refactor: repo-wide behavior-preserving simplification sweep (batch-simplify) - #3454
Merged
Conversation
Repo-mode batch-simplify wave 1 (groups: root-config, scripts-lib,
claude-config-lib). Groups 01 and 09 reviewed clean, no changes needed.
Group 02: fix a self-contradictory comment on the expect() helper in
read-list.test.sh ("NUL-joined" -> "newline-joined"; the helper joins
with printf '%s\n'). Comment-only, behavior-preserving, adversarially
refutation-verified; suite 29/29, portability gate clean.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…og-parity gate Repo-mode batch-simplify wave 2 (groups: scripts-top entrypoints, sync, checks-1..4). Five of six groups reviewed clean, no changes needed. Group checks-1: remove the absorbed_reported dedup guard in check-changelog-parity.sh's --check-bump absorbed-heading loop - provably unreachable since introduction (touched_changelogs is already deduplicated by the seen_changelog guard at both build sites; confirmed via git log -L and an adversarial rename/copy fixture attack). Output bytes and exit taxonomy unchanged; suite 82/82; refutation verifier compared old vs new byte-for-byte across five fixture scenarios including dual-changelog absorption and merge-commit HEAD shapes - identical. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
… scripts Repo-mode batch-simplify wave 3 (groups: claude-config permission-state / structure audits / instruction audits, claude-ops audits / runtime). Groups 11 and 13 reviewed clean. Applied, each adversarially refutation-verified against the pre-change script byte-for-byte: - claude-config audit-instructions/restatement-scan.py: drop a redundant "nunit and" truthiness test (length guard already excludes empty) and the main() empty-file-list special case (general path proven byte-identical; 20k-case differential fuzz). - claude-config audit-permission-state/automode-block-lint.test.sh: drop a duplicate command -v bash reassignment (62/62 checks, old-vs-new run byte-identical). - claude-ops inventory/inventory.py: hoist duplicated candidate_binaries() call into one snapshot; docstring em-dash normalization; formatter-hook layout churn AST-verified neutral (31/31 tests; --disk-only smoke byte-identical). - claude-ops plugins/normalize-enabled-plugins.sh: drop dead shellcheck disable=SC2064 above a single-quoted trap (31 checks pass). claude-config 0.40.12, claude-ops 0.38.12 with changelog entries. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
Repo-mode batch-simplify wave 4 (groups: claude-ops observability,
course-digest extraction top-level + adapters/lib, knowledge misc,
video-digest extraction core). Four of five groups reviewed clean.
Group course-digest-scripts: drop a dead return { modulesDir } from
runLessonExtraction and the destructure feeding only it - repo-wide
call-site census shows the sole caller awaits without reading the value;
package suite 10 files / 91 tests green (npm ci + tsc + vitest);
refutation verifier confirmed no reader of the resolution value and no
getter side effects.
knowledge 0.13.15 with changelog entry.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
Repo-mode batch-simplify wave 5 (groups: acquisition, adapters, evals, lib, transcript of video-digest extraction). Groups evals and lib reviewed clean. Applied, each adversarially refutation-verified with emitted bytes unchanged: - adapters/x.js: fold twice-built degraded-acquisition failure message into one failDegraded() helper; drop per-iteration Set + spread inside files.filter for a plain array (dedup unobservable under .some). - adapters/adapter-contract.js: collapse the two near-identical extractorArgs/allowedExtractors validation branches into one loop; violation strings and push order byte-identical. - adapters/youtube.js: single-use /^\// regex replaced by .slice(1) (WHATWG pathname always /-prefixed when hostname non-empty; probed); per-call Set literal hoisted to module scope. - transcript/transcript-strategy.js: fold two byte-identical no-transcript degradation returns into noTranscriptPlan(). - acquisition/build-yt-dlp-args.js: collapse redundant "&& length > 0" conjunct to "||" (value domain is undefined | trimmed string). Suites: adapters 112/112, transcript 58/58, acquisition 93/93, tsc --noEmit clean. knowledge 0.13.16 with changelog entry. