feat: add docs-triage-guard skill to prevent auto-coding design docs - #153
Open
maruiz93 wants to merge 1 commit into
Open
feat: add docs-triage-guard skill to prevent auto-coding design docs#153maruiz93 wants to merge 1 commit into
maruiz93 wants to merge 1 commit into
Conversation
ADR and design-doc issues were being auto-promoted to ready-to-code via the documentation category, triggering the code agent on work that requires human discussion. This adds a skill that teaches the triage agent to classify these as category "other" with design-doc and documentation labels via label_actions, routing them to triaged instead. Uses base composition (ADR 0045) to extend the upstream triage harness with just the skill — no schema fork or post-script override. Replaces the approach from #149 (closed per review feedback). Relates to fullsend-ai/agents#260, fullsend-ai/agents#618. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> Signed-off-by: Marta Anon <manon@redhat.com>
PR Summary by QodoAdd docs-triage-guard skill to route design docs to human review
AI Description
Diagram
High-Level Assessment
Files changed (3)
|
Code Review by Qodo
Context used✅ Compliance rules (platform):
4 rules 1. Hash updater breaks
|
![qodo-code-review[bot]](https://avatars.githubusercontent.com/in/484649?s=60&v=4){kind=small}
Comment on lines
+43
to
+44
| - name: triage | ||
| source: harness/triage-custom.yaml |
There was a problem hiding this comment.
1. Hash updater breaks 🐞 Bug ⚙ Maintainability
hack/update-agent-hashes.sh greps any harness/*.yaml substring from config.yaml and then tries to fetch it from the fullsend-ai/agents repo; the new local source: harness/triage-custom.yaml matches this and will cause the script to curl a non-existent URL and exit non-zero. This breaks the repo’s standard workflow for bumping the pinned fullsend-ai/agents SHA.
Agent Prompt
## Issue description
`hack/update-agent-hashes.sh` currently assumes every `harness/*.yaml` occurrence in `config.yaml` is a remote path inside the `fullsend-ai/agents` repo, but `config.yaml` now includes a repo-local harness (`harness/triage-custom.yaml`). The script should only hash/update remote agents harness URLs (or at least skip local `source:` entries), otherwise `curl -sfL` fails and the script exits.
## Issue Context
The config change is intentional (local wrapper harness). The maintenance script should adapt so future SHA bumps remain smooth.
## Fix Focus Areas
- hack/update-agent-hashes.sh[14-24]
- config.yaml[43-44]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
Comment on lines
+1
to
+2
| base: https://raw.githubusercontent.com/fullsend-ai/agents/a0fec953cd3c8b877af9821551b503df88d394d8/harness/triage.yaml#sha256=f56cbb0a3c1cb71563aa1b2fb80ca9ecb922e4c832f886e8154e2e9beb6b56c6 | ||
| role: triage |
There was a problem hiding this comment.
2. Triage base sha drifts 🐞 Bug ☼ Reliability
harness/triage-custom.yaml pins its own base: URL to a specific fullsend-ai/agents commit+sha256, but hack/update-agent-hashes.sh only rewrites agent SHAs/hashes inside config.yaml. After this PR, bumping the agents SHA via the script will leave triage inheriting an older upstream triage harness (and/or require manual base+sha256 updates), causing version skew.
Agent Prompt
## Issue description
The new wrapper harness (`harness/triage-custom.yaml`) embeds a pinned `base:` URL to `fullsend-ai/agents/<sha>/...#sha256=...`. The existing bump script only updates pins found in `config.yaml`, so triage’s base stays on the old SHA when the rest of the repo is updated.
## Issue Context
This doesn’t necessarily break current runtime behavior, but it will cause drift on the next agents SHA bump.
## Fix Focus Areas
- harness/triage-custom.yaml[1-1]
- hack/update-agent-hashes.sh[12-26]
ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
docs-triage-guardskill that teaches the triage agent to detect ADR and design-doc issues and route them to human review instead of auto-codingcategory: "other"(which falls through totriagedin the post-script's*catch-all) combined withlabel_actionsto applydesign-docanddocumentationlabelsHow it works
category: "other"instead of"documentation", which routes totriagedvia the existing*)default case inpost-triage.shlabel_actionsto adddesign-docanddocumentationlabels for trackingcategory: "documentation"and auto-promote toready-to-codeas beforeWhy this approach
This replaces the approach from #149 (closed per review feedback — forking the post-script was too invasive). By working within the existing category routing, we avoid overriding the schema or post-script entirely.
A more general solution for user-defined categories is tracked in fullsend-ai/agents#618.
Files changed
config.yamlharness/triage-custom.yamlskills/docs-triage-guard/SKILL.mdTest plan
issue-labels)triaged+design-doc+documentationready-to-code+documentationtriaged+design-doc+documentationRelates to fullsend-ai/agents#260, fullsend-ai/agents#618
🤖 Generated with Claude Code