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 ai-analyst-lab/ai-analyst-plugin --skill evalgit clone --depth 1 https://github.com/ai-analyst-lab/ai-analyst-pluginWrote 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/ai-analyst-lab/ai-analyst-plugin/eval)<a href="https://agentmods.dev/skills/ai-analyst-lab/ai-analyst-plugin/eval"><img src="https://agentmods.dev/badge/skills/ai-analyst-lab/ai-analyst-plugin/eval/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ai-analyst-lab/ai-analyst-plugin/eval"><img src="https://agentmods.dev/badge/skills/ai-analyst-lab/ai-analyst-plugin/eval.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00108 | $0.02319 |
| Opus 5.5 | $0.00043 | $0.00928 |
| Sonnet 5.5 | $0.00022 | $0.00464 |
| Haiku 4.5 | $0.00011 | $0.00232 |
Grade A, and why
eval 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 1mo 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 — 154 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Skill: Eval (live gold-suite runner)
Purpose
Run the analyst on every question in a held-out gold suite, then score the answers against the blind gold: accuracy (the analyst's number vs the recomputed gold), query similarity (its SQL vs the trusted query), and cost/latency. This is the system-level eval: the number that tells you whether the context you are adding is paying off, and the number a model comparison turns on.
Two honest properties:
- Blind by construction. The analyst runs see the question only, never the gold SQL or value. The gold is read only at grading, after the answers are locked.
- Real, not staged. Each answer is produced by actually running the analyst now. Nothing is pre-filled.
The gold-case file
The suite is a YAML file you author and keep in your working folder (for example
gold-cases.yaml). It never lives in shared context, because the analyst must not be able to see
it. Each case is a question you already know the right answer to, paired with the query you trust
and the value it returns:
cases:
- id: rev-2025-q4 # short unique id
question: "What was total net revenue in Q4 2025?"
split: train # train (the set you iterate on) or test (held out)
gold_sql: "select sum(net_revenue) from orders where order_date between '2025-10-01' and '2025-12-31'"
gold_value: 4823910.55 # what gold_sql returns; a reference point, recomputed at grading
# tolerance: 0.01 # optional relative tolerance; default 0.005 (0.5%)
The easiest way to start: take five queries your team already trusts (month-end numbers you have
reported, dashboard tiles you have verified) and record each as a case with the exact SQL and the
value it produces. Mark three or four of them train and keep at least one as test. Grading
recomputes the gold by re-running gold_sql against the live connection at eval time, so the
suite does not rot as new data arrives; the stored gold_value is a sanity reference.
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.
- 1mo ago First seen · 154 lines · 108 tokens per session scan A 9dcf47644faf
eval is a skill published in the GitHub repository ai-analyst-lab/ai-analyst-plugin (33 stars, last pushed 1mo ago), licensed MIT. It adds 108 tokens to every session and 2,319 once invoked, about $0.0004 per session on Opus 5.5. 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.
eval-dataset-design
A guide to creating reliable test sets for AI agents. Each evaluation task includes the request, the simulated user's behaviour, a reset starting state, and a success check that can be independently verified.
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".
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
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.