Get Started

AgenticIdentity Docs

Cryptographic identity, trust, and accountability for AI agents.

What is AgenticIdentity?

AgenticIdentity gives AI agents a persistent, verifiable identity. Every action is signed, every permission is tracked, and the agent can prove its entire history.

Core capabilities:

Cryptographic identity anchors
Signed action receipts
Trust web (permissions)
Competence proof (demonstrated ability)
Negative capability proof (what agent CAN'T do)
Temporal continuity (unbroken experience chain)
Identity inheritance (spawn lineage)

Quick Install

MCP (Claude Desktop, Cursor, Windsurf):

curl -fsSL https://raw.githubusercontent.com/agentralabs/agentic-identity/main/scripts/install.sh | bash

Python:

pip install agentic-identity

Rust:

cargo install agentic-identity-cli

JavaScript:

npm install @agenticamem/identity

Core Capabilities

1. Identity Anchor

A cryptographic root identity for the agent.

aid init --name "my-agent"

Creates a keypair. The public key IS the agent's identity. Portable, verifiable, permanent.

2. Action Receipts

Every significant action gets signed.

aid sign --action "deployed v1.2.0 to production"

Creates tamper-evident proof: this agent did this action at this time.

3. Trust Web

Permissions granted between identities.

aid trust grant --to agent-123 --capability "deploy:staging" --expires 30d

Delegatable, revocable, time-bounded authority.

4. Competence Proof

Demonstrated ability, not just authorization.

aid competence record --domain "rust-development" --outcome success
aid competence prove --domain "rust-development"

Tracks success rates, streaks, and generates verifiable competence proofs.

5. Negative Capability Proof

Prove what the agent CANNOT do (structural impossibility).

aid cannot prove --capability "access:production-db"

Not policy — structural proof that the agent lacks the keys/authority.

6. Temporal Continuity

Unbroken chain of experience since genesis.

aid experience record --content "completed first deployment"
aid continuity prove

Proves continuous existence — no gaps, no imposters.

7. Identity Inheritance

Track spawn lineage (parent → child agents).

aid spawn create --name "worker-agent" --authority "subset"

Children inherit bounded authority from parents. Full lineage tracked.

MCP Tools (30 total)

Category	Tools
Anchor	`identity_create`, `identity_show`, `identity_verify`, `identity_export`, `identity_import`
Receipts	`action_sign`, `action_verify`, `action_list`, `action_chain`, `action_receipt`
Trust	`trust_grant`, `trust_revoke`, `trust_verify`, `trust_list`, `trust_delegation`
Competence	`competence_record`, `competence_show`, `competence_prove`, `competence_verify`, `competence_list`
Negative	`negative_prove`, `negative_verify`, `negative_declare`, `negative_list`, `negative_check`
Continuity	`experience_record`, `experience_list`, `continuity_prove`, `continuity_verify`, `continuity_chain`
Spawn	`spawn_create`, `spawn_list`, `spawn_verify`, `spawn_lineage`, `spawn_authority`

Use Cases

Multi-agent systems: Agents verify each other's identity before collaboration.

Audit trails: Every action signed — complete accountability.

Permission management: Grant, revoke, delegate authority with cryptographic proof.

Agent reputation: Competence proofs build verifiable track record.

Security boundaries: Negative proofs show what agent cannot access.

File Format

Identity data stored in .aid files — portable, encrypted, user-owned.

~/.agentic/identity/
├── anchor.aid       # Root identity
├── receipts.aid     # Action history
├── trust.aid        # Permission grants
└── continuity.aid   # Experience chain