
CLI Developer Experience expert panel. Use when user says 'Ask the DX council' or needs expert opinions on CLI design, developer experience, error messages, command structure, or UX decisions. Spawns parallel research for multiple CLI design experts.
Community Lead and Technical Writer for NeoHaskell. Use for documentation, tutorials, GitHub issues, PR descriptions, release notes, and contributor guides. Writes for "Jess" - a time-constrained junior developer. Handles pipeline phase 14 (PR body).
Developer Experience review for NeoHaskell APIs. Use when evaluating API intuitiveness, naming, pipe-friendliness, and consistency. Handles pipeline phase 4 (DevEx Review).
Orchestrates NeoHaskell feature implementation using a structured 17-phase pipeline. Use when implementing new features, types, or modules for NeoHaskell. Triggers on requests like 'implement feature', 'new NeoHaskell feature', 'feature pipeline', 'implement issue
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-ground description: Applies the grounding loop to raw security findings to filter overkill and surface real blockers. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Security Findings Filters raw security findings through the 4-question ground
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 15-final-verify description: Runs build and test as a strict no-iteration gate before PR creation; records final hlint warnings for the PR body. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Final Verify Strict gate on `cabal build all` and
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-submit description: Commits and pushes the implementation, updates the phase-3 draft PR's title/body, and marks it ready for review. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Submit PR (mark draft ready) Commits and pushes the implement
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-security-adr description: Runs the ADR-level security review (threat-model → ground → record). kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Security review (ADR) Runs a threat-model pass, then the mandatory grounding pass against the fea
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-decide description: Judges the feature's complexity tier from the ADR, issue, and diff scope, with explicit rationale. kind: leaf executor: opus model: claude-opus-4-7 --- # Decide complexity tier Returns the single tier that best describes the feature's blast
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 05-performance-adr description: Runs the performance review on the ADR, producing grounded findings against the 50k req/s target. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Performance ADR Review Produces `.pipeline/findings-05.json` with
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 14-fix-findings description: Addresses every blocker finding from phases 12 and 13 and re-runs build and tests. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Findings Reads the implementation security and performance findings and resolves every bl
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-fix-bot-comments description: Reads CodeRabbit and other bot review comments, addresses them, and pushes fix-up commits. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Bot Comments Reads CodeRabbit / bot comments on the PR, addresses each, and p
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: integration-pipeline-preview description: Use when implementing a new outbound NeoHaskell integration end-to-end — from design draft through PR merge — and the work needs the structured 17-phase pipeline with design, security, and performance reviews (each grounded
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-implementer description: NeoHaskell code implementation guide. Use when implementing features, writing tests, build/test loops, or any task requiring NeoHaskell code. Handles pipeline phases 7-9 (tests, implementation, build loop), 12-13 (fix reviews, fin
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 17-ci-cycle description: Waits on CI, addresses bot comments, and waits for maintainer merge. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # CI Cycle Waits on CI checks, addresses bot review comments, and waits for the maintainer to merge. #
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-await-merge description: Prints maintainer merge instructions and closes the pipeline. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Await Merge Prints merge instructions to the maintainer and marks phase 17 complete. Approval is reserved f
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-deep-audit description: Reads the implementation source and applies the security methodology to produce file-line findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Deep Security Audit Reads changed source files plus siblings and applies every
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-deep-audit description: Reads the implementation source and applies the performance methodology to produce file-line findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Deep Performance Audit Reads changed source files plus siblings and applies
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 13-performance-impl description: Runs the implementation-phase performance review with a static scan, deep audit, grounding, and record. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Performance Review (Implementation) Produces `.integration
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-submit description: Commits and pushes the implementation, updates the phase-3 draft PR's title/body, and marks it ready for review. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Submit PR (mark draft ready) Stages and commits the implement
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-hlint description: Runs hlint on changed files and captures warnings for later surfacing in the PR body. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Hlint Runs `hlint` on the changed Haskell files and writes the output to `.integration-pi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 11-build-loop description: Iterates build and test until both are green, then captures hlint warnings without gating. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Build Loop Drives the implementation to a state where `cabal build all` and `
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-record description: Persists grounded implementation security findings and marks phase 12 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Implementation Security Findings Reads grounded findings on stdin, writes `.integration
Performance review for NeoHaskell targeting 50k req/s throughput. Use when reviewing code for performance implications, INLINE pragmas, strictness, and allocation patterns. Handles pipeline phases 3 (ADR review) and 11 (implementation review).
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-ground description: Applies the grounding loop to merged static and deep security findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Implementation Security Findings Filters merged static + deep findings through the 4-question grounding
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-build description: Runs cabal build all and streams the output to the pipeline log. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Build Runs `nix develop --command cabal build all` and writes the output to `.integration-pipeline/build.log`.
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-static-scan description: Runs the static security checks script across changed files and emits raw findings JSON. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Static Security Scan Runs `sec-static-checks.py` against the implementation's ch
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-static-scan description: Runs the static performance checks script across changed files and emits raw findings JSON. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Static Performance Scan Runs `perf-static-checks.py` against the implementati
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 12-security-impl description: Runs the implementation-phase security review with a static scan, deep audit, grounding, and record. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Security Review (Implementation) Produces `.integration-pipeline
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-fix-iter description: Reads build and test logs, identifies the root cause, and edits implementation files to fix it. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Iter Reads the latest build and test logs and applies a targeted fix to the impl
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-record description: Persists grounded implementation performance findings and marks phase 13 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Implementation Performance Findings Reads grounded findings on stdin, writes `.integ
Language design expert panel. Use when user says 'ask the council' or needs expert opinions on language syntax, DX tradeoffs, or ergonomics questions for NeoHaskell. Selects 3 relevant experts from a 19-person roster and researches their actual positions.
