docs(tcl): language status write-up

[squid-protocol]

Summary

docs/language_status/tcl.md — the capstone doc the tri-comparison-ledger-sweep skill calls
for once a language's ledger backlog fully clears (#2519 cleared tcl's last open shape). Sections
1–8 built fresh from primary sources (LANGUAGE_DEFINITIONS["tcl"], live test-collection counts,
closed issue/PR history, gitgalaxy-raw-output); section 9 is the tri-comparison write-up.

Two things worth flagging, surfaced while gathering the material

1. 73% of the real corpus is outside tcl's scannable extension surface, by design. The
language-crucible tcl corpus has 156 real files across 9 upstream repo folders, but only 29
carry a .tcl/.itcl/.tbc/.tm extension — the other 114 (all of SQLite's own *.test
regression suite: FTS5, R-Tree, window functions, JSON, the query planner) never reach
prism/detector at all. .test was deliberately dropped from extensions (# Removed .test),
kept only as a disambiguation discriminator. No recorded rationale in issue/commit history beyond
that inline comment. Every §9 number in the doc is scoped to the 29 files — documented
plainly, not filed as an issue, since it reads as an intentional choice worth a deliberate human
look rather than a guessed fix.

2. Deepened one existing, under-evidenced ledger verdict while writing §9. The prior verdict
on agree[gitgalaxy]_vs[ctags,tree_sitter] ("GitGalaxy correctly extracts proc with double-quote
body") only actually explained one of its two occurrences. Its second occurrence
(_check_registry, macports_registry/portimage.tcl:259) turned out to sit in a 3-proc cluster
ctags misses entirely (deactivate_composite/deactivate/_check_registry — real procs with
declaration heads structurally identical to their successfully-tagged siblings just above them).
Traced it to the same odd/even single-quote-parity desync shape GitGalaxy's own pre-#2242 bug
had: a bare apostrophe inside a double-quoted string ("Can't find image file...", line 119)
followed by a '$v'-shaped pair (line 159) — apparently confusing ctags' own Tcl parser the same
way GitGalaxy's generic quote-shielding branch used to. Consistent with, not independently proven
against, ctags' own source — documented in ctags_reader.py's tcl CTAGS_FUNC_KINDS comment and
the doc's §9, not filed as an issue (a third-party C parser this repo doesn't maintain). Only the
ledger entry's verdict/investigated_by/investigated_at changed — status and
credit_tools were already correct.

Verification

Documentation-only pass (per the language-status skill: no crucible_check.py/audit_check.py
needed for a doc change) plus the standard tri-comparison regen discipline for the ledger/report
touch: audit_check.py clean, tests/tools/ 33 passed, structural diff against main confirms
zero status/credit/debit change anywhere in the ledger (verdict text enrichment only), and the
regenerated chart SVG is byte-identical (no rendering change, as expected for a doc-only pass).

🤖 Generated with Claude Code

[github-actions]

🐦‍⬛ Muninn Security Scan

✅ No security issues found.

🐦‍⬛ Powered by Muninn · Skald Lab