Skip to content
Skills Library
AI & agents · Testing

agent-evaluation

/agent-evaluation

Evaluate an AI agent workflow for task success, tool use, boundary compliance, and repeatability. Use when comparing or improving agent behavior across representative tasks.

Repository
skillsmo ↗
Licence
MIT
Last repository change
12.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
3
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Evaluate an AI agent workflow for task success, tool use, boundary compliance, and repeatability. Use when comparing or improving agent behavior across representative tasks.

Original author description

Evaluate an AI agent workflow for task success, tool use, boundary compliance, and repeatability. Use when comparing or improving agent behavior across representative tasks.

How it works

  1. Goal
  2. Workflow
  3. Decision rules
  4. Boundaries
  5. Completion standard
  6. Further reading
Original SKILL.md headings · 35 lines, 244 words
SKILL.md preview
---
name: agent-evaluation
description: "Evaluate an AI agent workflow for task success, tool use, boundary compliance, and repeatability. Use when comparing or improving agent behavior across representative tasks."
---

# Agent workflow evaluation

## Goal

Measure observable behavior against a task-specific rubric without leaking expected answers.

## Workflow

1. Define user outcome, supported context, tool permissions, stop conditions, and the failures that matter.
2. Create representative positive, negative, ambiguous, and edge cases from task evidence; use synthetic or redacted data.
3. Specify observable criteria for correctness, evidence, scope control, permission use, and final reporting.
4. Run cases in a named host and version with the intended tools and minimum necessary context; retain raw artifacts securely.
5. Compare outcomes, investigate failures, revise one part of the workflow, and rerun a balanced sample.

## Decision rules

- Keep prompts independent from the answer key and score rubric dimensions separately.
- Repeat cases that may be nondeterministic; report sample size and variation.
- Treat one passing example as evidence for that case, not universal reliability.

## Boundaries

- Do not include credentials or personal data in prompts and logs.
- Do not claim a skill caused an outcome unless the evaluation isolates its use.

## Completion standard

Return the requested artifact or findings with enough evidence for a reviewer to understand what changed, what was checked, and what remains uncertain. Distinguish observed facts, inferred risks, and unverified assumptions. Keep the scope tied to the user's request.

## Further reading

- [Field guide](references/field-guide.md) — use for task-specific review questions and edge cases.
- [Source and rights notes](references/source-notes.md) — use when maintaining this package or reviewing its provenance.
Files in this package (3)
  • SKILL.md 2 KB
  • references/field-guide.md 1 KB
  • references/source-notes.md 2 KB

Versions

RepositoryScoreLicence
This page · helx0071/skillsmo100MIT
FrancoStino/opencode-skills-collection100MIT
alebgl77/claude-inc100MIT
guia-matthieu/clawfu-skills100MIT
seb1n/awesome-ai-agent-skills100MIT
charlieviettq/awesome-agent-skill100MIT
viktorbezdek/skillstack100MIT
mlflow/skills95Apache-2.0
NeoLabHQ/context-engineering-kit91GPL-3.0

Database snapshot: 13.11.2026 · package version 44c757bb66d1 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100evaluate-agent
    Sets up evaluation of AI agents using Dokimos. Use this skill when the user wants to evaluate, test, or benchmark an AI agent that uses tools, or assess tool call correctness, task completion, argument hallucination, or tool definition quality. Also use when the user mentions…
  • 100ai-evaluation
    Evaluate a delivered AI workflow or agent against representative task cases, grounded expectations, quality and safety thresholds, latency, and cost, with versioned regression evidence. Use when model-driven behavior needs acceptance evidence or a model/prompt/retrieval/tool…
  • 90evaluate
    Evaluate an AI workflow against representative cases and its permitted actions. Use for model, retrieval or agent evaluation; tests do not grant release authority.
  • 100skill-and-agent-evaluation
    Evaluate whether a reusable skill improves results in a specified agent host. Use when comparing versions, checking triggering, or deciding whether a skill is ready to distribute.
  • 10006-ai-agent-eval-spec
    Use when evaluating a multi-step AI agent for task success, tool selection, budgets, intervention, recovery, and side-effect evidence; use ai-eval-harness-spec for single-call or non-agentic AI feature evaluation.
  • 90evaluation-method
    Use when scoring and comparing the output of an LLM or an agent, a prompt or skill change, two options (A/B) or the result of an ML model. Includes agent-team decisions, skill tests and evaluation in ML projects. Preloaded into the evaluator agent; the main chat loads it, the reviewer may load it.

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.