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 CSOAI-ORG/councilof-ai --skill evidence-bundlegit clone --depth 1 https://github.com/CSOAI-ORG/councilof-aiWrote 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/csoai-org/councilof-ai/evidence-bundle)<a href="https://agentmods.dev/skills/csoai-org/councilof-ai/evidence-bundle"><img src="https://agentmods.dev/badge/skills/csoai-org/councilof-ai/evidence-bundle/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/csoai-org/councilof-ai/evidence-bundle"><img src="https://agentmods.dev/badge/skills/csoai-org/councilof-ai/evidence-bundle.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 | $0.00157 | $0.00586 |
| Opus 5.5 | $0.00063 | $0.00234 |
| Sonnet 5.5 | $0.00031 | $0.00117 |
| Haiku 4.5 | $0.00016 | $0.00059 |
Grade A, and why
evidence-bundle 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 13d 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.
What it actually says
Evidence bundle preview: signed cards relevant to one obligation (free)
For ONE obligation (article-50, article-53 GPAI transparency, dora or cra) and an optional subject: the obligation record, its counsel-gate status and the already-signed measurement cards that are relevant to it, read from GET /api/evidence-bundle. The same module as the free MCP tool evidence_bundle_preview. Relevant-to is never a determination: no answer says satisfied or not satisfied, and zero relevant cards is answered EMPTY. Article 53 output is evidence for review, not a legal determination. The assembled OSCAL bundle is the x402 MCP tool evidence_bundle on https://councilof.ai/mcp; this skill never charges. Not a conformity assessment, certification, audit opinion or compliance determination; grants no status.
How to call it
This is an A2A skill of the Council of AI measurement agent (agent card: https://councilof.ai/.well-known/agent-card.json).
Send SendMessage to POST https://councilof.ai/api/a2a with header A2A-Version: 1.0; params.message carries a messageId and parts,
and one structured part selects the skill: {"data": {"skill": "evidence-bundle", "input": {...}}, "mediaType": "application/json"}.
Calling it needs no account, key or registration (see https://councilof.ai/auth.md).
Examples
- SendMessage with Part.data {"skill":"evidence-bundle","input":{"obligation":"article-50"}}
- SendMessage with Part.data {"skill":"evidence-bundle","input":{"obligation":"article-53","subject":"gpt"}}
Tags: evidence, oscal, eu-ai-act, dora, cra
Measurement, never certification: an answer is evidence about specific bytes at a specific time, not a grade or mark. Generated by scripts/build-agent-skills.mjs from the agent card; do not hand-edit.
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.
- 13d ago First seen · 26 lines · 157 tokens per session scan A 316f02854958
evidence-bundle is a skill published in the GitHub repository CSOAI-ORG/councilof-ai (3 stars, last pushed today), licensed MIT. It adds 157 tokens to every session and 586 once invoked, about $0.0006 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 2026-10-03.
Other skills, from other repositories
complyedge-eu-ai-act-runtime-deny
Runtime deny for EU AI Act Article 5 and Article 50 on this prompt or output via ComplyEdge TrustLint MCP. Use when the user wants agent I/O compliance enforcement, not system classification, FRIA forms, or C2PA watermarking.
eu-ai-act-compliance-mcp
EU AI Act compliance checking and risk classification. Automated compliance validation for high-risk AI systems with Article 11 technical documentation generation.
hold-triage
Review and process pending governance holds. Use when holds are waiting for human approval or when the user asks about blocked operations.
mima-close-gaps
Run the compliance copilot loop — observe posture, identify gaps, dry-run proposed evidence records where possible, present for human approval, then write approved records. Use when the user asks to close compliance gaps, improve their EU AI Act score, or fix failing governance gates.
mima-register-systems
Run the Art. 9 intake workflow — find every AI system without an airiskassessment record, gather the required fields, and register each one. Use when the user asks to register systems, fix Art. 9 gaps, or prepare for the EU AI Act deadline.
mima-quarterly-evidence
Run the quarterly evidence collection workflow — find the stalest controls by record type, collect completion details for each, dry-run to confirm controls earned, then write approved records. Use at the end of each quarter when a GRC manager has completed governance activities (risk assessments, model evaluations…