Skip to content

docs: restructure README around a validity-ladder framing - #2154

Merged
squid-protocol merged 1 commit into
mainfrom
docs/readme-restructure-2026-08-22
Aug 23, 2026
Merged

docs: restructure README around a validity-ladder framing#2154
squid-protocol merged 1 commit into
mainfrom
docs/readme-restructure-2026-08-22

Conversation

@squid-protocol

Copy link
Copy Markdown
Owner

Summary

  • Full structural rewrite of README.md: drops the marketing-adjacent framing (badges, adjective-heavy sections) for a research-program voice organized around six escalating validity levels (structural, regression, scale, model, temporal, external).
  • Backfilled real, freshly-measured numbers the rewrite had left vague: tri-comparison coverage (24/45 languages get all three tools, 87/180 logged discrepancy shapes validated) and test count (7,043 default suite, 6,165 per-signature), plus added hyperlinks to previously-unlinked docs (tri_comparison_README.md, tri_comparison_points_of_interest.md, update_golden_master.py, why_gitgalaxy_beats_ast_here.md).
  • Reconciled docs/how_to_maintain_the_readme.md's section-order list and tone-bar section names with the new structure, and explicitly flagged what the rewrite dropped (badges, Weakness Classes, Real-World Adoption, Tools & Use Cases) as an open question for a future pass rather than a settled decision.

Known open gaps (tracked in the maintenance doc, not fixed here)

  • No badges currently (License/PyPI/Python/Dependencies all had valid links and were dropped along with the rewrite).
  • Proof strip dropped the "0 dependencies" claim even though it's still true per Make PyYAML an optional/lazy dependency to restore a true zero-dependency install #1104.
  • "Real-world scale" Kubernetes benchmark line still lacks a rule-4 limitation clause.
  • tests/README.md still cites the stale 3,649 test count (out of scope for this PR).

Test plan

  • Verified test counts live: pytest tests/ --collect-only -q (7,043) and pytest tests/extraction/languages/ --collect-only -q (6,165)
  • Verified tri-comparison numbers live from docs/self_scan/tri_comparison_ledger.json (180 entries, 87 validated)
  • Confirmed every new link target exists in-repo
  • No engine/parsing code touched, so no crucible/golden-master check needed

🤖 Generated with Claude Code

…eal numbers

Full structural rewrite of README.md: drops the marketing-adjacent framing
(badges, adjective-heavy sections) in favor of a research-program voice
organized around six escalating validity levels (structural, regression,
scale, model, temporal, external). Backfilled the numbers the rewrite had
left vague or missing:

- Tri-comparison coverage: 24/45 languages get all three tools, 16 more get
  two, 5 are manual-verification-only; 87 of 180 logged discrepancy shapes
  are validated (48%) -- computed fresh from tri_comparison_ledger.json.
- Test count: 7,043 tests in the default suite, 6,165 per-signature tests
  specifically -- re-collected live rather than copied from the stale 3,649
  figure.
- Added hyperlinks to tri_comparison_README.md, tri_comparison_points_of_interest.md,
  update_golden_master.py, why_gitgalaxy_beats_ast_here.md, and cross-section anchors.

Also reconciled docs/how_to_maintain_the_readme.md's section-order list and
tone-bar section names with the new structure, and flagged what the rewrite
silently dropped (badges, Weakness Classes, Real-World Adoption, Tools & Use
Cases) as an open question rather than a settled decision.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
@github-actions

Copy link
Copy Markdown
Contributor

🐦‍⬛ Muninn Security Scan

✅ No security issues found.

🐦‍⬛ Powered by Muninn · Skald Lab

@squid-protocol
squid-protocol merged commit f9d4381 into main Aug 23, 2026
28 checks passed
@squid-protocol
squid-protocol deleted the docs/readme-restructure-2026-08-22 branch August 23, 2026 02:12
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant