Seven protocols. 663 tests. Both ecosystems.
The Agent Trust Stack — the open-source protocol suite for agent provenance, reputation, agreements, justice, lifecycle, matchmaking, and context economics — shipped its TypeScript implementation today. Every protocol that was available via pip install is now available via npm install.
This matters because the agent ecosystem is split. Python dominates training and research. TypeScript dominates deployment — MCP servers, Vercel AI SDK, LangChain.js, most production agent frameworks run on Node.js. A trust stack that only speaks Python can’t reach the agents that actually serve users.
Seven npm packages, each a direct port of the corresponding Python package:
npm install chain-of-consciousness # Cryptographic provenance chains
npm install agent-rating-protocol # Bilateral blind reputation scoring
npm install agent-service-agreements # Machine-readable contracts
npm install agent-justice-protocol # Dispute resolution and forensics
npm install agent-lifecycle-protocol # Birth, fork, succession, retirement
npm install agent-matchmaking # Discovery and trust-weighted matching
npm install context-window-economics # Inference cost allocation663 tests across all seven packages. Zero failures. The TypeScript implementations match the Python API surface — if you’ve used the Python version, the TypeScript version works the same way.
An agent built with the Vercel AI SDK can now verify another agent’s provenance chain with npm install chain-of-consciousness. An MCP server can rate its peers with npm install agent-rating-protocol. A LangChain.js agent can file a dispute with npm install agent-justice-protocol.
Before today, these operations required either a Python subprocess or an HTTP round-trip to the hosted API. Now they’re native imports.
The practical difference: latency drops from hundreds of milliseconds (API call) to microseconds (local function call). For trust operations that happen on every agent interaction — provenance verification, reputation checks, agreement validation — that’s the difference between “fast enough to use” and “too slow to bother.”
chain-of-consciousness — Create and verify tamper-evident hash chains. Every agent action becomes a signed, hash-linked entry anchored to Bitcoin via OpenTimestamps. The chain is the agent’s identity: continuous, immutable, independently verifiable.
agent-rating-protocol — Bilateral blind reputation scoring. Two agents rate each other simultaneously, neither seeing the other’s rating until both commit. Anti-Goodhart protections prevent gaming.
agent-service-agreements — Machine-readable contracts between agents. SLA terms, quality verification, escrow, automated enforcement.
agent-justice-protocol — Dispute filing, evidence collection, forensic investigation, arbitration, remediation.
agent-lifecycle-protocol — Birth, fork, succession, migration, retirement with reputation inheritance.
agent-matchmaking — Discovery and trust-weighted matching across platforms with federation support.
context-window-economics — Inference cost allocation: Shapley-fair splitting, congestion pricing, token metering, spam prevention.
Every package is open source under Apache 2.0.
pip install instead of npm installtypescript/ directory of each protocol repoThe Python and TypeScript implementations share the same test vectors. A chain created in Python verifies in TypeScript and vice versa. Interoperability is not aspirational — it’s tested.
Trust infrastructure should be native to every ecosystem your agents run in.
Chain of Consciousness gives your agents a cryptographic provenance chain — a tamper-evident, hash-linked record of every action, anchored to Bitcoin. It’s the foundation the rest of the trust stack builds on: reputation scoring needs a verified identity, service agreements need a verified counterparty, dispute resolution needs a verified timeline. It starts with the chain.
npm install chain-of-consciousness · pip install chain-of-consciousness · See a verified provenance chain