Skip to content

docs: Phase 8.8 — prior-art evaluation mechanism (planning hook) - #9

Closed
artyhoo wants to merge 5 commits into
mainfrom
docs/phase-8.8-planning-hook
Closed

docs: Phase 8.8 — prior-art evaluation mechanism (planning hook)#9
artyhoo wants to merge 5 commits into
mainfrom
docs/phase-8.8-planning-hook

Conversation

@artyhoo

@artyhoo artyhoo commented May 8, 2026

Copy link
Copy Markdown
Owner

Summary

Planning hook for Phase 8.8 — documentation/process phase mirror of Phase 7.5 pattern. Inserts Phase 8.8 description into EXECUTION-PLAN.md §6 between Phase 8 (acceptance) and Phase 9+ (Path B / niches / AIF). Zero implementation; this PR is the planning anchor only — full design (~7-8 atomic commits) executes in dedicated Phase 8.8 session post-Phase-8-close.

2 atomic commits, docs-only.

What Phase 8.8 will deliver (6-level construction)

  • Уровень 0 (NEW in commit 2) — Promote prior-art consult to first-class framework principle. New entry in skills/rules-as-tests/references/overview.md (canonical SSOT per self-application.md §2 row L2) + matching factory/rules-manifest.json entry with check.type: "manual". Inherits SSOT pipeline: catalogued, RULES.md auto-render, Phase 2 meta-test drift detection. Equal footing с «every rule has executable check» / «AST > grep» / «no tautology». Consumers inherit via RULES.md emission.
  • Уровень 1 — docs/meta-factory/prior-art-evaluations.md SSOT (shipped reference, ≤500 lines, table-driven). Initial entries: Autogrep, Netlify framework-info, fitness functions vocabulary + what surfaces in Phase 8 capabilities.
  • Уровень 2 — §5.5 Step 1.5 mandatory consult gate (between «List capability areas» and «Resolve candidates»).
  • Уровень 3 — PHASE-ENTRY-PROMPT-TEMPLATE.md Task 0 «consult prior-art-evaluations.md» as mandatory first step for Phase N entry sessions.
  • Уровень 4 — CLAUDE.md / CONTRIBUTING.md build-vs-reuse invariant + Prior-art: commit trailer convention + pre-push WARN hook (new dep / new module ≥50 LOC without trailer).
  • Уровень 5 — periodic staleness refresh routine spec (impl Phase 8.X+ tooling); CI gate framework-prior-art-staleness (Phase 8.X+ as well).

Plus retroactive audit clause: Phase 8.8 retro documents post-hoc consult for each Phase 8 build-vs-reuse decision (Next 16 detection, regen diff metric, recipe expansion strategy R12/R14/R20, gate 5 invocation mode); gaps where Phase 8 reinvented existing solution are explicit findings.

Numbering rationale

  • Phase 8.X (already in plan) = parallel sub-phase for self-diagnostics implementation, post-Phase-8-acceptance.
  • Phase 8.8 (this PR) = sequential, blocks Phase 9 entry. Numbering avoids 8.5 collision per Art's direction.

Motivation

Phase 4-7 retros + Phase 7.5 analog research (2026-05-08) surfaced 3 closest prior-art candidates not in EXECUTION-PLAN.md §3.3:

  • Autogrep (Semgrep+LLM, Feb 2025) — closest LLM-driven rule generation pipeline; security-only domain.
  • Netlify framework-info — production-ready multi-framework detection.
  • Fitness functions (Evolutionary Architecture, Ford et al.) — existing vocabulary for our principles-as-tests pattern.

Without structural enforcement, every future phase risks the same skip pattern. Phase 8.8 closes the gap by making «consult prior-art before building» a process gate (§5.5 Step 1.5), a commit-trailer convention, AND a first-class framework principle (Уровень 0) — not just shipped-doc text references.