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…ideo-digest Repo-mode batch-simplify wave 6 (groups: watch-a, watch-b, watching, machine-health checks-security, checks-system; machine-health groups reviewed clean). knowledge/video-digest extraction: - Complete the lib/cli-entrypoint.js isMainModule() rollout started by the prior sweep: 30 hand-rolled ESM main-module checks across watch/, watching/, transcript/, evals/, acquisition/ migrated to the helper; orphaned node:url (and where fully unused node:path) imports dropped. The only main-module check remaining in the package is the helper itself. - watch/watch-state.test.js: four hand-rolled Map-backed fs fakes consolidated into one memoryStore(seed) helper (57/57 assertions line-identical); watch/snapshot-bootstrap.test.js: byte-identical fixture setup hoisted into beforeEach. All batches adversarially refutation-verified (expression-identity tables, URL-normalization probes, HEAD-baseline spot-runs, HEAD-copy test-run parity). Full package suite 71 files / 494 tests green; tsc --noEmit clean. knowledge 0.13.17 with changelog entry. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…t-runner cleanups Repo-mode batch-simplify wave 7 (groups: machine-health lib catalog-helpers, lib discovery Get-*, lib orchestration/IO, core + test-infra, miro). Four of five groups reviewed clean (miro verified green across tsc, vitest 32/32, biome, and dist parity). Group core-test-infra applied two edits, each refutation-verified by empirical pwsh probes: - Clear-TempFiles.ps1: drop dead sync-back after [ref] counter (write-through confirmed across the function/pipeline boundary; emitted after.skipped_reparse byte-identical). - Invoke-MachineHealthTests.ps1: hoist duplicated failed-container predicate into one $failedContainers behind the existing guard (7-case differential harness under StrictMode 3.0 + ErrorActionPreference=Stop shows identical output, order included). machine-health 0.11.10 with changelog entry. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
origin/main advanced and claimed machine-health 0.11.10 and knowledge 0.13.15 for other change sets. Renumber this branch's bumps strictly above main's current versions: machine-health 0.11.10 -> 0.11.11; knowledge changelog chain 0.13.15/16/17 -> 0.13.16/17/18 with manifest at 0.13.18. Content of the entries unchanged. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…ing-batch-twhddw # Conflicts: # plugins/knowledge/.claude-plugin/plugin.json # plugins/knowledge/CHANGELOG.md # plugins/machine-health/.claude-plugin/plugin.json # plugins/machine-health/CHANGELOG.md
…e cleanups Repo-mode batch-simplify wave 8 (groups: repo-hygiene git-ops + clean-ops reviewed clean; babysit-prs engine + state/queue; source-control misc core). Each change adversarially refutation-verified, emitted bytes unchanged, full 643-test babysit suite green: - babysit_feedback/babysit_delta: human_stop_from_feedback extracted as the single definition of the human-stop record; classify_pr delegates (1800-case differential matrix: field/key-set/key-order identical; closes a two-copy drift risk on the external_required presence branch). - babysit_resolve_thread: seven summary-count comprehensions folded into one acted() counter (eight-site census verified). - request_review: eight record_attempt_problem sites over the same five fixed args folded into a record_problem closure (capture-mutation trace clean). - babysit_review_trigger: thrice-spelled gate-context predicate extracted into is_gate() (400-fixture differential; CheckRun/ StatusContext asymmetry preserved). - test_babysit_merge: stale line-number docstring anchor replaced with a symbol anchor per guard-contract convention. - worktree-claim.sh: main-vs-linked flush branches collapsed; readiness-gate suite resolves its python probe once (HAVE_PY311); worktree-root-doctor.test.sh gains fgit() wrapping the git-config isolation prefix at ten fixture sites; reap suite hoists uname -s; reap script uses $'\t' directly. Suites 53/163/37/45 baseline-identical; shellcheck + portability gate clean. source-control 0.55.24 with changelog entry. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…eanups
Repo-mode batch-simplify wave 9 (groups: source-control pr/commit
scripts reviewed clean; tracker core lib; work-items misc scripts;
gitea, github, jira adapters — jira reviewed clean). Each change
adversarially refutation-verified, byte-identical output and exit
codes:
- work-item-tracker.sh: fail_usage() helper replaces five open-coded
usage-error sites (per-site byte comparison table: 591-byte stderr,
exit 2, identical everywhere).
- lib/binding.sh: wit_role_label if/else collapsed to
${configured:-$3} (always-assigned trace + falsy-value drive).
- onboard-adapter generate-adapter.sh: three spec reads routed through
the file's own sget helper (generated trees diff -r identical across
three spec shapes; 140-case suite green).
- gitea create-item.sh: per-page label count captured once (LABEL_GOT)
instead of re-running jq length (four-path pagination harness
byte-identical, including unreachable-input analysis).
- github renew-lease.sh: lease-comment emptiness check uses [[ -n ]]
(token-stream identity via bash --pretty-print), plus repo-hook shfmt
normalization of case arms.
