ai-act-governance
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| AIACT_USER | No | The user identity used to scope which departments and applications are visible (e.g. "j.nowak", "k.schmidt" for HR Director, "t.yilmaz" for Marketing). Users are defined in data/users.json. | |
| ANTHROPIC_API_KEY | No | Optional Anthropic API key used by the Claude classifier; without it, classify_use_case falls back to the keyword baseline and says so in the engine field. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Governance layer for the EU AI Act on top of the company's application inventory. Use it to find which AI use cases are high-risk or prohibited, which obligations apply, and which evidence is missing. Results are limited to the departments the current user may see. Assessments marked 'proposed' are not yet approved by compliance. classify_use_case gives a first proposal for a new use case; it is not a legal opinion and needs human review.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| whoamiA | Who the current user is and which departments they can see. |
| list_applicationsA | List applications in the user's scope with their highest AI Act tier. end_of_life_before is an ISO date, e.g. '2028-01-01' for apps retiring by end of 2027. |
| get_applicationB | Full record of one application: owner, lifecycle, data, AI use cases and their assessments. |
| obligations_for_use_caseA | AI Act obligations for one AI use case, when they apply, and whether evidence exists for each. |
| evidence_gapsA | AI use cases with missing compliance evidence, strictest tier first. tier: prohibited, high, limited or minimal. obligation: an ID or words from its title, e.g. 'oversight', 'logs', 'D-WORKERS'. end_of_life_before: ISO date. |
| portfolio_exposureA | Management summary: use cases per tier and department, prohibited systems in use, assessments awaiting review, and the size of the evidence gap. |
| classify_use_caseA | Propose an AI Act risk tier for a new AI use case described in plain language. ai_role: 'deployer' if the company uses a bought system, 'provider' if it built it. Uses Claude when an API key is configured, otherwise the keyword baseline. The result is a proposal that compliance must review before it enters the inventory. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| timeline | EU AI Act application dates, including the Digital Omnibus changes. |
TDQS
Scored across 7 tools
Each tool targets a distinct object: identity (whoami), application list/detail (list_applications, get_application), per-use-case obligation detail (obligations_for_use_case), gap aggregation (evidence_gaps), portfolio rollup (portfolio_exposure), and tier proposal (classify_use_case). The main potential confusion is list_applications vs portfolio_exposure and obligations_for_use_case vs evidence_gaps, but the descriptions clearly mark one as detail and the other as aggregate, so an agent can choose correctly.
All names are snake_case and readable, with action tools following verb_noun (list_applications, get_application, classify_use_case) and reporting tools following noun phrases (evidence_gaps, portfolio_exposure, obligations_for_use_case). The split between verb-led and noun-led conventions is a minor deviation, and whoami is a conventional outlier rather than a real inconsistency.
Seven tools is well within the sensible range and each one covers a distinct governance need (identity, inventory, detail, obligations, gaps, rollup, classification). Nothing appears redundant or tacked on for the stated compliance scope.
The surface is read-heavy: it can inspect applications, obligations, gaps and exposure, and classify_use_case proposes a tier, but there is no tool to persist that proposal, attach/refresh evidence, or update an application or assessment. This leaves a notable dead end, since the classify tool explicitly says compliance must review before it enters the inventory and no review or write path exists.