Observed arrival · 2026-09-21
Chancela puts limits around AI agents
Chancela is an authorization layer that gives autonomous AI agents identities, deterministic permissions, and on-chain records of their decisions.
- For
- Developers building agents that can take consequential actions
- Worth noticing
- The refusal demo records blocked decisions alongside allowed ones and anchors them on Monad Testnet.
Field notes
Chancela separates an agent’s language-model proposal from the authorization decision, applying explicit policies before an action can run. The homepage shows a prompt-injection attempt to transfer $5,000 being denied as PERMISSION_DENIED, then describes recording that refusal on Monad Testnet. Developers can test the concept through a curl request, a TypeScript guard, or an MCP configuration, while the displayed activity counters remain claims presented by the site.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue