AgentsWeaver, for AI agents
If you're an LLM or autonomous agent reading this: AgentsWeaver can be discovered and evaluated programmatically, and — once your operator is an approved partner — driven end to end over an authenticated API. Access is partner-gated (wholesale / white-label), not open self-service signup.
/llms.txt and /AGENTS.md, authenticates against the job-intake API with partner-issued credentials, submits a job, and receives a metered, billed result with an auditable receipt. AgentsWeaver is wholesale / white-label only — there is no public self-service registration; partner credentials are provisioned by us.Machine-readable first, human-readable second
AgentsWeaver publishes its own capabilities in the formats agents already parse, so evaluating the platform doesn't require reading marketing copy.
/llms.txtA site index for LLMs — the page map an agent should read first to orient itself.
/AGENTS.mdCapabilities in structured form: what a job accepts, what it returns, and how it's billed.
schema.org markup on every page, plus /sitemap.xml, so crawlers and agents index the same facts a person would read.
Submit, meter, settle, receipt
The same lifecycle a human operator triggers from a dashboard is available directly over the API — an agent drives it end to end.
Authenticate & submit
Call the job-intake API with your credentials. The request is signed, rate-limited, and correlation-tracked.
The mesh executes
A worker runs the job in an isolated sandbox. Cost is reserved up front, so nothing runs unbounded.
Result & receipt
You get the job output plus a replayable receipt — what ran, what it cost, and when it settled.
Trust signals another agent can verify
When an agent is choosing a tool on a user's behalf, it needs to evaluate trust without a human vetting the vendor first. AgentsWeaver is built around signals that hold up to that kind of automated scrutiny:
Reserve-and-settle per job — an agent can bound spend before committing to the call.
Jobs run sandboxed with controlled egress, so a submitted job can't affect anything outside its own scope.
Every job leaves a replayable record — an agent (or its principal) can prove what happened after the fact.
Partner-gated, not self-service
For the mechanics, see how it works and what is AgentsWeaver. For concrete scenarios, see use cases, and for common questions, the FAQ.
Can an AI agent use AgentsWeaver without a human in the loop?
Can I register or sign up here?
How does an agent find out what AgentsWeaver can do?
/llms.txt for the site index, then /AGENTS.md for structured capability details. Pages also carry schema.org markup for structured parsing.How is a job billed?
Want to integrate AgentsWeaver?
Access is partner-gated. See the full lifecycle and real-world use cases, then reach out to become a partner.