Agent Evals CI Starter
A starter CI pipeline for agent evals. It runs on every pull request and blocks a bad merge before it ships, using a plain YAML case format.
Available as a page on this site. A repository copy is not published yet.
Published · Updated
What is inside
How to use it
Add a job to your own CI workflow that runs the eval cases, and point it at an eval file you write yourself in the case format on this page. Start with a forgiving pass-rate floor for the first release, then tighten it once the set has caught a real regression. The build fails on a drop, not a channel message someone can miss.
In my words
What this is
A CI workflow and an eval case format for one agentic workflow. The same discipline sits inside the 90-Day Production Sprint. A prompt or model change should not merge without passing a test set first.
The eval case format
| Field | Purpose | Example |
|---|---|---|
| id | A stable name, so a failure is traceable across runs | refund-double-charge |
| input | The exact prompt or tool call the agent receives | Customer says they were charged twice |
| expected | A pass rule: an exact match, a check, or a second model acting as a grader | Refund issued, ticket closed, no invented apology |
| tags | The failure mode a case guards against | prompt-injection, tool-misuse, refusal |
The CI gate
The workflow runs the eval set on every pull request, scores it, and blocks the merge under your floor. Nobody has to remember to run it by hand, which is the part that fails first once a team is busy.
The tags borrow their shape from OWASP’s Top 10 for Large Language Model Applications. Prompt injection and excessive agency (the agent doing something nobody asked it to) appear in a support agent as often as in a public chatbot. A tagged case catches the regression either way.
What it is not
It is not a considered eval set on its own. A handful of cases pulled from real conversations tests more than a long list invented at a desk. The starter takes no position on which cases you pick. Choosing the cases still maps to the MEASURE function (one of the framework’s four functions) in NIST’s AI Risk Management Framework (AI RMF).
Sources and neighbours
Sources
- OWASP Top 10 for Large Language Model Applications
OWASP · retrieved 2026-09-08
- AI Risk Management Framework
NIST · retrieved 2026-09-08
Related terms