Acceptance

  • 2 atomic commits, conventional-commits, English subjects
  • Docs-only (only EXECUTION-PLAN.md modified)
  • EXECUTION-PLAN.md still parses; Phase 8.8 inserted at §6 between Phase 8 and Phase 9+ headings
  • Phase 8.X (self-diagnostics) preserved as parallel sub-phase, distinct from Phase 8.8 sequential

Test plan

  • grep ^### Phase 8\.8 returns 1 match
  • grep Уровень 0 returns 1 match (new in commit 2)
  • Phase 9+ heading still exists, untouched
  • Fresh-session reader test: a fresh Claude session reading EXECUTION-PLAN.md after Phase 8 close should identify Phase 8.8 as next mandatory step before Phase 9 entry

Related

artyhoo added 5 commits May 8, 2026 17:42
@artyhoo
docs(execution-plan): Phase 8.8 prior-art evaluation mechanism (manda…
290099f 
…tory between Phase 8 close and Phase 9 entry)

Phase 8.8 = documentation/process phase mirror of Phase 7.5 pattern.
Sequential, post-Phase-8-acceptance. Numbering 8.8 (not 8.5 — slot
reserved per Art's explicit direction; 8.X parallel = self-diagnostics).
Blocks Phase 9 entry — Phase 9 implementation risks reinventing
existing solutions без structural prior-art consult mechanism.

Scope brief (full design lives in Phase 8.8 session): 5-level
construction.

- L1 — prior-art-evaluations.md SSOT (table-driven, ≥3 initial entries
  Autogrep / Netlify framework-info / fitness functions vocabulary +
  what surfaces in Phase 8 capability decisions)
- L2 — §5.5 Step 1.5 mandatory consult gate
- L3 — phase entry prompt template Task 0
- L4 — CLAUDE/CONTRIBUTING build-vs-reuse invariant + Prior-art commit
  trailer convention + pre-push WARN hook
- L5 — periodic staleness refresh routine + CI gate (Phase 8.X+ tooling)

Retroactive audit clause: Phase 8.8 retro documents post-hoc consult for
each Phase 8 build-vs-reuse decision; gaps where Phase 8 reinvented
existing solution are explicit findings.

Motivation: Phase 4-7 retros + Phase 7.5 analog research (2026-05-08)
surfaced 3 closest prior-art candidates not in §3.3 — without enforced
consult mechanism each future phase risks the same skip pattern.
@artyhoo
docs(execution-plan): Phase 8.8 add Уровень 0 — promote prior-art con…
56f8afb 
…sult to first-class framework principle

Extends Phase 8.8 scope (5 → 6 levels) with Уровень 0: promote
prior-art consult from process discipline to framework-level invariant
via:

- New principle entry in skills/rules-as-tests/references/overview.md
  (canonical principles SSOT per self-application.md §2 row L2):
  «before building/introducing dependency, MUST run context7 query
  (≥3 phrasings) AND consult prior-art-evaluations.md; document via
  Prior-art: trailer or phase-N-research.md citation».
- Matching factory/rules-manifest.json entry with check.type=manual.
- Inherits SSOT pipeline benefits: catalogued, RULES.md auto-render,
  Phase 2 meta-test drift detection.
- Equal footing с «every rule has executable check» / «AST > grep» /
  «no tautology».
- Consumers inherit via RULES.md emission (when first consumer ships).

Atomic commit count for Phase 8.8 session bumped 6-7 → 7-8.
@artyhoo
docs(execution-plan): Phase 8.8 recursive self-validation — P-build-v…
ef05b73 
…s-reuse passes its own framework gates

Extend Phase 8.8 description with explicit recursive self-validation
section. P-build-vs-reuse principle is subject to the same gates as
every other rule in the framework (recursive thesis per
aif-comparison.md §10 differentiator #4).

Three forms of recursion documented:

1. L4 6-gate validation table — gate-by-gate status for the new
   principle. Gate 1 (schema) ✓ — manifest entry; Gate 2 (rule-tester)
   n/a — manual rule (G2 Pages-Router precedent); Gate 3 (mutation)
   n/a v1; Gate 4 (tautology) ✓ MUST — pre-push hook companion must
   fire when violated; Gate 5 (two-AI review) DEFER per §13.10 entry
   #4; Gate 6 (cross-rule conflict) ✓.

2. Self-application at creation — chicken-and-egg resolved: the
   2026-05-08 analog research session IS the consult evidence for
   creating P-build-vs-reuse. Strongest form recursion: rule about
   prior-art consult is itself prior-art-consulted. Documented in
   Phase 8.8 retro as «self-application at-creation evidence».

3. 6 Уровней as defense-in-depth — each level catches a distinct
   class of violation failure mode (drift / staleness / phase-entry /
   session-start / implementation-time / time-decay). Violation must
   pass all 6 layers undetected to slip through.

Mutation testing note: when gate 3 activates (Phase 9+ Path B), invert
pre-push trailer regex → meta-test must fail. Otherwise the test of
the test is itself tautological.

Acceptance criteria extended: P-build-vs-reuse passes applicable L4
gates + self-application-at-creation evidence documented in retro.
@artyhoo
docs(execution-plan): Phase 8.8 quality pass — close no-consumers cav…
b867fdb 
…eat + Task ordering + capability commit definition + Gate 4 independent meta-test

Reframing: Art = первый реальный consumer (2026-05-08), framework
используется для собственных проектов. §1 «no-consumers caveat» —
CLOSED. Recursive-self-validation thesis применяется в полную силу;
Phase 8.X / 8.8 — больше не v2 deferred, а v1 invariants.

Four substantive fixes to Phase 8.8 description:

1. Time estimate honest — 1 day → 2-3 дня; commit count 7-8 → 8-10.
   Документная база + Уровень 4c hook impl + Уровень 5 CI gate skeleton
   + recursive validation tests = day-and-half realistic, not «1 day
   doc phase».

2. Task ordering subsection — explicit T1-T11 sequence resolves
   chicken-and-egg structurally:
   T1 SSOT skeleton → T2 first entry (Autogrep, with Prior-art trailer
   ссылающийся на 2026-05-08 research) → T3 P-build-vs-reuse principle
   created через consult of existing T2 entry → T4-T5 more entries →
   T6-T11 levels + retro. Self-application at creation = T3 commit
   body, не post-hoc retroactivity.

3. Capability commit — formal grep-based definition. Detection logic:
   new dep в package.json OR new file ≥50 LOC under
   packages/core/<new-dir>/ OR new file ≥80 LOC anywhere в packages/.
   Thresholds — initial guesses per §5 numerical thresholds caveat;
   tune via Phase 8.8 retro on actual false-positive rate (>30% =
   review thresholds, не drop layer).

4. Gate 4 independent meta-test spec —
   tests/principles/p-build-vs-reuse-tautology.test.sh. 4 sub-tests:
   positive (trailer present → exit 0), negative (no trailer → warn),
   mutation 1 (invert regex → positive must fail), mutation 2 (drop
   detection → negative must fail). Tests instrument hook через
   subprocess в isolated tmpdir — independent от pre-push hook self-
   call (closes the «Gate 4 circular» concern). New CI job
   framework-self-build-vs-reuse depends on principles-meta-tests.

Net effect: Phase 8.8 description now production-grade for first-
consumer use, not minimal POC handwave.
@artyhoo
docs(execution-plan): Phase 8.8 split design — IR7 artifact rule (man…
a26a272 
…ifest) + process discipline (CONTRIBUTING) — resolves «выбивается from остальных» concern

User concern (2026-05-08): single P-build-vs-reuse rule mixes artifact
validation (research file content) с process discipline (commit
metadata, developer behavior). Manifest schema fits artifact rules;
check.type=manual для process part = dodge, not fit. Concern is real:
existing rules check static artifacts (R-rules on consumer code, IR
on manifest, meta-principles on framework files); P-build-vs-reuse
checked dynamic process — categorical mismatch.

Resolution — split into 2 entities by surface:

(a) IR7-research-cites-prior-art (ARTIFACT rule) → promoted to
manifest cleanly:
- applies-to: docs/meta-factory/phase-*-research.md
- check.type: command (validation script)
- Negative corpus: research files без citations OR с broken refs
- Passes Gates 1/2/4/6 ✓ MUST; 3/5 n/a/DEFER per §13.10
- Real artifact rule, не decoration. Subject to Phase 2 meta-tests
  on principle drift; rendered to RULES.md
- Lives в overview.md as canonical principle (equal footing с every
  rule has executable check / AST > grep / no tautology)

(b) Process discipline `Prior-art:` commit trailer → NOT in manifest.
Lives в CLAUDE.md + CONTRIBUTING.md. Enforced via:
- .husky/pre-push HARD FAIL с structured escape hatch
  (Prior-art: skipped — <rationale> permitted; absent = exit ≠ 0)
- PR review checklist (Prior-art trailers verified)
- Independent mutation-tested meta-test (anti-tautology guard for
  the hook itself)

Different surface (process, not artifact), different enforcement
model (developer-time hook + PR review, not L4 validator). Hard
fail with escape hatch, not soft warn — soft warn = self-defeating
если правило важно. Friction fires только на capability commits
(≈1 per 5-10 commits), не каждый.

