triaging-feedback
Skill3 starsChanged 16 days agoMITView on GitHub
use when deciding which review feedback to address/skip
In this repository15 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-browser-diffSkill
- ai-code-review-catalogSkill
- claude-parallel-writersSkill
- cross-harness-sub-agentSkill
- helper-sub-agentsSkill
- llm-evaluationSkill
- loopSkill
- orchestrate-multi-pr-workSkill
- planSkill
- propose-refactorSkill
- prSkill
- record-videoSkill
- upgrade-dependenciesSkill
- writing-skillsSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 22
ask the user only when a verdict truly needs them: a product/UX/architecture call the goal doesn't answer, where a wrong guess is expensive …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
applies to any batch of review feedback: PR comments, reviewer reports, handoff findings
<what-to-do>- group feedback into unique comments
- cross-check each comment and its evidence against the current code; treat reported severity and confidence as inputs, not the address/skip decision
- label each comment against the original goal:
- "better": closes a gap between the code and the goal (defects, missed cases), or improves the code without changing core behavior (simplification, restructuring, code judo)
- "more": grows behavior beyond the goal - features, abstractions, capabilities, speculative hardening
- skip every "more" comment, however good the idea; record it in your report so it can be promoted later
- skip every
<comments-to-skip>comment (unless you're in round 1/2 of a /loop) - weigh each "better" comment as a trade:
- payoff: for fixes it's
severity * likelihood of occurrence - price: complexity, LOC, maintenance burden, code churn, scope creep, time to implement and review, risk of introducing new problems
- "address" when payoff > price, "skip" when payoff <= price
- payoff: for fixes it's
- feedback from humans outranks feedback from agents
- ask the user only when a verdict truly needs them: a product/UX/architecture call the goal doesn't answer, where a wrong guess is expensive to undo. anything else, decide
- done when every comment is marked "address" or "skip", with why
</what-to-do><comments-to-skip>- pre-existing issues that don't block the requested change
- equally valid style, naming, or type-encoding preferences
- non-blocking changes in non-production code (tests/fixtures/stories/comments etc.) and logging
- non-reproducible theoreticals
- accessibility polish
- speculative performance work, including indexes without evidence
</comments-to-skip>Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.