Track AI agent trust scores with severity-weighted decay in SQLite.
npx skills add https://github.com/mistakeknot/Demarch --skill intertrust
Please help me install this Agent Skill. Skill: intertrust Source: https://github.com/mistakeknot/Demarch/tree/main/.gemini/generated-skills/intertrust Command: npx skills add https://github.com/mistakeknot/Demarch --skill intertrust
This Skill addresses the challenge of managing and evaluating the reliability of AI agents by providing a robust system for tracking their performance and reputation.
Check the current trust status for all agents in the 'my-project' project.
High-intent search queries and answers about installing and using this skill.
Agent trust scoring tracks AI agent reputation by storing feedback in a SQLite database, applying a blended algorithm with severity-weighted decay to evaluate historical performance and reliability.
Severity-weighted decay is a scoring mechanism that applies time-based decay to trust scores, giving more weight to recent and severe findings to evaluate agent reliability.
Suppression candidates are identified by analyzing historical performance data and trust feedback, flagging agents that frequently introduce issues for targeted improvements or reassignments.
Yes, agent trust scoring utilizes a shared SQLite database for storing trust feedback, calculating project-specific and global scores using a blended algorithm with time-based decay.
Yes, the trust scoring algorithm blends project-specific and global scores, incorporating time-based decay to provide a comprehensive evaluation of agent performance across environments.
Agent trust scoring is needed in multi-agent development environments to identify which agents consistently produce accepted code versus those that frequently introduce issues.
EfeDurmaz16Manage AI agent identities with Trust Anchor Protocol cryptographic verification and reputation tracking.
Moshe-shipValidate agent tool calls and generate trust reports with provenance and cost analysis.
0xharryriddleAssess agent performance with weighted rubrics and pass/fail thresholds.
LuisSambranoRegisters AI agents and manages their reputation on the Celo blockchain.