Updated sections:
- Уровень 0: split into (a) IR7 promoted + (b) process discipline
  staying в CONTRIBUTING. Rationale: framework identity preservation.
- Уровень 4: HARD FAIL spec с escape hatch syntax (Prior-art:
  skipped — rationale).
- Recursive self-validation: split into (a) IR7 L4 6-gate validation
  table + (b) process discipline 4-axis table (PR review + local
  hook + independent meta-test + periodic staleness). Different
  axes, intentional design boundary.
- Task ordering T1-T12: T1-T2 SSOT first, T3 IR7 artifact rule, T4-T5
  more entries, T6-T7 phase entry mechanics, T8-T9 process discipline
  (CONTRIBUTING + hook), T10-T11 meta-tests + CI, T12 retro. No
  chicken-and-egg для either part.
- Gate 4 meta-tests: split into (a) IR7 corpus extension to existing
  gate-tautology.test.ts + (b) prior-art-trailer-hook.test.sh для
  process discipline. Two CI jobs framework-research-cites-prior-art
  + framework-prior-art-trailer-hook.
- Acceptance: explicit split criteria (IR7 passes manifest gates;
  process discipline has hard-fail hook + 5 mutation-tested cases +
  PR template checkbox).
@artyhoo

artyhoo commented May 8, 2026

Copy link
Copy Markdown
Owner Author

C-park per ultraview verdict (2026-05-08).

5 commits = iterative architectural pivots, не build history. Multi-agent review confirmed M1/M2/M3 (IR namespace mismatch, manifest scope violation, Phase 2 principles directory is right home) + 10 new findings (notably N1 «Mirror Phase 7.5» false claim, N3 ghost test path, N5 aif-evolve overlap missed, m4 broken bash heuristic).

Threading 6+ revision commits through wrong architectural foundation > clean restart with verified findings as input. Sunk cost = 5 doc commits на ветке (preserved locally for design exploration reference, not shipped infra).

Replacement: fresh Phase 8.8 session prompt (separate PR) incorporating all verified findings + new findings + reviewer recommendations:

  • Phase 2 principle pattern (packages/core/principles/08-*.test.ts), NOT manifest entry
  • Tighter framing «every framework artifact claiming new capability cites prior-art evaluation»
  • Honest framing: implementation phase (not docs-only), 3-5 days estimate
  • Real tested bash heuristic for capability detection
  • aif-evolve overlap explicit cross-ref в aif-comparison.md §9
  • STRONGLY RECOMMENDED for Phase 9 entry (не HARD BLOCKER preemptively)

Branch docs/phase-8.8-planning-hook stays local for reference; not deleted.

@artyhoo artyhoo closed this May 8, 2026
@artyhoo artyhoo mentioned this pull request May 8, 2026
Closed
9 tasks
artyhoo added a commit that referenced this pull request May 8, 2026
@artyhoo
fix(diff): globToRegex handles **/ zero-dirs + recipe glob redundancy…
9fe5a5b 
… cleanup

PR #11 review M1: globToRegex emitted `^src/app/.*/[^/]*\.tsx$` for
`src/app/**/*.tsx`, requiring at least one directory between `src/app/` and the
leaf file. Files at the base (src/app/page.tsx, src/app/layout.tsx) were
silently excluded from glob expansion, undercounting glob overlap when
comparing real divergent plans. Phase 8 acceptance trivially held at
similarity=1.0 (regen vs frozen are identical inputs, so the bug was invisible
to canonical-regen.test.ts), but Phase 9 entry research (Path A LLM gen ROI
scoping) consumes this metric on divergent plans where the bug would
understate glob overlap.

Fix: special-case `**/` as `(?:.*/)?` in `globToRegex` (+4 LOC). Tightened
`'glob-overlap edge'` test to assert exact 3/5 ratio (was loose `>0 <1`,
would have passed with broken regex). Added regression guard
`'**/ matches zero or more directory segments'`.

