# MCP Tools Reference

> Complete reference for all 47 Attestix MCP tools across 9 modules, with parameters, return types, and EU AI Act article mapping.

MCP tools exposed via the Model Context Protocol. Grouped by the 9 modules. For REST and Python library signatures see the [API reference](/docs/reference/api-reference).

Every tool is discoverable by any MCP-compatible client (Claude Desktop, Cursor, Continue, VS Code, Windsurf) once Attestix is installed:

```bash
pip install attestix
attestix mcp
```

## 01 / Identity

Unified Agent Identity Tokens bridging MCP OAuth, A2A, DIDs, and API keys.

| Tool | Purpose |
|------|---------|
| `create_agent_identity` | Issue a UAIT with a fresh did:key and Ed25519 keypair |
| `get_identity` | Retrieve a stored agent identity by ID |
| `list_identities` | List all agent identities in the local store |
| `verify_identity` | Verify an identity's Ed25519 signature against its DID |
| `revoke_identity` | Revoke an identity (soft delete with audit entry) |
| `translate_identity` | Translate between UAIT formats (MCP, A2A, DID, OAuth) |
| `resolve_identity` | Resolve an agent by DID or agent card URL |
| `purge_agent_data` | Local erasure: remove managed records and parseable local backups; external copies are separate |

## 02 / Agent Cards

A2A-compatible agent card generation and discovery.

| Tool | Purpose |
|------|---------|
| `generate_agent_card` | Generate an A2A agent card JSON from an identity |
| `parse_agent_card` | Parse and validate a remote agent card |
| `discover_agent` | Fetch `/.well-known/agent.json` from a URL |

## 03 / DID

W3C Decentralized Identifiers.

| Tool | Purpose |
|------|---------|
| `create_did_key` | Create a `did:key` identifier with Ed25519VerificationKey2020 |
| `create_did_web` | Create a `did:web` identifier for a domain |
| `resolve_did` | Resolve any DID method (key, web, universal resolver) |

## 04 / Delegation

UCAN v0.9 capability delegation with signed JWTs.

| Tool | Purpose |
|------|---------|
| `create_delegation` | Issue a UCAN delegating capabilities with attenuation |
| `verify_delegation` | Verify a UCAN chain end-to-end |
| `list_delegations` | List delegations issued by an agent |
| `revoke_delegation` | Revoke a delegation (propagates to children) |

## 05 / Reputation

Recency-weighted trust scoring.

| Tool | Purpose |
|------|---------|
| `record_interaction` | Record a signed interaction outcome for scoring |
| `get_reputation` | Get the current reputation for an agent |
| `query_reputation` | Search and filter reputation across a cohort |

## 06 / Compliance

EU AI Act risk profiles, Article 43 conformity, Annex V declarations.

| Tool | Purpose |
|------|---------|
| `create_compliance_profile` | Classify risk and unfold obligations per tier |
| `get_compliance_profile` | Retrieve a stored compliance profile |
| `list_compliance_profiles` | List all compliance profiles |
| `record_conformity_assessment` | Record Article 43 assessment (blocks self on high-risk) |
| `get_compliance_status` | Get the current completeness status |
| `generate_declaration_of_conformity` | Generate Annex V declaration as a W3C VC |
| `update_compliance_profile` | Iteratively update an existing profile (only non-empty fields change) |

## 07 / Credentials

W3C Verifiable Credentials 1.1 with Ed25519Signature2020.

| Tool | Purpose |
|------|---------|
| `issue_credential` | Issue a W3C VC with Ed25519Signature2020 proof |
| `verify_credential` | Verify a VC signature offline |
| `list_credentials` | List credentials for a subject or issuer |
| `get_credential` | Retrieve a credential by URN |
| `revoke_credential` | Revoke a credential with reason code |
| `create_verifiable_presentation` | Bundle VCs into a VP with audience binding |
| `verify_presentation` | Verify a VP including replay protection |
| `verify_credential_external` | Verify a VC supplied as raw JSON, without local storage |

## 08 / Provenance

Article 10 data governance, Article 11 lineage, Article 12 audit trail.

| Tool | Purpose |
|------|---------|
| `record_training_data` | Document training dataset with bias tests |
| `record_model_lineage` | Record base model, fine-tunes, eval metrics |
| `log_action` | Append an action to the hash-chained audit trail |
| `get_provenance` | Get the provenance record for an agent |
| `get_audit_trail` | Retrieve the audit trail, optionally scoped |

## 09 / Blockchain

Anchor to Base L2 testnet via Ethereum Attestation Service. Mainnet schema registration is planned.

| Tool | Purpose |
|------|---------|
| `anchor_identity` | Anchor an identity hash to Base L2 testnet |
| `anchor_credential` | Anchor a credential hash to Base L2 testnet |
| `anchor_audit_batch` | Merkle-batch audit entries and anchor the root |
| `get_anchor_status` | Query the on-chain status of an anchor |
| `verify_anchor` | Verify an anchor resolves to a local artefact |
| `estimate_anchor_cost` | Estimate gas cost for a pending anchor |

## Claim the MCP server

Clients that support MCP servers can add Attestix via `stdio`:

```json
{
  "mcpServers": {
    "attestix": {
      "command": "attestix",
      "args": ["mcp"]
    }
  }
}
```

Or connect over HTTP with `attestix mcp --transport http --port 8501`.

See the [integration guides](/docs/guides/integration-guide) for framework-specific wiring into LangChain, OpenAI Agents SDK, and CrewAI.