Validate worker task output against TaskContract with tiered gating verdicts.
npx skills add https://github.com/choam2426/geas --skill evidence-gate
Please help me install this Agent Skill. Skill: evidence-gate Source: https://github.com/choam2426/geas/tree/main/plugin/skills/evidence-gate Command: npx skills add https://github.com/choam2426/geas --skill evidence-gate
AI-driven missions require an objective, reproducible way to confirm that a worker's output satisfies the defined TaskContract. Manual reviews are error‑prone and cannot enforce consistent gating criteria across many tasks.
Invoke the evidence‑gate skill on a specific task to automatically validate its output against the contract and receive a pass, fail, block, or error verdict.
High-intent search queries and answers about installing and using this skill.
Automated verification of AI task outputs validates worker results against a TaskContract using a three-tiered gate, running prechecks, mechanical eval commands, and rubric assessments to return a deterministic pass or fail verdict.
Tiered verification for AI agent missions is a three-step gating process that validates prerequisites, executes mechanical eval commands, and applies rubric checks to enforce objective acceptance criteria before marking a task complete.
Enforce gating rules in a pipeline by applying tiered verification to worker outputs, which analyzes evidence schemas and returns pass, fail, block, or error verdicts with integrated retry budgets and escalation policies.
Yes, configurable gate profiles support artifact-only tasks by enabling selective tier execution, allowing you to skip mechanical eval commands and run specific precheck or rubric assessments based on the task type.
Automated task contract evaluation returns deterministic verdicts of pass, fail, block, or error, integrating with decision records to enforce consistent acceptance criteria across many tasks.
Yes, tiered task verification operates as a standalone skill with no external dependencies, executing predefined eval commands and analyzing evidence schemas internally to produce objective gating decisions.
rongaredeEvaluate Claude Code sessions with capability and regression evals.
egorfedorovCoordinate multi-agent workflows with deterministic task sequencing and dependency resolution.
OceanEyeFFOrchestrate a closed-loop harness control system with evidence-based gating.
eai-toolsValidate implemented code and configurations using a 110-point rubric.
