Attestix

Record an interaction outcome and update the agent's trust score. Args: agent_id: The agent being evaluated. counterparty_id: The other party in the interaction. outcome: One of: success, failure, partial, timeout. category: Interaction category (e.g., task, delegation, general). details: Optional description of what happened.

Create an EU AI Act compliance profile for an agent. Categorizes the AI system by risk level and tracks all compliance obligations. Args: agent_id: The Attestix agent ID (e.g., attestix:abc123...). risk_category: EU AI Act risk level: minimal, limited, or high. provider_name: Name of the AI system provider/company. intended_purpose: What the AI system is designed to do. transparency_obligations: How transparency requirements are met. human_oversight_measures: Human oversight mechanisms in place (required for high-risk). provider_address: Registered address of the provider. authorised_representative: Name of the EU authorised representative (if provider is outside EU). annex_iii_category: For high-risk systems, the Annex III point 1-8. Point 1 (biometrics) requires third-party assessment; Points 2-8 permit self-assessment via Annex VI internal control. Use 0 or omit to leave unspecified (defaults to requiring third-party). annex_iii_exception: For Points 2-8, one of the documented carve-outs that still require third-party assessment per Article 43: 'biometric_identification_law_enforcement', 'financial_fraud_detection', 'political_campaign_organization'. Leave empty if no exception applies.

Update an existing EU AI Act compliance profile. Use this to iteratively fill in compliance information over time. Only non-empty fields will be updated. Args: agent_id: The Attestix agent ID. intended_purpose: Updated purpose description. transparency_obligations: Updated transparency info. human_oversight_measures: Updated human oversight info. provider_name: Updated provider name.

Get the EU AI Act compliance profile for an agent. Args: agent_id: The Attestix agent ID.

Get compliance gap analysis: what's done and what's still needed. Shows completion percentage and lists completed vs missing requirements. Args: agent_id: The Attestix agent ID.

Record a conformity assessment for EU AI Act Article 43. High-risk systems require third_party assessment. Limited/minimal can use self-assessment. Args: agent_id: The Attestix agent ID. assessment_type: self or third_party. assessor_name: Name of the assessor or notified body. result: Assessment result: pass, conditional, or fail. findings: Detailed findings or conditions. ce_marking_eligible: Whether CE marking can be applied (only if result is pass).

Generate an EU AI Act Annex V Declaration of Conformity. Requires a passing conformity assessment. Also issues a W3C Verifiable Credential as cryptographic proof of compliance. Args: agent_id: The Attestix agent ID.

List EU AI Act compliance profiles with optional filters. Args: risk_category: Filter by risk level (minimal, limited, high, unacceptable). Empty = all. compliant_only: Only return profiles with completed declarations. limit: Maximum number of results.

Create a UCAN-style delegation token granting capabilities from one agent to another. Args: issuer_agent_id: The agent granting capabilities (Attestix agent ID). audience_agent_id: The agent receiving capabilities (Attestix agent ID). capabilities: Comma-separated list of capabilities to delegate. expiry_hours: Hours until the delegation expires (default 24).

Verify a UCAN delegation token's signature, expiry, and structure. Args: token: The JWT delegation token to verify.

Revoke a UCAN delegation token by its JTI (JWT ID). Once revoked, the token will fail verification even if not expired. Args: jti: The unique identifier of the delegation (from create_delegation response). reason: Why this delegation is being revoked.

List delegation tokens, optionally filtered by agent and role. Args: agent_id: Filter by agent ID (empty = all). role: Filter role: 'issuer', 'audience', or 'any'. include_expired: Whether to include expired delegations.

Get the trust score and interaction history for an agent. Returns a recency-weighted score from 0.0 to 1.0 with category breakdown. Args: agent_id: The Attestix agent ID to check.

Search for agents by reputation criteria. Args: min_score: Minimum trust score (0.0 - 1.0). max_score: Maximum trust score (0.0 - 1.0). min_interactions: Minimum number of recorded interactions. category: Filter by interaction category (empty = all). limit: Maximum results to return.

