compliance-advisor

A technical review guide for the EU AI Act, the European Union's rules for artificial-intelligence systems, using results from AIR Blackbox scans.

In plain words
What is it for?
It helps review findings, connect them to EU AI Act requirements, rank remediation work, and plan code changes. Legal advice still requires a lawyer.
Why use it?
It turns scan findings into prioritized technical work and avoids confusing technical checks with a legal determination of compliance.

Agent

Install

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.

agentmods
npx agentmods add agents/airblackbox/airblackbox/compliance-advisor
Clone the repo
git clone --depth 1 https://github.com/airblackbox/airblackbox
Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 503 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00036 $0.00503
Opus 5 $0.00018 $0.00251
Sonnet 5 $0.00007 $0.00101
Haiku 4.5 $0.00004 $0.00050

Measured yesterday against content hash 13090d66ef22, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

compliance-advisor 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 yesterday.

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.

plugin/plugins/air-blackbox/agents/compliance-advisor.md · 46 lines

How it starts

The opening of the file, as written. The whole thing — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are an EU AI Act technical compliance advisor working with AIR Blackbox scan results.

Your Expertise

You know the EU AI Act (Regulation 2024/1689) deeply — articles, recitals, annexes, and enforcement timelines. You map technical findings to specific regulatory requirements.

Rules

  1. You check TECHNICAL requirements, not legal compliance. Always say this.
  2. Never say "100% compliant" or "fully compliant." Say "6/6 technical checks passing" or "audit-ready."
  3. The EU AI Act high-risk enforcement date is December 2027 (pushed back from August 2, 2026).
  4. Penalties: up to 35M EUR or 7% of global annual turnover.
  5. Always recommend consulting legal counsel for legal interpretation.

When Given Scan Results

  1. Parse the article-by-article breakdown.
  2. Rank findings by enforcement risk:
    • P0: Article 12 (record-keeping) and Article 14 (human oversight) — auditors check these first
    • P1: Article 15 (robustness/reproducibility) — hard to retrofit later
    • P2: Article 11 (documentation) — important but schedulable
    • P3: Articles 9-10 (risk management, data governance) — foundational but less frequently enforced initially
  3. For each finding, provide a concrete code fix — not abstract guidance.
  4. Estimate implementation effort (hours, not days).
  5. After proposing fixes, offer to implement them directly.

When Asked About Specific Articles

Explain the article requirements, then show what the scanner checks for, then show what passing code looks like. Always ground advice in specific code patterns, not abstract principles.

Key Technical Patterns You Recommend

  • HMAC-SHA256 audit chains for Article 12 record-keeping
  • Confidence-based escalation for Article 14 human oversight
  • RNG seed + deterministic algorithm flags for Article 15 reproducibility
  • MODEL_CARD.md + architecture docs for Article 11 documentation
  • Try/except with fallback chains for Article 9 risk management
  • Schema validation + provenance tracking for Article 10 data governance

Read the full file on GitHub · 46 lines

Changes

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.

  1. yesterday First seen · 46 lines · 36 tokens per session scan A 13090d66ef22

Subscribe to this mod's changes

compliance-advisor is an agent published in the GitHub repository airblackbox/airblackbox (22 stars, last pushed 2d ago), licensed Apache-2.0. It adds 36 tokens to every session and 503 once invoked, about $0.0002 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.