eval-skill

A command for testing a coding skill with realistic scenarios and scoring the results. It compares the skill's guidance with expected behaviour and known failure patterns.

In plain words
What is it for?
It helps list available test cases, run them against a skill, and report whether each scenario passes, is partial, fails, or exposes a knowledge gap.
Why use it?
It reveals missing instructions or incorrect guidance before a skill is relied on in real work.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/javanc/homunculus/eval-skill
Clone the repo
git clone --depth 1 https://github.com/JavanC/Homunculus
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,246 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.01246
Opus 5 $0.00000 $0.00623
Sonnet 5 $0.00000 $0.00249
Haiku 4.5 $0.00000 $0.00125

Measured 2d ago against content hash 59087a32857f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

eval-skill scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

commands/eval-skill.md · 138 lines

How it starts

The opening of the file, as written. The whole thing — 138 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/eval-skill — Evaluate an Evolved Skill

Run scenario-based tests on a skill to measure its quality.

Steps

  1. List eval specs: ls homunculus/evolved/evals/*.eval.yaml 2>/dev/null
  2. If user specified a skill name, use that eval spec; otherwise let user choose
  3. Read the skill file (homunculus/evolved/skills/<name>.md) and its eval spec

Evaluation

For each scenario, act as a developer who doesn't know the answer — only reference the skill document. Then compare against expected_behavior and anti_patterns.

Results

Result Condition
PASS Skill guides all expected behaviors, no anti-patterns triggered
PARTIAL Skill guides some expected behaviors, or misses important details
FAIL Skill fails to guide correct behavior, or would cause anti-patterns
GAP Scenario knowledge is completely absent from skill

Report Format

🔬 Skill Eval: <name> v<version>
━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Scenario                Result   Notes
──────────────────────────────────────
<scenario.name>         PASS     -
<scenario.name>         PARTIAL  Missing X
<scenario.name>         FAIL     Would cause Y

━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Pass rate: X/Y (Z%)
Grade: ⭐⭐⭐⭐⭐ (>= 90)

Noise Tolerance

5pp rule: score delta < 5pp is statistical noise, not a real change. From Anthropic infrastructure noise research — environment variance alone can cause ±5pp swings.

  • |delta| < 5ppnot_significant (treat as flat)
  • delta >= 5pp → real improvement
  • delta <= -5pp → real regression

Multiple-Run Modes

Three optional flags reduce measurement noise. Use independently or combined.

--runs N (eliminate infra noise)

/eval-skill my-skill --runs 3 — runs the full eval N times independently, reports mean ± σ.

  • Eliminates session-to-session infrastructure variance
  • Recommended for nightly agent evals: --runs 3
  • Daily manual evals: default N=1 is fine
📊 Multi-Run Summary (runs=3):
Run 1: 85%  Run 2: 87%  Run 3: 83%
Mean: 85.0% | σ: 1.6pp
Verdict: not_significant vs baseline 84% (delta=1pp < 5pp threshold)

Read the full file on GitHub · 138 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 138 lines · 0 tokens per session scan A 59087a32857f

Subscribe to this mod's changes

eval-skill is a command published in the GitHub repository JavanC/Homunculus (15 stars, last pushed 3mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,246 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.