Trust, safety & payments for the agent economy.

Independent infrastructure for AI agents that pay, fetch, and transact on their own.

Agents are starting to pay for API access autonomously — the x402 protocol turns any HTTP 402 into a machine-payable endpoint. That creates two questions an agent can't answer alone before it spends: is this endpoint reliable? and is this asset a scam? — and a third for the operator running the agent: how do I keep it from overspending? Forum Labs builds the three pieces that answer them, with published, reproducible methodology.

Endpoint Trust Score

Objective, reproducible quality-of-service & reliability scoring for x402 endpoints — so an agent can check an endpoint before it pays it.

GET /v1/trust/score · $0.005

Methodology →

Token Safety Screen

Pre-trade screen that flags serial rug-pullers by their on-chain deployer history — before an agent buys a token.

POST /v1/safety/screen · $0.01

How it works →

Payfetch

The agent's paying-fetch client: pays x402 endpoints and returns the content, under spending caps you set. BYO wallet, local-first.

npm · MCP · CLI · free

Overview →

We measure and we tool — we don't take custody of funds, don't accept payment from the endpoints or tokens we measure, and every score ships with the raw counts behind it. Metrics are mechanical and informational, not financial advice.


For endpoint operators

Our reliability monitor identifies itself honestly (User-Agent portico-trust-prober · +https://forum-labs.com) and is a good citizen: conservative cadence, exponential backoff, honors 429/Retry-After, and probes only publicly-listed endpoints in the manner they advertise.

Re-probe, correction, or opt-out: email ops@forum-labs.com — opt-outs honored within 24 hours; corrections are appended (our record is append-only, never silently edited).

Contact: ops@forum-labs.com · @shopforumlabs
Metrics and tools are informational and mechanical. Not financial advice.