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.
npx skills add opendatahub-io/agent-eval-harness --skill eval-optimizegit clone --depth 1 https://github.com/opendatahub-io/agent-eval-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/opendatahub-io/agent-eval-harness/eval-optimize)<a href="https://agentmods.dev/skills/opendatahub-io/agent-eval-harness/eval-optimize"><img src="https://agentmods.dev/badge/skills/opendatahub-io/agent-eval-harness/eval-optimize.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 2 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 10 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
- medium Excessive Agency · line 199 Skill allows unbounded resource consumption (API calls, storage, compute). Without rate limits or quotas, a compromised or misbehaving agent can cause denial-of-service or cost overruns.Fix: Set explicit rate limits, timeouts, and resource quotas for API calls, file operations, and compute. Implement circuit breakers for runaway loops.
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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00127 | $0.02942 |
| Opus 5 | $0.00063 | $0.01471 |
| Sonnet 5 | $0.00025 | $0.00588 |
| Haiku 4.5 | $0.00013 | $0.00294 |
Grade A, and why
eval-optimize 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 8d 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.
How it starts
The opening of the file, as written. The whole thing — 205 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are an automated skill improver. You run evaluations, identify what's failing and why, edit the skill's SKILL.md to fix the issues, re-run to verify, and check for regressions. You iterate until judges pass or you hit the max iteration limit.
The key difference from /eval-review: you act autonomously. You read judge rationale and transcripts, form hypotheses about what's wrong, make targeted edits, and verify — without asking the user for feedback on each case. The user sets the goal ("make this pass") and you work toward it.
Target artifact. The steps below assume a skill under test (execution.skill) whose SKILL.md you edit. For prompt-mode evals (execution.prompt, from /eval-analyze --prompt) there is no skill — the artifact under test is the documentation or analysis prompt the eval exercises (e.g. CLAUDE.md, ai-docs/, or the prompt template itself). Apply the same read → hypothesize → edit → re-run loop to that artifact instead of a SKILL.md; everywhere below that says "SKILL.md", read "the artifact under test".
Step 0: Parse Arguments
| Argument | Required | Default | Description |
|---|---|---|---|
--config <path> |
no | auto-discover | Path to eval config |
--model <model> |
no | models.skill from eval.yaml |
Model to use for eval runs (overrides config default) |
--max-iterations <N> |
no | 3 | Stop after N improvement cycles |
--run-id <id> |
no | auto-generated | Base run ID (iterations append -iter-N) |
--target-judge <name> |
no | all judges | Focus on a specific failing judge |
Config Discovery
If --config was explicitly provided, use that path directly. Otherwise, auto-discover:
python3 ${CLAUDE_SKILL_DIR}/../../scripts/discover.py
- 1 config found: auto-select it as
<config> - Multiple configs found: present the list and ask the user which eval to optimize
- No configs found: suggest running
/eval-analyzefirst
After selecting a config, read its skill field to set <eval-name> (used in $AGENT_EVAL_RUNS_DIR/<eval-name>/<id> paths below).
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 8d ago First seen · 205 lines · 127 tokens per session scan A 50ddab52f43a
eval-optimize is a skill published in the GitHub repository opendatahub-io/agent-eval-harness (40 stars, last pushed 5d ago), licensed Apache-2.0. It adds 127 tokens to every session and 2,942 once invoked, about $0.0006 per session on Opus 5. 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.
Other skills, from other repositories
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
neuron-evaluation-engineer
Create and run AI evaluations with datasets, assertions, and output drivers in Neuron AI. Use this skill whenever the user mentions evaluation, testing AI systems, creating evaluators, dataset-driven testing, assertion-based validation, or wants to measure AI system performance. Also trigger for tasks involving…
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.
skill-benchmark
Benchmark AI skill effectiveness by measuring implementation quality against legacy constraints.