Skip to content

v7.0.0: Mutation testing, structured repair, goal scoring, repo maps, optimize mode#9

Merged
bc1plainview merged 2 commits into
mainfrom
loop/v7-mutation-scoring
Mar 13, 2026
Merged

v7.0.0: Mutation testing, structured repair, goal scoring, repo maps, optimize mode#9
bc1plainview merged 2 commits into
mainfrom
loop/v7-mutation-scoring

Conversation

@bc1plainview

Copy link
Copy Markdown
Owner

Summary

  • Mutation testing gate: scripts/mutate-contract.sh applies 20 sed-level mutation operators to contracts, blocks loop exit below 70% kill rate
  • Structured repair (Agentless): scripts/localize-failure.sh + R1/R2/R3 phases replace full agent re-dispatch with targeted Localize→Patch→Validate
  • Goal-oriented scoring (CodeClash): scripts/score-build.sh scores across 4 dimensions (spec coverage, security delta, mutation score, code health) — ALL must meet thresholds
  • Cross-layer repo map: scripts/build-repo-map.sh generates shared interface map from ABI, referenced by all 12 domain agents
  • Autoresearch optimize: commands/buidl-optimize.md runs autonomous optimization loops for measurable goals without human gates

Changes

  • 24 files changed, 2193 insertions, 28 deletions
  • 6 new files: 5 scripts + 1 command
  • All 12 domain agents updated with repo-map reference
  • 110 new tests (607/607 passing)

Verification

  1. mutate-contract.sh with empty test dir → mutation_score: 0, verdict: FAIL
  2. score-build.sh → progress-tracker.yaml with 4 dimension scores
  3. build-repo-map.sh with minimal ABI → repo-map.md with Contract Layer
  4. buidl-optimize.md exists with correct frontmatter
  5. grep -r "repo-map.md" agents/ → 12 agent files

Test plan

  • bash tests/plugin-tests.sh — 607/607 PASS
  • All 5 completion criteria verified

… optimize mode

- scripts/mutate-contract.sh: 20 sed-level mutation operators, 70% threshold gate
- scripts/localize-failure.sh: structured failure localization (Agentless pattern)
- scripts/extract-requirements.sh: spec requirements extraction for coverage tracking
- scripts/score-build.sh: 4-dimension build scoring (spec/security/mutation/health)
- scripts/build-repo-map.sh: hierarchical cross-layer interface map from ABI
- commands/buidl-optimize.md: autonomous optimization loop (Karpathy/Shopify pattern)
- Phase R1/R2/R3 structured repair replacing full agent re-dispatch
- Cross-layer repo map shared across all 12 domain agents
- 607/607 tests passing (110 new tests for v7 features)
…stash inversion

- MAJOR: build-repo-map.sh — init variables before isinstance check (NameError fix)
- MAJOR: mutate-contract.sh — add source restoration to trap handler (corruption guard)
- MINOR: Remove no-op cp, add /g flag to gte-to-lte operator, fix stash to git checkout
- 607/607 tests passing
@bc1plainview bc1plainview merged commit c8e00e1 into main Mar 13, 2026
1 check failed
@bc1plainview bc1plainview deleted the loop/v7-mutation-scoring branch March 13, 2026 13:45
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