Take Interest Trust Layer for Agents
takeinterest.ai
Two surfaces for permitted agents. GuardClaw: deterministic 7-layer runtime security for agent tool calls (shell, filesystem, HTTP), no LLM in the security path, runs locally, available now via MCP. Take Interest context layer: working knowledge prepared as agent-ready, consent-gated context that approved agents can use with the owner's permission (preview). This is a discovery card; the live machine integration today is MCP via GuardClaw. See documentationUrl and /.well-known/mcp.json. An A2A JSON-RPC endp
What we verified
Everything in this section is a check agenttru.st performed itself.
- Assurance
- bronze Bronze — agent card fetched over HTTPS with a valid certificate
- Certificate
- Valid for this hostname
- DANE / TLSA
- Not verified (TLSA query returned RCodeNameError)
- Discovery
- Well-known document
- Protocols
- A2A verified by handshake or card fetch, not merely advertised
- Hosted in
- 🇺🇸 US · Fastly, Inc. (AS54113)
- First seen
- 1 Aug 2026
- Last checked
- 16m ago
What the agent claims
Copied from the agent's own card. Not verified — the operator of takeinterest.ai controls every value below.
- Provider
- TAKE INTEREST Inc.
- Protocol
- a2a
- Version
- 0.1.0
- Capabilities
- pushNotifications stateTransitionHistory streaming
- Auth schemes
- apiKey
- Agent card
- https://takeinterest.ai/.well-known/agent-card.json
Advertised skills
-
Runtime security for tool-using agents
Intercept and check an agent's shell, filesystem, and HTTP tool calls against policy, threat intelligence, and anomaly detection before they run. Deterministic enforcement, no LLM in the deny/allow pa…
-
Consent-gated context graph (preview)
Use a person's private context only with their consent, so an approved agent acts from real life and work instead of public defaults. Personal context stays per-owner and never reaches an outside prov…
Operate this agent and would rather not be listed? Request removal.