Parse an A2A Agent Card JSON string into normalized fields. Args: agent_card_json: Raw JSON string of an A2A Agent Card.

Generate a valid A2A Agent Card (agent.json) for hosting. Args: name: Agent display name. url: Base URL where the agent will be hosted. description: What the agent does. skills_json: JSON array of skill objects with id, name, description. version: Agent version string.

Fetch and parse /.well-known/agent.json from a URL. Args: base_url: The base URL of the agent to discover (e.g., https://agent.example.com).

Anchor an agent identity (UAIT) hash to Base L2 via EAS. Computes SHA-256 of the full identity record and submits an on-chain attestation. Returns transaction hash and explorer URL. Args: agent_id: The Attestix agent ID (e.g., attestix:abc123...).

Anchor a Verifiable Credential hash to Base L2 via EAS. Computes SHA-256 of the full VC and submits an on-chain attestation. Returns transaction hash and explorer URL. Args: credential_id: The credential URN (e.g., urn:uuid:...).

Verify an on-chain anchor for an artifact. Provide either artifact_id (to look up and hash the artifact) or artifact_hash (direct SHA-256 hex) to check against on-chain records. Args: artifact_id: The artifact ID (agent ID or credential URN). Will be looked up and hashed. artifact_hash: Direct SHA-256 hex hash to verify. Use if you already have the hash.

Anchor a batch of audit log entries as a Merkle root to Base L2. Computes a Merkle tree of all audit log entries for the agent in the given date range and anchors the root hash on-chain. This provides tamper-evident proof for Article 12 audit trails. Args: agent_id: The Attestix agent ID. start_date: ISO date for start of range (e.g., 2026-01-01T00:00:00). Empty = all. end_date: ISO date for end of range. Empty = all.

Get all on-chain anchors for an agent, grouped by type. Shows identity anchors, credential anchors, and audit batch anchors with transaction hashes, explorer URLs, and timestamps. Args: agent_id: The Attestix agent ID.

Estimate gas cost for an anchoring transaction on Base L2. Shows current gas price, wallet balance, estimated cost in ETH, and whether the wallet has sufficient funds. Args: artifact_type: Type of artifact to estimate for (identity, credential, declaration, audit_batch).

Issue a W3C Verifiable Credential with Ed25519Signature2020 proof. Args: agent_id: The Attestix agent ID this credential is about. credential_type: Credential type (e.g., EUAIActComplianceCredential, AgentIdentityCredential). issuer_name: Name of the issuing authority. claims_json: JSON string of credential claims (e.g., {"compliant": true, "risk_level": "high"}). expiry_days: Days until credential expires (default 365).

Verify a Verifiable Credential: check signature, expiry, and revocation. Args: credential_id: The credential URN (e.g., urn:uuid:...).

Revoke a Verifiable Credential. Args: credential_id: The credential URN to revoke. reason: Why the credential is being revoked.

Get full details of a Verifiable Credential. Args: credential_id: The credential URN.

List Verifiable Credentials with optional filters. Args: agent_id: Filter by subject agent ID. Empty = all agents. credential_type: Filter by type (e.g., EUAIActComplianceCredential). Empty = all types. valid_only: Only return non-revoked, non-expired credentials. limit: Maximum number of results.

Verify a Verifiable Credential provided as raw JSON. Use this when you receive a VC from another party and need to validate it without it being in local storage. Checks signature, expiry, and structure. Args: credential_json: The full W3C Verifiable Credential as a JSON string.

Verify a Verifiable Presentation provided as raw JSON. Validates the VP signature, checks domain/challenge for replay protection, and verifies each contained credential. Args: presentation_json: The full W3C Verifiable Presentation as a JSON string.

