Add evaluating-llm-output skill#286
Open
vikast908 wants to merge 1 commit into
Open
Conversation
Adds the regression-testing discipline for LLM output that ordinary unit tests miss: a versioned golden set, cheapest-valid grading (assertions plus LLM-as-judge), behavior/tool-use assertions, negative and safety cases, CI gating on a pass threshold, and metric tracking over time, with a Verification checklist. Follows the standard skill anatomy and references test-driven-development and ci-cd-and-automation instead of duplicating them. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
4 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
A new
evaluating-llm-outputskill on building evals so LLM and agent changes don't silently regress.Why
The repo has
test-driven-development, but TDD targets deterministic code. LLM output is free-form text that ordinary unit tests can't assert on, so a prompt tweak or a model bump can quietly degrade quality with nothing in CI to catch it. There's no skill covering the regression discipline for model output.What's in it
Each step has code, and the skill closes with a Verification checklist.
Conventions followed
name+descriptiononly; the description leads with what it does, thenUse whentriggers.SKILL.md, no supporting files.test-driven-development(deterministic code) andci-cd-and-automation(pipeline setup) rather than duplicating them.Related: this pairs with my other PR adding
reliable-agent-loops. Happy to adjust naming, scope, or split anything.🤖 Generated with Claude Code