iCreatAI News
Skip to content
Trending storyUpdating

Applied Ingenuity releases agent rule enforcement framework

1 reports1 sources13 hr agoUpdated

Understand this story

AI overview

Applied Ingenuity released Project Chaperone, a rule-enforcing agent constraint framework: it allows agents to call tools while enforcing high-level rules, such as prohibiting personal data from leaving the network. The framework audits all workflow paths before running and has humans confirm exceptions; at runtime it blocks violating calls through data labels such as UNTRUSTED and PERSONAL and module attributes. The repository includes five shared rules and examples such as deliveries and safe_rag. In this framework the LLM is optional: modules can call an LLM, and the state can also let an LLM choose the next state of the workflow. The runtime requires Python 3.13 or newer; versions before 3.13.5 have not been verified; the default --setup demo uses fictional data and simulated tool calls and does not connect to a real mail server or LLM. When a Google project is in the Testing stage, sign-in expires after seven days and the signin command must be run again.

AI overview based on source reports · Updated 12 hr ago

Coverage timeline

Follow the coverage to explore different perspectives.

Nov 20
  1. Hacker News AI
    AI Agentic harness with explicit risk analysis

    Project Chaperone is a harness that permits agentic AI tool calling while enforcing high-level policies such as "no personal data may leave the network," adding a pre-run audit that checks every route of all workflows for rule violations and runtime enforcement through data labels (UNTRUSTED, PERSONAL) and module/tool properties (OUTPUT, LEAVES_NETWORK). Humans approve scoped exceptions to the rules, which are recorded so identical questions aren't asked again, and a possible violation not worth designing away can be accepted as a risk, with the call staying forbidden and refused if attempted. The repo ships five shared rules (no destructive actions, nothing leaves the network, no personal or untrusted data output, no decision made from untrusted data) plus examples including deliveries, share_notes, ask_my_mail and safe_rag.

Activity for this story

Current comparable activity
7
Peak comparable activity
10Nov 20 08:00
Change in the past 24 hours
–

Trends compare the same fully observed participants, which may cover fewer participants than current activity. Move the pointer or tap the chart to inspect hourly activity; use the left and right arrow keys with a keyboard.