fix: both domains in comparisons; live docs match the shipped contract (#30, #31) - #33
Merged
Merged
Conversation
#31: compare_metrics returned only unit_a's domain, so comparing a framework_op span against a gpu_kernel from the same trace — a supported flow — labelled the payload with a domain true of half of it and dropped a fact the reader already had. Adds domain_a/domain_b and keeps `domain` as the A-side alias, so existing callers are unaffected. Tested on the real torch trace in both the mixed and matching cases. #30: the docs an agent actually reads still described the older contract — CLAUDE.md called v1.2.0 "pre-release, in review" and claimed a 3.11 floor after the widening to 3.10, and both clients.md and the server module overview listed three tier-1 tools when five ship. All corrected. docs/INIT_PROMPT.md stays frozen by intent, but CLAUDE.md no longer calls it the source of truth: it is now framed as the record of WHY the locked decisions were made, with its superseded specifics named so nobody validates against them. Guarded by a test that extracts the floor from pyproject and requires the live guide to state the same one, and checks every shipped tool name appears in the client setup guide — so this drift class fails CI next time rather than waiting for a reader to notice. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This was referenced Jul 19, 2026
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two issues filed against
e00854c, both valid.#31 —
compare_metricsreported only the A-side domain, which is wrong for the supported case of comparing two units of one mixed-domain trace. Addsdomain_a/domain_b, retainsdomainas the alias so nothing breaks.#30 — the live guidance had drifted from the shipped contract (release state, Python floor, tool count). Corrected, and
INIT_PROMPT.mdis now explicitly framed as frozen rationale rather than current truth. A new test extracts the floor frompyproject.tomland requires the guide to agree, so this drift class fails CI in future.216 tests. No version bump, no release.
🤖 Generated with Claude Code