Skip to content

Quality campaign tooling: TS standards, coverage, checkers, RSI gate - #131

Merged
w4ffl35 merged 3 commits into
masterfrom
feat/codeintel-batch-tools
Oct 13, 2026
Merged

w4ffl35 merged 3 commits into
masterfrom
feat/codeintel-batch-tools

Conversation

@w4ffl35

@w4ffl35 w4ffl35 commented Oct 13, 2026

Copy link
Copy Markdown
Member

Merges the quality-campaign tooling onto master:

  • docs/typescript-standards.md (defined: HQ/Commons had no TS size/complexity rules)
  • scripts/coverage.mjs + npm run coverage (c8; current lines 24.59%)
  • scripts/check-code-standards.mjs + baseline ratchet (23,599 entries)
  • scripts/check-architecture.mjs + baseline ratchet (cycles/clones/complexity/nesting/params/console/exit)
  • scripts/rsi-gate.sh (coverage floor 24 + both ratchets) wired into CI
  • Grouped quality issues Quality: RSI controller decomposition (tests-first) #126-129

Includes the determinism fix for cycle enumeration (sorted DFS; diamond
fixture test) that unblocks PR #130's architecture gate.

Evidence: tsc clean, 248/248 unit, 9/9 e2e, both self-tests green, arch gate
verified clean on two independent checkouts of this tip.

RSI smoke added 3 commits October 12, 2026 18:01
…eline

Cycle enumeration used codeintel graph order (raw readdir order, differs
per checkout); DFS done-pruning then reported different cycle sets for
identical graphs, failing the ratchet on fresh clones (PR #130 CI).
Sort nodes and adjacency lists; pin canonical output with a diamond
fixture test.
Cycle baseline keys embedded absolute checkout paths, so every cycle
reported as NEW on any other checkout (CI, worktrees). Keys are now
workspace-relative like all other finding classes; the self-test copies
a fixture to a second dir and asserts no new: cycle lines.
@w4ffl35
w4ffl35 merged commit 73701ca into master Oct 13, 2026
2 checks passed
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