Run predefined evaluation cases against prompts to validate agent behavior and generate Markdown reports.
npx skills add https://github.com/nocoo/deca --skill eval-nocoo
Please help me install this Agent Skill. Skill: eval Source: https://github.com/nocoo/deca/tree/main/skills/eval Command: npx skills add https://github.com/nocoo/deca --skill eval-nocoo
Evaluate Deca prompts by running a suite of test cases, collecting results, judging outcomes manually, and generating audit-ready reports to measure alignment with guidelines.
Start the evaluator from the repository root and run the full eval workflow to produce a new report.
High-intent search queries and answers about installing and using this skill.
Deterministic prompt evaluation is performed by executing a predefined suite of test cases against prompts like IDENTITY.md and SOUL.md. The runner validates agent behavior, collects results, and aggregates outcomes into audit-ready reports.
Audit-ready QA reports are generated by running evaluation cases against prompts, applying manual judgments to pending results, and producing human-readable Markdown alongside machine-readable JSON outputs for collaboration and audit trails.
Agent behavior validation is achieved by running predefined evaluation cases against channel-specific prompts such as AGENTS.md. The evaluation workflow executes tests deterministically, aggregates results, and enables manual judging to verify alignment.
Deterministic testing measures prompt alignment by running predefined eval cases against prompts, collecting outcomes, and applying manual judgments. Results are aggregated into final evaluation reports verifying compliance with established guidelines.
The prompt evaluation workflow produces human-readable Markdown reports for manual review and machine-readable JSON outputs for automated audit trails. These formats support end-to-end collaboration across the testing, judging, and reporting lifecycle.
Manual judging is required in deterministic prompt evaluation to assess pending test results that automated runners cannot validate alone. It enables human oversight of agent behavior outcomes before aggregating final audit-ready reports.
Signet-AICreate SOUL.md, IDENTITY.md, USER.md, and AGENTS.md seed files defining agent voice and behavior.
euCannConvert OSCAL documents between JSON, YAML, and XML formats.
jmaglyValidate phase-gate criteria and generate pass/fail gate reports.
EntityProcessOrchestrate AgentV evaluations by simulating LLM responses without API keys.
