodule from commit cd6df4e to 5def9f1. The PR was created at 04:16 UTC on 2026-07-17, approved by human reviewer rh-hemartin at 06:18 UTC, and merged at 06:23 UTC. Total time to resolution: ~2 hours.

Agent interaction: none. No triage, code, review, or fix agents ran on this PR. The fullsend shim workflow fired 3 times (on opened, pull_request_review/submitted, and closed), but only the closed event dispatched an agent — the retro agent (this run). The opened event exited early because renovate-fullsend[bot] fails the is_event_actor_authorized write-permission check that gates review dispatch. The pull_request_review event exited early because the review state was approved (not changes_requested).

Workflow quality assessment

The PR lifecycle was clean. A trivial bot dependency update was correctly identified as low-risk, manually approved by a human, and merged. CI passed on all checks (unit tests, build, e2e, functional tests, vouch). No rework iterations occurred.

The only inefficiency is this retro run itself. The retro agent (running on Opus with a 30-minute timeout) was dispatched to analyze a 1-line submodule pointer update that had zero agent interaction. There are no triage, code, review, or fix workflow runs to trace — nothing to retrospect on. The retro dispatch routing unconditionally sets STAGE="retro" for every pull_request_target/closed event (lines 280-282 of reusable-dispatch.yml and lines 212-214 of the scaffold dispatch.yml), with no guards for bot authorship, PR triviality, or prior agent involvement.

Existing issue coverage — no new proposals needed

This exact pattern (retro dispatched on a zero-agent-interaction bot dependency PR) is extensively tracked across 12+ open issues and 10+ closed issues in this repo. The most relevant open issues are:

• Skip retro agent for automated bot dependency PRs #3833 — Skip retro agent for automated bot dependency PRs
• Extend bot PR retro skip to use user.type detection instead of hardcoded bot list #5185 — Extend bot PR retro skip to use user.type detection instead of hardcoded bot list
• Evidence for fullsend#5185: retro dispatched on zero-agent-involvement bot digest PR #5220 — Evidence for Extend bot PR retro skip to use user.type detection instead of hardcoded bot list #5185: retro dispatched on zero-agent-involvement bot digest PR (literally this exact scenario)
• Skip retro agent for bot-authored trivial PRs with no agent interaction #3226 — Skip retro agent for bot-authored trivial PRs with no agent interaction
• Skip retro stage for automated dependency PRs (Renovate, Dependabot, MintMaker) #4006 — Skip retro stage for automated dependency PRs (Renovate, Dependabot, MintMaker)
• Skip fullsend agent dispatch for bot dependency PRs #5067 — Skip fullsend agent dispatch for bot dependency PRs

Additional overlapping open issues include #3317, #3647, #3951, #4154, #4177, #4975, #4989. Closed issues (#3614, #4388, #4501, #4513, #4650, #4661, #4974) suggest partial progress on specific bot types, but the core dispatch unconditional routing persists.

Roadmap discrepancy noted

The roadmap (docs/roadmap.md, line 37) lists "retro dispatch skip guards for bot dependency PRs" as shipped in June 2026, attributed to PRs #2679 and #2764. However, neither PR implemented retro skip guards: #2679 added label-based handoffs for bot-to-bot dispatch authorization, and #2764 improved API retry resilience with jitter. The dispatch code on main still shows the unconditional closed) STAGE="retro" routing with no bot/trivial filtering at any layer (dispatch routing, pre-retro.sh, or harness config). This is likely a roadmap attribution error rather than a revert — the referenced PRs delivered adjacent dispatch-hardening work, not the specific skip guard.

Recommendations (no proposals filed — all are covered by existing issues)

1. Prioritize the bot-PR retro skip guard. The cluster of 20+ issues suggests this has been identified repeatedly but not yet resolved. Issue Extend bot PR retro skip to use user.type detection instead of hardcoded bot list #5185 (extend to user.type detection) or Skip retro agent for automated bot dependency PRs #3833 (general bot skip) appear to be the canonical tracking issues.
2. Consolidate duplicate issues. The 20+ open issues on this topic create noise. Consider closing duplicates and consolidating evidence onto 1-2 canonical issues.
3. Fix the roadmap attribution. Line 37 of docs/roadmap.md incorrectly credits PRs fix(dispatch): add label-based handoffs for bot-to-bot dispatch paths #2679/fix(github): increase retry attempts and add jitter to backoff #2764 with delivering retro skip guards. This should be corrected to avoid confusion about feature completeness.