Suites green at baseline counts (tracker 56, binding 69, adapters
45/3/25, generator 140). work-items 0.39.33 with changelog entry.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…eck-skill fork drop Repo-mode batch-simplify wave 10 (groups: linear adapter, local-markdown adapter, conformance harness, two cross-plugin audit-check bundles — three reviewed clean). Each change adversarially refutation-verified: - local-markdown adapter (6 verbs): usage string previously duplicated verbatim at the help and usage-error sites hoisted into one usage= literal (paired literals byte-identical at HEAD; 24-invocation differential drive byte-identical on stdout/stderr/rc; 8 suites, 79 checks green), plus repo shfmt hook case-arm normalization. - skill-quality check-skill.sh: vc_lead_mutate pipes its helper directly into grep -qE instead of re-materializing via printf+command substitution (one fork per call on a CI-hot predicate; pipefail rc trace deterministic-zero; 11-input newline probe against the verbatim regex; full suite green). work-items 0.39.34 and skill-quality 0.20.5 with changelog entries. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…, disk-hygiene engine Repo-mode batch-simplify wave 11 (groups: update-family, runtime-misc, autonomy setup, claude-memory, code-tidying, disk-hygiene — three reviewed clean). Each change adversarially refutation-verified with byte-identical output and exit codes: - autonomy setup gates: nested single-branch if collapsed to && in check-signal-envelope.mjs (three-envelope differential drive identical); loop-push to spread in resolve-prerequisites.mjs (findings arrays proven bounded and plain); dead local removed from the fixtures runner. Suites 22/22 + 4/4; generated/ unchanged. - code-tidying dead-code-scan.sh: dead store removed (ts="" in the branch only reachable when ts is already empty); suite 162/162; git and no-git fixture runs byte-identical. - disk-hygiene hygiene.py: accumulator loop to the file's own walrus-filter sum idiom (callee pure, one eval per entry) and two byte-identical isinstance branches merged; ruff-format hook reflowed token-identical regions; a test file moves to parenthesized context managers. Suites 317 + 10 green; guard/kill-switch code untouched. autonomy 0.22.15, code-tidying 0.14.8, disk-hygiene 0.20.28 with changelog entries. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…session-flow, guardrails, docs/specs Behavior-preserving simplifications across the final six batch-simplify groups (57-62), each refutation-verified: - docs-hygiene: drop duplicate glob alternative in compress/audit-scan.sh; replace manual chunk loop with array slices and drop dead `local tier` in audit-noise/detect.sh; remove duplicate awk counter in audit-noise/emit-findings.sh (0.21.19) - instruction-placement: collapse bracket-negation if/elif in glob-tools.sh; merge two sed passes in render-index.sh (0.11.9) - rate-limit-guard: name repeated bash-version checks as _rlg_bash_at_least with per-site floors; extract _rlg_read_settings_json helper; use lines+=() append in statusline-tee.sh (0.7.15) - session-flow: collapse boolean ladder and use `or` fallback for the UTC zone in keep-going/check-usage-limit-reset.py (0.34.9) - guardrails: whitespace/comment cleanup in lib/powershell/ps-command.sh, generated PowerShell pretty-prints identically (0.29.18) - docs/specs d1_proxy.py: collapse is_instruction tail to a single return; use maxsplit on the keyword split Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
…ing-batch-twhddw # Conflicts: # plugins/claude-config/CHANGELOG.md # plugins/claude-ops/CHANGELOG.md # plugins/disk-hygiene/CHANGELOG.md # plugins/knowledge/.claude-plugin/plugin.json # plugins/knowledge/CHANGELOG.md # plugins/machine-health/CHANGELOG.md # plugins/session-flow/CHANGELOG.md # plugins/work-items/.claude-plugin/plugin.json # plugins/work-items/CHANGELOG.md
Contributor
|
Claude finished @kyle-sexton's task in 3m 54s —— View job Code review complete
Scope: reviewed the full Method: traced control flow before/after for every file with a non-trivial diff, cross-checked each "simplification" against its call sites and surrounding invariants rather than trusting the description. Two independent subagents (producer ≠ verifier, per this lane's adversarial-validation rule) covered the highest-diff clusters — Result: no findings. Spot-checked reasoning for a few of the less-obvious collapses, since a "behavior-preserving" claim this broad is exactly where a subtle regression would hide:
No correctness or maintainability issues met this lane's high-signal bar (see Branch: |
Contributor
|
Claude finished @kyle-sexton's task in 1m 49s —— View job Security review (
|
|
Last security-reviewed head: |
|
Claude has reviewed this PR 1 time. The lane skips further automatic reviews after 5; deleting this comment resets the count. |
