Agent Evaluation Reporting
Write agent evaluation reports that keep successes, failures, timeouts and invalid runs distinct and comparable
- Repository
- sickn33/agentic-awesome-skills
- Stars
- 49,023
- Licence
- MIT
- Files
- 1 file
What it does
When you have results from testing an AI agent, this skill makes the agent write up the results in a strict, transparent way. It separates runs the agent finished on its own from runs where a human helped, and keeps failures, timeouts and broken test runs visible instead of hiding them. It also shows the counts and denominators behind every percentage, so readers can check the numbers and see whether two agent setups were compared fairly.
Great for
- Engineers or evaluators who run benchmarks or regression tests on AI agents and need an honest summary
- Teams deciding whether a new agent version or prompt is really better than the old one
- People preparing pilot or production readiness reviews that need declared decision gates
Skip it if
- Simple one-off questions about a result with no need for a formal report, since the process is heavy and rigid
- Situations where no real run records exist: the skill does not recreate missing data or validate the evaluator, so output built on incomplete logs can look more precise than it is
- Concluding production safety from a success rate alone; the skill itself says statistical significance does not establish safety or user value
Try asking
- “Summarize these 120 agent runs into a report separating autonomous successes, assisted successes, failures, timeouts and invalid runs”
- “Compare our new prompt against the old one and tell me whether the comparison is valid before claiming an improvement”
- “Calculate first-attempt completion and eventual task completion from this retry log, with denominators shown”
- “Check whether we have enough declared readiness gates to say this agent is ready for production”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
agent-evaluation-reporting.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip agent-evaluation-reporting.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip agent-evaluation-reporting.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Provide the full run log including retries, invalid runs and intervention counts; the skill will not reconstruct missing records
- Define your readiness thresholds (minimum success, maximum timeout rate, latency bounds) before asking for the verdict, otherwise the result will be inconclusive
- Ask for the attempt-level and task-level metrics separately, and avoid calling attempt rates workflow completion
What's inside
- SKILL.md
Redistributed unmodified from sickn33/agentic-awesome-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.