Repository navigation
chore: remove the QualityFlow harness from the fullsend config - #63
Conversation
The pinned QualityFlow source (qualityflow-fullsend@b26c947e) no longer resolves: harness/qualityflow.yaml references qualityflow/agents/qualityflow.md, which returns 404 at that commit, so the harness fails to load agents on every fullsend event in this repository. Drop the remote agent source and its allowed_remote_resources prefix; the repository falls back to the fleet default agents. The matching entry is removed from the fleet manifest in fullsend-ai/.fullsend so a sync does not re-add it. Assisted-by: Claude Signed-off-by: Wayne Sun <1912684+waynesun09@users.noreply.github.com>
PR Summary by QodoRemove broken QualityFlow harness from Fullsend configuration
AI Description
Diagram
High-Level Assessment
Files changed (1)
|
Code Review by Qodo🐞 Bugs (0) 📘 Rule violations (0) 📎 Requirement gaps (0)
Great, no issues found!Qodo reviewed your code and found no material issues that require reviewTip of the day💡 Did you know, you can pick a display mode instead of tweaking every Display option by hand |
|
🤖 Finished Review · ✅ Success · Started 11:58 PM UTC · Completed 12:05 AM UTC Commit: Runtime: claude · Model: opus → claude-opus-5-5 · Effort: high · Cost: $2.63 |
#189) fullsend-ai/experiments#63 removes the QualityFlow agent source and its allowed_remote_resources prefix from the repository config because the pinned source (qualityflow-fullsend@b26c947e) no longer resolves and the harness fails to load agents on every event there. Remove the same two keys from the manifest entry so the next sync converges to the same state instead of re-adding them. The entry keeps its org-wide create_issues target. Assisted-by: Claude Signed-off-by: Wayne Sun <1912684+waynesun09@users.noreply.github.com> Co-authored-by: Wayne Sun <1912684+waynesun09@users.noreply.github.com>
|
Risk Assessment: low (1/5) DetailsSingle 4-line config-only deletion with no security sensitivity, no protected paths, and no CI impact; minor multi-author contention and moderate file age raise Tier 2 slightly but the composite (1.13×0.62 + 1.57×0.38 ≈ 1.30) rounds to 1. |
|
Looks good to me |
|
🤖 Finished Retro · ✅ Success · Started 12:40 PM UTC · Completed 12:52 PM UTC Commit: Runtime: claude · Model: opus → claude-opus-5-5 · Effort: high · Cost: $3.55 |
Retro: PR #63 (remove QualityFlow harness)Outcome: went well. A human opened this 4-line config deletion. The fullsend review agent approved it (risk 1/5, no findings) in about 9 min, and the maintainer merged it about 12h later. No rework, and the review made the right call. Agents repo, from the review run log: Root cause of the broken harness (found during this retro):
Evidence for existing issues (no new proposal filed for these):
Housekeeping: experiments#55 and PR #56 (filed Aug 22 for this same 404) are still open. This PR resolves them, so they can probably be closed. One new proposal: check remote agent sources on PRs that change Proposals filed
|
Adopts this file into fullsend-ai/.fullsend/repos.yaml management (ADR 0122): the marker is the only change here, so the next fleet sync can rewrite the file to the manifest's rendering instead of stopping on "adoption required". Effective configuration change after adoption: gains the six fleet agents and status reactions (this repository ran code defaults). The broken QualityFlow source was removed separately in #63. Assisted-by: Claude Signed-off-by: Wayne Sun <1912684+waynesun09@users.noreply.github.com> Signed-off-by: Greg Allen <gallen@redhat.com> Signed-off-by: Codex <codex@openai.com>
The pinned QualityFlow source (qualityflow-fullsend@b26c947e) no longer resolves:
harness/qualityflow.yamlpoints atqualityflow/agents/qualityflow.md, which 404s at that commit, so the harness fails to load agents on every fullsend event here (see the Harness dispatch failure on #62).This removes the remote agent source and its
allowed_remote_resourcesprefix. The repository falls back to the fleet default agents.The experiments entry in the fleet manifest (fullsend-ai/.fullsend
repos.yaml) loses the same two keys in a companion PR, so the next sync does not re-add them. Independent of #62, which only adds the managed-file marker; the two apply cleanly in either order.