PR #11 review M2: removed redundant globs in three new recipes:
- next-r12-no-server-imports-in-client: dropped src/app/**/*.tsx and
  src/components/**/*.tsx (subsets of src/**/*.tsx)
- next-r14-require-form-safe-parse: dropped src/app/**/actions.ts
  (subset of src/app/**/*.ts)
- next-r20-require-use-server-directive: same cleanup as r14

Frozen canonical-v15 snapshot regenerated from updated recipes;
canonical-regen test still green at similarity=1.0.

Phase 8 retro updated: verification block #7 (LOC 84 -> 88), Created/modified
files block (84 LOC -> 88 LOC, 9 tests -> 10 tests), Self-reflection #2
appended with post-fix bump note, new Self-reflection #9 documenting the
PR #11 review fix and the meta-lesson (acceptance test invariance under
self-comparison is necessary but not sufficient for metric correctness).

Verified: 239/239 tests pass, make self-audit green.
@artyhoo artyhoo mentioned this pull request May 8, 2026
Merged
10 tasks
artyhoo added a commit that referenced this pull request May 8, 2026
@artyhoo
fix(husky): broaden semver detection + bump-vs-add differentiation + …
e982975 
…cleanup

PR #12 review M1+M2+L2.

M1 — broaden semver-prefix coverage in pa_is_new_dep_added:
  Old regex: `"\^?[0-9]` (caret or digit-prefix only).
  New regex: `"(\^|~|>=?|<=?|=|[0-9*])`.
  Now catches: `"^X"`, `"~X"`, `">=X"`, `"<=X"`, `">X"`, `"<X"`, `"=X"`,
  digit start, `"*"`. Still misses pure dist-tag versions ("latest" /
  "next" / "beta") and URL specifiers ("file:" / "git+" / "github:") —
  documented in the function comment as a known limitation.
  Rationale: scoping to broader patterns (any `[^"]+` value) would
  false-positive on package.json `scripts` entries which also follow
  the `"name": "value"` shape. Phase 9+ entry research can tighten via
  jq-based JSON parse if observed in practice.

M2 — bump-vs-add differentiation:
  Pre-fix: any `+` line matching the dep-shape regex flagged the commit
  as capability, including pure version bumps where the same key
  appears in both `+` and `-` lines.
  Post-fix: extract dep keys from both `+` and `-` lines via grep+sed;
  a key is "newly added" only if present in `+` but absent in `-`.
  Pure bumps no longer false-positive into capability detection.
  Implementation discipline: standard `for key in $added_keys` loop +
  `printf '%s\n' "$removed_keys" | grep -qFx "$key"` membership test.
  No subshell-return / pipe-while patterns (PR #9 m4 finding stays
  enforced).

L2 — remove unused `stripped` local var declaration in pa_check_trailer
  (leftover from earlier draft of the placeholder check; superseded by
  the word-tokenizer loop).

T9 hook test — added 2 sub-tests to validate the M1/M2 fixes:
  Test 7 (M2): seed an existing dep, then commit a version bump
    without a trailer. Hook must NOT flag as capability (exit 0).
  Test 8 (M1): commit a new dep with tilde-prefix version (`~1.0.0`)
    + valid trailer. Hook must flag as capability AND accept the
    trailer (exit 0). Pre-M1-fix would have missed the tilde and
    silently passed the commit through.
  Plus 2 new helper functions (seed_existing_dep, bump_existing_dep,
  add_capability_commit_with_tilde) to keep sub-test bodies thin.

Verified locally: bash -n on both files clean; 8/8 sub-tests pass
(original 6 still green + 2 new for M1/M2); principle 08 still 7/7;
make self-audit green (34/34 across all 8 principles).

Prior-art: prior-art-evaluations.md#1 (PR review tightening — same Autogrep entry rationale; bump-vs-add differentiation is mechanical refinement of an existing detector, not new capability area, no separate analog research needed).
artyhoo added a commit that referenced this pull request May 8, 2026
@artyhoo
docs(phase-9): T6 — review delta-fix (M1 calibration framing + m2 PR …
5fb5bf1 
…pointer + m3 forward-finding)

Closes 1 MAJOR + 2 MINOR review findings on Phase 9 prompt drafting:

- M1 — A8 «calibration» framing was overclaim. PHASE-9-PROMPT.md §4
  T3 said weights become «data-backed» — but 5 mutants × 3 dimensions
  ≈ 1 sample per dimension = regression guard for the formula, NOT
  statistical calibration of 0.40/0.40/0.20 values. Updated §4 T3
  «Why» paragraph + sub-task #4 doc-note text + verification grep
  probe (regression guard, not data-backed) to distinguish «formula
  behaves as designed» (what corpus proves) from «weight values are
  right» (still requires real divergent-plan data per Phase 8 retro
  Self-reflection #9).