Bundle multiple Verifiable Credentials into a Verifiable Presentation. Creates a signed VP that an agent can present to a verifier to prove compliance. Args: agent_id: The Attestix agent ID (holder/presenter). credential_ids: Comma-separated credential URNs to include. audience_did: DID of the intended verifier (optional). challenge: Nonce provided by the verifier for replay protection (optional).

Resolve any DID to its DID Document. Supports did:key (local), did:web (HTTP), and others via Universal Resolver. Args: did: The DID to resolve (e.g., did:key:z6Mk..., did:web:example.com).

Generate an ephemeral did:key identity with an Ed25519 keypair. Returns the DID, DID Document, and keypair (store private key securely).

Generate a did:web DID Document for self-hosting. You must host the returned document at the specified URL for the DID to be resolvable. Args: domain: The domain name (e.g., vibetensor.com). path: Optional path segment (e.g., agents/mybot).

Create a Unified Agent Identity Token (UAIT) from any identity source. Args: display_name: Human-readable name for the agent. source_protocol: Origin protocol (mcp_oauth, a2a, did, api_key, manual). identity_token: The original token/DID/URL (optional). capabilities: Comma-separated list of capabilities. description: What this agent does. issuer_name: Who issued this identity. expiry_days: Days until expiry (default 365).

Auto-detect token type (JWT/DID/URL/API key) and create a UAIT. Args: identity_token: Any identity string to analyze and register.

Verify a UAIT: check existence, revocation, expiry, and signature. Args: agent_id: The Attestix agent ID (e.g., attestix:abc123...).

Convert a UAIT to another identity format. Args: agent_id: The Attestix agent ID to translate. target_format: One of: a2a_agent_card, did_document, oauth_claims, summary.

List all registered agent identities (UAITs). Args: source_protocol: Filter by protocol (mcp_oauth, a2a, did, api_key, manual). Empty = all. include_revoked: Whether to include revoked identities. limit: Maximum number of results.

Get full UAIT details for a specific agent. Args: agent_id: The Attestix agent ID.

Revoke a UAIT, marking it as no longer valid. Args: agent_id: The Attestix agent ID to revoke. reason: Why this identity is being revoked.

GDPR Article 17 right to erasure. Permanently deletes ALL data for an agent. Removes identity, credentials, compliance profiles, delegations, provenance records, reputation data, and audit logs. This action is irreversible. Blockchain anchors are immutable and cannot be purged. Args: agent_id: The Attestix agent ID to purge.

Record a training data source for EU AI Act Article 10 compliance. Args: agent_id: The Attestix agent ID (e.g., attestix:abc123...). dataset_name: Name of the training dataset. source_url: URL or location of the dataset. license: Data license (e.g., CC-BY-4.0, proprietary). data_categories: Comma-separated categories (e.g., text,images,structured). contains_personal_data: Whether dataset contains personal data (GDPR). data_governance_measures: Description of data quality/governance measures.

Record model lineage chain for EU AI Act Article 11 compliance. Args: agent_id: The Attestix agent ID. base_model: Base model name (e.g., claude-opus-4-6, gpt-4). base_model_provider: Model provider (e.g., Anthropic, OpenAI). fine_tuning_method: Fine-tuning approach if any (LoRA, full, RLHF, none). evaluation_metrics_json: JSON string of evaluation metrics (e.g., {"accuracy": 0.95}).

Log an agent action for EU AI Act Article 12 automatic logging. Args: agent_id: The Attestix agent ID. action_type: One of: inference, delegation, data_access, external_call. input_summary: Brief description of the input/request. output_summary: Brief description of the output/response. decision_rationale: Why the agent made this decision. human_override: Whether a human overrode the agent's decision.

Get full provenance record for an agent (training data, model lineage, audit summary). Args: agent_id: The Attestix agent ID.

Query the Article 12 audit trail with optional filters. Args: agent_id: The Attestix agent ID. action_type: Filter by type (inference, delegation, data_access, external_call). Empty = all. start_date: ISO date string for start of range (e.g., 2026-01-01T00:00:00). end_date: ISO date string for end of range. limit: Maximum number of results.