Repository navigation
Verify every published number in CI, redo the README, fix the skill references - #30
Merged
Merged
Conversation
…eferences evals/check_numbers.py recomputes each number in README.md and the rebuild results page from the raw files, checks that the rule block is identical in the two mirrors and present in SKILL.md, and checks that the five version strings agree. It exits 1 on the first mismatch. A GitHub Actions workflow runs it with the three test suites on every push and pull request. README.md is rewritten: no bold lead-ins (the skill's own rule 11), the 663-character history sentence is three sentences, the reply table says what its columns count, the documents table carries the noise caveat from its results page, and the judge line names 2.0.1 instead of "the skill". Every number passes the checker. skills/simple-english: references/checklist.md is deleted. Nothing loaded it, and its content restates the self-check, the rule catalog, and the linter. references/use-cases.md loses its em-dashes and bold lead-ins and gains a load trigger in the Document section. The never-touch rule moves from last to second. The self-check searches for bold and short-lived headings. CHECK mode opens the catalog first. The two mirrors move the never-touch sentence to match. evals/ste_lint.py --type reply --gate no longer crashes. The 32 raw files under package/ that leaked from the closed PR #17 are removed. Claude-Session: https://claude.ai/code/session_018LyDhBTHmY3mTwD5dBoEsm
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
evals/check_numbers.py recomputes each number in README.md and the
rebuild results page from the raw files, checks that the rule block is
identical in the two mirrors and present in SKILL.md, and checks that
the five version strings agree. It exits 1 on the first mismatch. A
GitHub Actions workflow runs it with the three test suites on every
push and pull request.
README.md is rewritten: no bold lead-ins (the skill's own rule 11), the
663-character history sentence is three sentences, the reply table says
what its columns count, the documents table carries the noise caveat
from its results page, and the judge line names 2.0.1 instead of "the
skill". Every number passes the checker.
skills/simple-english: references/checklist.md is deleted. Nothing
loaded it, and its content restates the self-check, the rule catalog,
and the linter. references/use-cases.md loses its em-dashes and bold
lead-ins and gains a load trigger in the Document section. The
never-touch rule moves from last to second. The self-check searches
for bold and short-lived headings. CHECK mode opens the catalog first.
The two mirrors move the never-touch sentence to match.
evals/ste_lint.py --type reply --gate no longer crashes. The 32 raw
files under package/ that leaked from the closed PR #17 are removed.
Checks run locally: ste_lint self-test, 12 hook tests, 13 lint-hook tests, skills-ref validate, claude plugin validate, check_numbers (and it fails on a one-digit change).
https://claude.ai/code/session_018LyDhBTHmY3mTwD5dBoEsm