- m2 — retro Versioning section did not point to next-step push +
  gh pr create. Added explicit «Next:» line at end of Versioning +
  2026-05-09 entry recording T6 review delta-fix landing. Also
  corrected the branch-name reference (drafting work is on
  docs/phase-9-prompt-draft after Phase 8.8.1 T0 split, not on
  docs/phase-9-coverage-gap).

- m3 — drafting session did not apply (yet-to-be-formalized) Phase
  8.8.1 search-coverage methodology to its own decisions (own-stack
  sweep on A8 corpus shape; adversarial check on N=15; surfaced
  Open Q from drafting itself = 0). Circular: Phase 8.8.1 not yet
  merged → methodology not formally required. Documented as Open Q
  #4 forward-finding for post-Phase-8.8.1-merge audit.

Reviewer-flagged BLOCKER B1 (drafting work «lost») was misdiagnosed
— drafting commits live on this branch (origin synced at 1c422cd);
T0 of Phase 8.8.1 meta-prompt split correctly. Reviewer's coverage
was bounded to one branch, missed the parallel docs/phase-9-prompt-draft.

PHASE-9-PROMPT.md = 307 LOC ≤500 cap unchanged; retro 119 → 124 LOC
(+5 from m2 closing + m3 Open Q #4) ≤200 cap.
@artyhoo artyhoo mentioned this pull request May 9, 2026
Merged
9 tasks
@artyhoo artyhoo mentioned this pull request May 21, 2026
Merged
artyhoo added a commit that referenced this pull request May 21, 2026
@artyhoo
fix(hooks): merge end-of-turn hook — deliver recap via reason (supers…
156b8a9 
…ede #81), 3 branches + visible marker (#89)

#81 (cfa28a3, on main) delivered the recap INSTRUCTION via systemMessage, which
is user-only and never reaches the model — verified dual-channel (CC hooks docs
Stop decision-control: "reason must be provided for Claude to know how to proceed"
+ WebSearch). So #81's recap prompt was never seen by the model. d695ac5/redesign
correctly use `reason` (injected; agent gets one more turn). This merges best-of-all-three:

- reason-delivery (model receives) + short systemMessage human note   [supersede #81 delivery bug]
- forced visible «## 🟢 Простыми словами» header                      [from #81 — human scannability]
- session-goal anchor (aiTitle→first-msg fallback) + 3-way drift verdict + recommendation-first  [from redesign]
- 3 branches incl. combined long+question (keeps recommendation-first when a fork is on the table)
- stateless-fire preserved (no /tmp, no aggregation/isMeta)

Tested 2026-05-21: bash -n; branches A/B/C; silent on bare tool_use; AskUserQuestion→B;
anchor aiTitle-primary + first-user-msg fallback + graceful default; §1.8 skip smoke-test exit 0.

Prior-art: skipped — modifies existing <50-LOC-logic CC hook (delivery-channel merge), no new capability per dual-implementation-discipline §2(iv); SSOT #8/#9/#20 cover the anchoring/hook family.
@artyhoo
artyhoo deleted the docs/phase-8.8-planning-hook branch May 22, 2026 19:15
artyhoo added a commit that referenced this pull request May 26, 2026
@artyhoo
research(bundle-autonomous): reframe B3 — Oracle composition over BUILD
983bca0 
Maintainer correction 2026-05-26 caught 4 invariant-level violations in
first draft of B3 sub-mechanism:

1. CI cast as primary evaluator (violates README CI=last-resort invariant)
2. Literal "halt-on-first-CI-red" despite REVISE=3 in same kickoff row
3. False claim "companion-projects only in memory" (§1.13 codified 2026-05-22)
4. "Oracle via subagent = expensive/complex" while session was using it for
   cold-review (91k tokens / 3 min / caught 2 MAJOR)

All four are nested T-traps on self (T3 + T16 + T11/T13 + recursive-self-
application gap). Cold-review subagent missed all four because its prompt
was structural-check, not invariant-check.

