Search

Harness for measuring LLM agent resistance to indirect prompt injection and comparing defense effectiveness.

Multi coding-agent task dispatcher — a CLI proxy for claude that sends coding tasks to configurable AI backends

Python SDK for the blockfill agent execution

Kubernetes Dashboard for Agent Systems — CLI + web control plane for multi-agent LLM blueprints

Lifelong, contradiction-aware, curiosity-driven memory infrastructure for large language models. Local-first, LLM-agnostic, symbolic.

ERC-4337 smart-account substrate: deterministic addressing, factory deployment, ERC-1271 signing, UserOp building.

Append-only audit event schema + sink interface. Transport-agnostic; consumers wire concrete sinks (console / D1 / Cloud Logging / etc.).

User authentication (passkey + SIWE + Google OAuth), JWT sessions, and pluggable signer interfaces.

Agentic Connect — the SSO broker (spec 224 / ADR-0014). Mints CAIP-10-subject, no-owner AgentSessions (asymmetric + JWKS); drives entry-flow convergence (connect-auth verify → identity-directory resolve → issue/bootstrap/disambiguate) with assurance floor

Custody-layer SDK: CustodyPolicy ABI, CustodyAction enum + arg builders, EIP-712 typed-data helpers, and Custodian/Trustee/CustodyCouncil types. Speaks custody-domain vocabulary only — no delegation / steward / caveat concepts (those belong in @agenticpri

Agent relationships SDK: trust-fabric edge store (subject, object, relationshipType) with role taxonomy. Phase 1 ships edge-ID derivation + relationship-type constants + client skeleton; AgentRelationship.sol + RelationshipTypeRegistry.sol ports land in P

Agent Naming Service SDK: hierarchical naming for Smart Agents (.agent TLD). Owns name normalization, namehash/labelhash, records schemas, and the AgentNamingClient (resolve / reverse / registry). Speaks naming-domain vocabulary only — no delegation / cav

Solidity contracts + ABIs + flattened sources + per-network deployment addresses for the agenticprimitives stack.

Evidence-backed read model over canonical agents and their facets (ADR-0015 / spec 223). Declares the ports (NamingPort / OnChainReadPort / IndexerPort) + the query API with provenance + assurance. Not an authority; no eth_getLogs; no fallback. Adapters l

Pluggable envelope encryption + signers + HMAC providers (local-AES / AWS KMS / GCP KMS). Pure crypto primitives; no session lifecycle (that's in delegation).

Monorepo-wide formal vocabulary (RDFS/OWL T-box + SHACL/SKOS C-box + A-box fixtures) for agentic primitives. Off-chain source of truth the on-chain ontology (ADR-0009) instantiates. Ships TTL/JSON-LD artifacts + typed IRI constants.

Port implementations for @agenticprimitives/identity-directory: NamingPort (wraps agent-naming), OnChainReadPort (viem readContract), IndexerPort (in-memory; SPARQL/GraphDB later). The one composition layer allowed to import agent-naming (spec 100 §4 / AD

Cross-cutting branded types and chain primitives shared across @agenticprimitives/* packages.

Guarded RAG: grounded answers, refuse-when-unsupported, PII redaction, and an eval harness with metrics. Stdlib core, bring-your-own model.

Promptbook: Create persistent AI agents that turn your company's scattered knowledge into action