qa-review
Skill0 starsChanged 19 days agoMITView on GitHub
Review a support ticket, chat or email reply against a weighted QA scorecard and produce a 0 to 100 score with per criterion reasoning and coaching ready feedback. Use when the user pastes a support conversation and asks to QA it, grade it, review it, or check it against the scorecard.
12 other projects use a version of this.
Adopted by
Most recently taken up by
- drosam/agentstheir own version, first seen Nov 15, 2026
- fushinoryuu/claude_skillstheir own version, first seen Nov 1, 2026
- drvoss/everything-copilot-clitheir own version, first seen Oct 26, 2026
- TzolkinB/skillstheir own version, first seen Oct 22, 2026
- ieuanign/skillstheir own version, first seen Oct 13, 2026
- nimadorostkar/Claude-Skills-collectiontheir own version, first seen Oct 13, 2026
Other versions12 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 12
- nearai/ironclaw★ 13ksame name
- arjun988/blender-skills★ 2.2ksame name
- hellowind777/helloagents★ 704same name
- PackmindHub/packmind★ 321same name
- shuotao/REVIT_MCP_study★ 110same name
- drvoss/everything-copilot-cli★ 48same name
- nimadorostkar/Claude-Skills-collection★ 33same name
- ieuanign/skills★ 2same name
- TzolkinB/skills★ 2same name
- drosam/agents★ 1same name
2 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository3 other files it gives its agents.
- calibrationSkill
- coaching-noteSkill
- nesting-reviewSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 9 edited versions we read keep these.
2
Distinctive
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
- Distinctive: line 12
Read the whole conversation once before scoring.
- Distinctive: line 17
Write the feedback in SBI form (situation, behavior, impact), two strengths and one thing to change.
- Very specific: line 18
If the user gives a CSV of many reviews, do not score by hand.
“Common” is measured: kept by at least 6 in 10 of the 9 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
QA review
Score the reply the agent wrote, not the customer's behavior. Default scorecard: ${CLAUDE_PLUGIN_ROOT}/scorecards/default.yaml. If the user's repo or message contains another scorecard, use that one.
Procedure
- Read the whole conversation once before scoring. Identify what the customer actually asked (there is often more than one thing).
- For each criterion give 0, 1 or 2 and one sentence of evidence quoting the ticket. No evidence, no score.
- Auto fail criteria at 0 set the final score to 0. Say so plainly at the top.
- Compute: criterion points = weight × value / 2, then sum. Show the per category subtotal. Pass threshold is in the scorecard (default 85).
- For every criterion scored 0 or 1, tag the root cause with exactly one of:
agent(knew the right thing, did not do it),knowledge(did not know; KB or training gap),process(followed a process or macro that produced the bad outcome),tooling(system, access or data problem). Onlyagentandknowledgebelong in the agent's coaching note;processandtoolinggo to the team lead's own backlog. Say this split explicitly, it is the thing most QA programs get wrong: agents get coached for macro errors. - Write the feedback in SBI form (situation, behavior, impact), two strengths and one thing to change. The thing to change must be an observable behavior tied to a criterion id, not a trait.
- If the user gives a CSV of many reviews, do not score by hand; run
python3 ${CLAUDE_PLUGIN_ROOT}/scripts/score.py --scorecard <file> --input <csv> --by agent.
Output shape
Score: 78.5 / 100 (pass threshold 85) FAIL
Auto fail: none
Accuracy 28/35 · Communication 25/30 · Efficiency 13/20 · Empathy 12.5/15
correct_info 2 "the 14 day window is stated correctly"
resolved 1 "refund promised but no date given" root cause: agent
policy 1 "macro M12 still says 7 days, agent used it" root cause: process (fix the macro, not the agent)
...
Feedback for the agent
Strengths: ...
Change one thing: ...Do not soften a 0. Do not invent context the ticket does not contain. If a criterion cannot be judged from the text (for example internal notes you cannot see), mark it "not visible" and exclude its weight from the denominator, and say that the score is rescaled.
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.