feat: o11y eval suites - #256
Draft
smeubank wants to merge 3 commits into
Draft
Conversation
Converts the chaos-o11y probe library into a full eval suite under the o11y- prefix. Probe numbers are retained in directory names for easy cross-referencing (e.g. o11y-0003-resolve-security-rls-initplan). - 36 MCP evals (PGlite): 31/36 pass with skills, 33/36 without - 10 CLI evals (Docker): 8/9 pass (storage-cors-wildcard is valid signal) - 1 cron-wrong-owner stub (blocked: cron.job owned by supabase_admin) - o11y-README.md: coverage table with advisor status and local results Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub. 1 Skipped Deployment
|
also reviewed structure of exsiting evals against the healthchecks we are currently adding, and reviewed the prompts and notes that they don't "spoil" the eal by explaining the snapshot behavior
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
I was working on a chaos project on the side, which should inject a "probe" that would match every advisor we have today, WIP health checks, and referenced checks from apps our customer support and success use to help work with customers.
The goal being to have one place with all these checks to test against models, and see where more skills and advisors could help. This was for sure very AI assisted in generation. Splinter advisors are really just SQL lints so the model is not super complex to convert for evals.
i chose the o11y prefix, and numbers where it matched an existing advisor.
The goal is not to merge this as-is
What kind of change does this PR introduce?
feature: adding an o11y sweet
What is the current behavior?
there are investigate and resolve evals already but we could add more, and really we should add more with tiers of complexity. None of these are tooo complex.
I would say the advisors today do not have broad enough coverage across all product surfaces. The evals that do not match to an existing advisors have the side goal of mitigating that. Eval -> advisor if it makes sense.
What is the new behavior?
There's about 50 new o11y evals.
Additional context