managed-agents-cma-verify-with-outcome-grader
Build a grade-and-revise loop with Outcomes: a writer drafts a cited research brief, a stateless grader fetches every URL and checks every quote against a rubric, and feedback drives revisions until the brief passes. Covers user.define_outcome, the span.outcome_evaluation_* events, and how to write a rubric the grad…
- Score
- Unranked
- Evidence
- No signals yet
- Last commit
- Not known
- Listed
Install
armory install managed-agents-cma-verify-with-outcome-grader --cli claudewrites the file to.claude/commands/managed-agents-cma-verify-with-outcome-grader.mdListed as compatible
# fetches the source and writes it to:
.claude/commands/managed-agents-cma-verify-with-outcome-grader.mdNeeds the armory CLI · not on npm yet, build it from cli/ in the repository
What it is
Official anthropics/anthropic-cookbook workflows component — Build a grade-and-revise loop with Outcomes: a writer drafts a cited research brief, a stateless grader fetches every URL and checks every quote against a rubric, and feedback drives revisions until the brief passes. Covers user.define_outcome, the span.outcome_evaluation_* events, and how to write a rubric the grader can act on.
When to use it
Build a grade-and-revise loop with Outcomes: a writer drafts a cited research brief, a stateless grader fetches every URL and checks every quote against a rubric, and feedback drives revisions until the brief passes. Covers user.define_outcome, the span.outcome_evaluation_* events, and how to write a rubric the grader can act on.
How to install / invoke
Official Anthropic cookbook recipe (notebook). Open it to follow the runnable walkthrough: https://github.com/anthropics/anthropic-cookbook/blob/main/managed_agents/CMA_verify_with_outcome_grader.ipynb
Notes
Recipe "Outcomes: agents that verify their own work" from the anthropic-cookbook managed_agents/ family (categories: Claude Managed Agents, Evals). License: MIT.