Patch changes:
- §0: add T-BA-E #ci-as-primary-evaluator trap (specialisation of T3+T16);
  enumerate it alongside existing T-BA-D #bundle-name-match
- §3 B3 table: add 4 new candidates with corrected scoring —
  oh-my-openagent (#8, ADOPT three-attempt+Oracle protocol, ~95% match,
  EXACT problem class), Aider --auto-test (#9, ADOPT success-path,
  ~85%), OpenHands AgentController (#10, ADOPT VOCABULARY + REFERENCE,
  ~80%), Claude Task Master autopilot (#11, ADOPT pause-not-halt
  semantics, ~80%); re-score Anthropic evaluator-optimizer ~40%→~90%
  with reframe (Oracle subagent IS our evaluator-LLM, not CI)
- §5 B3 verdict: rewrite from "ADOPT SDD + vocab from Argo/Anthropic" to
  ADOPT composition of 4 production patterns (oh-my-openagent +
  Aider + Task Master + Anthropic evaluator-optimizer + Superpowers SDD
  reviewer-loop); BUILD-delta ≤130 LOC (per-item-in-bundle wrapper +
  Oracle prompt template + bash glue), only what no upstream provides
- §7 forward-check: explicit T-BA-E listing + meta-insight section
  acknowledging this R-phase demonstrated the Oracle layer is missing in
  our own workflow (not just for B3); the very pattern this patch
  recommends building for B3 is the pattern this patch itself needed
  for its own discipline check

Companion memory codified:
- feedback_oracle_via_subagent_reuse_existing.md (4-checker before BUILD)
- feedback_cold_review_must_check_core_invariants.md (invariant-check
  template for future cold-review subagent prompts)
- feedback_b3_revise_not_halt_on_ci_red.md (REVISE semantics correction)
- feedback_ai_doc_research_priority_pool.md broadened scope (added
  runtime side; corrected oh-my-openagent canonical URL)

Research-only: zero SSOT row mutations, no helpers shipped. Per-sub-mech
verdict unchanged for B1/B2; B3 verdict substantially strengthened by
prior-art composition. Stage 2 maintainer GO/DEFER/DROP decision unblocks
Stage 3 I-phase.

Prior-art: skipped — research-patch only, no capability commit
@artyhoo artyhoo mentioned this pull request May 27, 2026
Merged
11 tasks
artyhoo added a commit that referenced this pull request May 28, 2026
@artyhoo
docs(meta-orch-no-arg-laziness): Phase -1 MINOR fixes — §4.5 AI-traps…
849326f 
… consolidation + F.6 self-contained rationale

Two of three MINOR findings from PR #261 Phase -1 cold-review applied (the third — SSOT #20 line-citation gap — was a false-positive; #20 already cites :L88 at patch:79 identical to peer rows).

MINOR #1 (#ai-traps-section-fragmented) — added new §4.5 "AI-traps applied (canonical + domain-specific)" consolidating the T-trap enumeration that was previously fragmented inline at §4 §1.7 line 333. Each T-number now names the specific action in this R-phase that honours it (T1 sampling floor=5 → §1.4 enumerated all 84 SSOT rows; T3 → file:line + verbatim excerpts at lines 140-143/87/93-95/101-105/109-113; etc.). T-N1/T-N2/T-N3/T-N4 listed with their honouring action too. The prior inline mention at §4 §1.7 line 333 now points forward to §4.5.

MINOR #3 (#f6-supplementary-rationale-relies-on-matrix-jump) — added "Supplementary, not subsumed by F.3 (self-contained statement)" paragraph at end of F.6 section. Now self-contained explanation: F.3-without-F.6 = Bug #3 fixed by side-effect but latent landmines remain; F.6-without-F.3 = Bugs #2/#4 unfixed → REJECT; combined F.3+F.6 = defence in depth on Bug #3. Reader no longer needs to jump to coverage matrix at lines 297-307 to understand why F.6 is additive not replacement.

MINOR #2 dropped — reviewer misread: #20 citation has explicit ":L88" line ref identical to peer rows (#9/#43/#68/#77). Verified at patch:79.

Prior-art: skipped — markdown polish on a research-patch; no new capability; per CLAUDE.md "Refactors, doc edits, test additions for existing capabilities ... NOT capability commits."
@artyhoo artyhoo mentioned this pull request Aug 1, 2026
Merged
This was referenced Aug 16, 2026
Merged
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

1 participant

@artyhoo