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 metraton/gaia --skill verification-rubricgit clone --depth 1 https://github.com/metraton/gaiaWrote 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/metraton/gaia/verification-rubric)<a href="https://agentmods.dev/skills/metraton/gaia/verification-rubric"><img src="https://agentmods.dev/badge/skills/metraton/gaia/verification-rubric/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/metraton/gaia/verification-rubric"><img src="https://agentmods.dev/badge/skills/metraton/gaia/verification-rubric.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 | $0.00061 | $0.01427 |
| Opus 5 | $0.00030 | $0.00714 |
| Sonnet 5 | $0.00012 | $0.00285 |
| Haiku 4.5 | $0.00006 | $0.00143 |
Grade A, and why
verification-rubric 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 10d 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 — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verification Rubric
Verification-rubric is the judging discipline for a task_gates entry whose
verification_type is semantic or self_review: it reads the gate's
evidence_shape as an explicit rubric, assesses the produced work against
each stated criterion independently, and emits a structured, justified
pass/fail verdict -- never a hollow "looks fine." It is the judgment-based
counterpart to the deterministic-oracle mode, which re-runs a command/code
gate and diffs the result; there is no command to re-run here, only evidence
to weigh against criteria.
Core principle
Two of the four verification_type values (gaia.state.VALID_VERIFICATION_TYPES,
gaia/state/__init__.py) route to this skill: semantic (evidence_shape
carries an externally authored rubric -- the contract-envelope's own
requires_human marker names the same discipline: "needs human/rubric
validation") and self_review (evidence_shape is the producer's own
statement of what it checked -- the envelope's reviewed field is the same
idea, and it is the honesty-rule floor from agent-protocol, not a shortcut
around it). Both are LLM-as-judge tasks, not code execution.
Three forces shape the judgment:
- Explicit criteria beat holistic impression. Read the rubric as a list of discrete, independently-checkable claims, not one paragraph to eyeball. A verdict built on an overall feeling collapses the moment someone asks "which part, exactly?"
- Judge each criterion on its own -- resist position and verbosity bias. Do not let a strong criterion compensate for a failing one (verdict inflation), do not let whichever criterion you read first anchor the whole read (position bias), and do not treat a longer or more polished artifact as automatically meeting more criteria (verbosity bias) -- effort and length are not what the rubric asks for.
- Confirmed beats assumed. A criterion is "met" only when the produced
work was actually observed to satisfy it. An unobserved criterion is not
met -- it is unknown, and unknown is not pass (mirrors
investigation's confirmed-vs-assumed line).
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.
- 10d ago First seen · 110 lines · 61 tokens per session scan A f01abc6808b6
verification-rubric is a skill published in the GitHub repository metraton/gaia (3 stars, last pushed today), licensed MIT. It adds 61 tokens to every session and 1,427 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
review-team
A multi-reviewer code review process that checks a change from several specialist viewpoints and combines the results into one report. It can cover bugs, security, tests, dependencies, frontend behavior, and continuous-integration workflows.
ultra
Fans the work out as a fleet of parallel Grok and Codex agents billed to their own subscriptions, then synthesizes one result. The peer engine equivalent of ultracode, adding intensity without spending Claude quota on the fleet. Use it for genuinely broad goals, not only explicit asks for intensity.
grok-prompting
Brief writing guidance for composing self contained Grok briefs for coding, review, diagnosis, and second opinion tasks.
smoke
Runs a three probe live smoke wave after a plugin update and reports gate chain health before real work rides it.
codex-result-handling
Internal contract for returning Codex companion output without alteration.
shipwrights-loop
Drive multiple Jira tickets sequentially through the /shipwrights-epic pipeline. Auto-picks the next ticket, watches the PR until merged, transitions Jira to shipped, repeats. Resumable across Claude Code sessions via .shipwrights/loop-state.json. Invoked as /shipwrights-loop [N] | --status | --abort.