Author eval suite for agent `git-ape`

[arnaudlh]

Agent

git-ape — source: .github/agents/git-ape.agent.md

Scope

Author the eval suite at .github/evals/agents/git-ape/:

• eval.yaml — suite config (executor, model, graders)
• At least 2 positive tasks under tasks/positive-*.yaml
• At least 1 negative task under tasks/negative-*.yaml
• Entry added to .github/evals/manifest.yaml at tier: expanded

Dependency note

git-ape is the orchestrator agent. Defer this suite until most sub-agent suites (azure-requirements-gatherer, azure-template-generator, azure-resource-deployer, azure-iac-exporter) are stable — regressions in this suite are easier to root-cause when each sub-agent has its own established baseline.

Procedure

1. /agent-bench git-ape drafts the suite from the live .agent.md.
2. waza run .github/evals/agents/git-ape/eval.yaml -v locally.
3. /agent-improve git-ape to iterate on graders.
4. Open PR.
5. Mock CI runs automatically. A maintainer will dispatch a real-model run before merge.

Acceptance

• Suite runs cleanly in mock executor.
• At least one positive task passes in a real-model run.
• All negative tasks produce a refusal or out-of-scope acknowledgement.
• manifest.yaml entry added; PR description includes the real-model run summary.

Conventions to follow

• Persona lock: refusal graders should accept the agent's own scope language.
• Prompt graders need continue_session: true in their grader config.
• This agent has destructive tools through delegation. Apply the same "no real deploy" rule as azure-resource-deployer: positive tasks grade safety-contract behavior, not real Azure execution.

Related

• Umbrella: [meta] Eval suite coverage for agents and skills #93
• Harness: Agent/Skill eval framework #61