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 UKGovernmentBEIS/inspect_evals --skill verify-scoring-changegit clone --depth 1 https://github.com/UKGovernmentBEIS/inspect_evalsWrote 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/ukgovernmentbeis/inspect_evals/verify-scoring-change)<a href="https://agentmods.dev/skills/ukgovernmentbeis/inspect_evals/verify-scoring-change"><img src="https://agentmods.dev/badge/skills/ukgovernmentbeis/inspect_evals/verify-scoring-change/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/ukgovernmentbeis/inspect_evals/verify-scoring-change"><img src="https://agentmods.dev/badge/skills/ukgovernmentbeis/inspect_evals/verify-scoring-change.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00095 | $0.05590 |
| Opus 5.5 | $0.00038 | $0.02236 |
| Sonnet 5.5 | $0.00019 | $0.01118 |
| Haiku 4.5 | $0.00010 | $0.00559 |
Grade A, and why
verify-scoring-change 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 — 305 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verify a Scoring Change
A scoring change moves reported numbers. The review has to establish three things: that the bug exists, that the fix removes it, and what else moves. The output is a verdict a maintainer can act on without repeating your work.
The seven questions to answer are the Scoring-Change Gate. The rules behind them are in BEST_PRACTICES.md. Read these four sections first:
- "Route sample-level failures to the right outcome"
- "Handle grader failures as instrument failures, not model failures"
- "Say what a metric does when its denominator is empty"
- "Dict-valued scores and metric registration"
The receipts rule
Every claim about how inspect_ai behaves must be checked against the installed pin before it goes in the verdict.
- Find the source with
uv run python -c "import inspect_ai, pathlib; print(pathlib.Path(inspect_ai.__file__).parent)". - Cite
file:line, or run a probe and paste its output. - A claim without a receipt is left out.
The same rule applies to the eval. Before judging, read its README changelog, its eval.yaml version, and git log --oneline -- src/inspect_evals/<eval>/<scorer file>. Whether a branch is reachable, whether a zero was a deliberate decision, and who produces the artefact being scored are answered by that history rather than by the diff.
Behaviours that have been cited wrongly from memory
references/framework-receipts.md lists the inspect_ai behaviours that reviewers have got wrong from memory, each with a file:line on the pin it was read on. Open it when a verdict is about to rely on how the framework treats empty inputs, nan, reducers, grader parse misses, redacted reasoning, sandbox exceptions, or mockllm. Line numbers drift between releases, so re-check any row you quote against the current pin.
Setup
-
Create a worktree on the PR head.
git fetch origin pull/<N>/head:pr-<N> && git worktree add .claude/worktrees/pr-<N> pr-<N>- Record the head SHA with
git rev-parse pr-<N>.
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.
- 1mo ago Changed 5b9e623a88a3
- 1mo ago Changed · +198 lines 7b1ae35b6ea2
- 1mo ago First seen · 107 lines · 95 tokens per session scan A e64e36c3a1eb
verify-scoring-change is a skill published in the GitHub repository UKGovernmentBEIS/inspect_evals (706 stars, last pushed yesterday), licensed MIT. It adds 95 tokens to every session and 5,590 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 2026-09-09.
Other skills, from other repositories
autoreview
Structured code review through the shared OpenClaw agent-skills installation.
omh-code-review
This is an OMH code-review workflow skill, projected for Agent Skills hosts (Claude Code, Codex, Cursor, opencode, OpenClaw, pi).
revdiff-plan
Review the last Codex assistant message (plan, analysis, or proposal) with inline annotations in a TUI overlay. Extracts the most recent response from Codex rollout files and opens it in revdiff for review and annotation. Activates on "revdiff-plan", "review plan with revdiff", "annotate plan", "review last response"…
asplos-writing-style
Use when drafting or revising ASPLOS prose — engineering the first two pages that rapid review actually reads, stating the cross-layer insight as one quotable sentence, structuring per-layer mechanism sections, writing bounded claims, and fitting everything into 11 self-contained pages that include figures and…
explore-unknowns
Guide the user through a quadrant walk that maps the unknowns of a task — open by listing the known knowns, then work through known unknowns, unknown knowns, and unknown unknowns one stage at a time, ending with a complete four-quadrant map in the user's hands. Use when a request is ambiguous or underspecified, the…
re0-merge
Review and land an external contribution the way this suite does: gate it against the thesis, land it with the author's credit intact, complete a new skill rather than merging it raw, then approve, credit, and explain before closing. Use when reviewing a pull request, as any collaborator or maintainer, not only the…