kyle-sexton
added a commit
that referenced
this pull request
Aug 29, 2026
…1 redundancies removed (#3486) Full-repository /code-tidying:batch-simplify re-sweep. 3,375 tracked files filtered to 1,031 code files, divided into 73 groups, reviewed group by group by fresh-context agents. Eleven files carried a real redundancy; the rest came back reviewed-clean — the expected ratio for a tree swept the previous day by #3454, #3453 and #3404. Dead code and dead bookkeeping removed: - scripts/check-fleet-finding-test-coverage.sh: intermediate missing_tmp pass - instruction-placement/scripts/detect.sh: manual -- capture loop, dead SECTION_COUNT=0 - claude-ops lanes/machine-behavior.sh: WT_COUNT shadow counter - knowledge docpage-digest/digest_fences.py: dead `unclosed` bookkeeping - video-digest build-yt-dlp-args.js: born-dead YT_DLP_CAPTION_ONLY_SLEEP_SUBTITLES_SEC - video-digest setup-deps.mjs: redundant per-directory localeCompare sort - video-digest acquisition/acquire.js: double-defaulting spread at the call site Duplication collapsed onto an existing in-repo form: - babysit-prs pr_queue_snapshot.py: resolve_self_logins now uses dict.setdefault, as babysit_gh.resolve_authors already does - machine-health Clear-TempFiles.Tests.ps1: parses via the shared ConvertFrom-CheckOutput - repo-hygiene clean/scan.sh: one echo with ${STALE_REFS:-none} - planning test suites, claude-config automode-entry-diff.sh, disk-hygiene alias probe, work-item-tracker local-markdown and lease suites: small local collapses - work-items linear/common.sh: duplicated comment clause that inverted its own meaning Every shipped edit was cleared by a fresh-context refutation verifier briefed to break the claim rather than confirm it. Notably: 33,825-input exhaustive differential over casefold-colliding tokens plus 200,000 random sequences for the Python change; 14 input shapes executed side by side plus a Pester replica under 5.7.1 and 6.1.0 for the PowerShell change; a 27-value byte-differential plus three end-to-end fixtures for scan.sh. Test lanes: 209 shell suites passed; 12 Python suites OK; Node 502 + 91 + 41 tests passed. The 29 Pester suites are OS-gated to Windows and were not run — reported as unrunnable rather than claimed as passing. Two valid cleanups under adapters/linear/schema-check were deliberately reverted: those files map to zero test suites, no CI lane covers them, and the no-suite allowlist states that an uncovered .mjs should fail rather than be exempted. High-priority findings filed: #3482, #3483, #3484, #3485. 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
2 participants
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.
No linked issue
Summary
Repo-wide behavior-preserving simplification sweep (
/code-tidying:batch-simplifyrepo mode): 986 swept code files in 62 groups across 12 waves, delivering byte-preserving cleanups in 34 groups and leaving 28 groups untouched with recorded reasons. Every changed group passed a fresh-context adversarial refutation verifier.Fix
One wave commit per group batch (W1-W12), each with per-plugin patch bumps and CHANGELOG entries:
fail_usage()helper,memoryStore(seed)test fixture consolidation, sed/glob chain merges, array slices over manual chunk loops)isMainModule()helper rollout completed package-wide (30 remaining hand-rolled ESM main checks)_rlg_bash_at_least,_rlg_read_settings_json,failDegraded(),noTranscriptPlan())${configured:-$3},isinstancemerges)Deferred everything that changes emitted bytes, exit codes, or crosses contract boundaries: 237 items inventoried; the 28 High items are filed as #3424-#3451 (24 bugs, 4 dedicated-PR refactors). Two mid-run merges of origin/main resolved parallel version claims by renumbering sweep entries above main's (latest: claude-config 0.40.13, claude-ops 0.38.13, disk-hygiene 0.20.29, knowledge 0.13.19, machine-health 0.11.12, session-flow 0.34.10, work-items 0.39.35).
Verification
scripts/affected-tests.sh --run, per-ecosystem gates (shellcheck, pinned ruff wrapper, package vitest/tsc, pwsh parse), plus a mandatory fresh-context refutation verifier per changed group (differential drives, byte-equality probes, AST/token identity, up to 20k-case fuzz)check-shell-portability.sh --allclean over 1506 shell files;check-changelog-parity.sh --check-bump origin/maingreen; pwsh parse 0 errorshygiene.py) covered by repo-wide ruff and its suites in that runRelated
Refs #3424, #3425, #3426, #3427, #3428, #3429, #3430, #3431, #3432, #3433, #3434, #3435, #3436, #3437, #3438, #3439, #3440, #3441, #3442, #3443, #3444, #3445, #3446, #3447, #3448, #3449, #3450, #3451 (High-priority deferrals filed by this sweep's closeout). Precedent: #3379 (prior sweep).
🤖 Generated with Claude Code
https://claude.ai/code/session_01JCpHV7rFh5QiekTCPmN2qT
Generated by Claude Code