feat(skill): scope/deps/idempotency block + shipped eval suite (schliff field study)#34
Merged
Merged
Conversation
…ff field study) Measured with schliff 8.4.0 (eval suite included): 71.0/C -> 76.5/B. - edges 82 -> 100: eval suite now ships IN the repo with 5 edge cases; the two new ones pin REAL documented behaviors (state.json tamper guard = malformed_input, windowed review on huge diffs = scale_extreme) - composability 56 -> 81: explicit positive+negative scope, dependency declaration (bash/git, Codex CLI for deep mode, gtimeout), and the idempotency guarantee (timestamped reports, never mutates reviewed code) All changes are purely additive prose/eval data — no operational instruction or bash block was touched. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
First schliff field case study on a real production skill. Baseline with the released schliff 8.4.0: 71.0/C (7 dims, eval suite) — the tool's own priorities were edges (3/5 edge-case categories), composability (missing positive scope, dependency decl, idempotency statement), and the token budget.
Applied (purely additive — zero operational content touched):
malformed_input) and windowed review on huge diffs (scale_extreme) → edges 82→100Result: 71.0/C → 76.5/B.
Deliberately NOT done (informed decline): compressing the 14k-token SKILL.md — the inline bash is Hydra's operational core, and extraction to references/ would trade correctness risk for a score. Also: schliff's dead-marker detector false-positived on the word "placeholder" in Hydra's template-hygiene prose — filed as a schliff bug, no change here.
🤖 Generated with Claude Code