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-hlint description: Runs hlint as a final pass that captures warnings for the PR body without gating the pipeline. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Hlint Runs `hlint` against the changed Haskell files one last time and wri
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-build description: Runs cabal build all as a strict gate with no retry. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Build Runs `nix develop --command cabal build all` and writes the output to `.integration-pipeline/final-build.log`.
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 15-final-verify description: Runs build and test as a strict no-iteration gate before PR creation; records final hlint warnings for the PR body. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Final Verify Strict gate on `cabal build all` and
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-pr-body description: Writes the PR body to a file with a conventional-commit title, a checklist, and any hlint warnings captured from earlier phases. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # PR Body Writes `.integration-pipeline/pr-body.md` wi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-test description: Runs cabal test as a strict gate with no retry. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Test Runs `nix develop --command cabal test` and writes the output to `.integration-pipeline/final-test.log`. Non-zero exi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 16-create-pr description: Finalizes the PR — writes the final body/title, commits and pushes the implementation, and marks the phase-3 draft PR ready for review. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Finalize PR (mark ready) Generate
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-test description: Runs cabal test with streaming details and writes the output to the pipeline log. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Test Runs `nix develop --command cabal test --test-show-details=streaming` and writes the outp
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-community-writer description: Community Lead and Technical Writer for NeoHaskell. Use for documentation, tutorials, GitHub issues, PR descriptions, release notes, and contributor guides. Writes for "Jess" - a time-constrained junior developer. Handles pip
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-performance-review description: Performance review for NeoHaskell targeting 50k req/s throughput. Use when reviewing code for performance implications, INLINE pragmas, strictness, and allocation patterns. Handles pipeline phases 3 (ADR review) and 11 (imp
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-adr-architect description: Interactive ADR Architect for NeoHaskell. Use when the user has a vague feature idea, GitHub issue, or half-formed thought and wants to produce a full ADR through iterative interviewing. Triggers on 'adr architect', 'help me wri
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-adr-template description: ADR (Architecture Decision Record) template for NeoHaskell. Use when creating a new ADR, drafting a decision document, or starting a new feature proposal. Triggers on 'new ADR', 'create ADR', 'architecture decision', 'decision re
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-security-review description: Security & Code Quality review for NeoHaskell. Use when reviewing code changes, PRs, or architectural decisions for security implications. Evaluates OWASP, NIST, EU compliance. Handles pipeline phases 2 (ADR review) and 10 (im
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-devex-review description: Developer Experience review for NeoHaskell APIs. Use when evaluating API intuitiveness, naming, pipe-friendliness, and consistency. Handles pipeline phase 4 (DevEx Review). --- # NeoHaskell DevEx Review You are the Developer Ex
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-style-guide description: NeoHaskell coding style reference and enforcement rules. Load when writing, reviewing, or modifying any Haskell code in the NeoHaskell project. Triggers on 'NeoHaskell style', 'NeoHaskell conventions', 'how to write NeoHaskell', '
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-qa-designer description: Testing QA Designer for NeoHaskell. Designs comprehensive test specifications with exhaustive edge cases, boundary conditions, and happy paths BEFORE implementation. Handles pipeline phase 6 (Test Spec Design). Outside-in TDD meth
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-wait-checks description: Watches GitHub PR checks until they pass or fail. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Wait Checks Runs `gh pr checks <pr_number> --watch` and exits 0 only when every check passes. ## Inputs - `pipeline.p
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: neohaskell-feature-pipeline description: Orchestrates NeoHaskell feature implementation using a structured 17-phase pipeline. Use when implementing new features, types, or modules for NeoHaskell. Triggers on requests like 'implement feature', 'new NeoHaskell feature
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 09-test-writing description: Translates the test spec into compiling hspec tests that all fail against stub implementations. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Test Writing Translates the test spec into Haskell `hspec` tests and stub implem
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 10-implementation description: Implements the feature per the architecture doc using nhcore-first, NeoHaskell style. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Implementation Replaces the phase 9 stubs with a real implementation that follows the ar
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the test specification covering happy paths, edges, errors, and properties. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce test spec Reads the architecture doc and writes `docs/architecture/<adr-number>-<slug>-tests
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-pr-body description: Writes the PR body to a file with a conventional-commit title, a checklist, and any hlint warnings captured from earlier phases. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # PR Body Writes `.pipeline/pr-body.md` with the PR ti
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 16-create-pr description: Finalizes the PR — writes the final body/title, commits and pushes the implementation, and marks the phase-3 draft PR ready for review. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Finalize PR (mark ready) Generate
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-persist description: Persists the classifier's JSON output to the pipeline state and marks phase 2 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Persist classification Reads the decide step's JSON output, calls `pipeline.py classi
ADR (Architecture Decision Record) template for NeoHaskell. Use when creating a new ADR, drafting a decision document, or starting a new feature proposal. Triggers on 'new ADR', 'create ADR', 'architecture decision', 'decision record', 'feature proposal', 'ADR draft'.
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-ground description: Applies the grounding loop to merged static and deep performance findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Implementation Performance Findings Filters merged static + deep performance findings through the 4-
NeoHaskell coding style reference and enforcement rules. Load when writing, reviewing, or modifying any Haskell code in the NeoHaskell project. Triggers on 'NeoHaskell style', 'NeoHaskell conventions', 'how to write NeoHaskell', 'code style', 'style guide'.
NeoHaskell code implementation guide. Use when implementing features, writing tests, build/test loops, or any task requiring NeoHaskell code. Handles pipeline phases 7-9 (tests, implementation, build loop), 12-13 (fix reviews, final build), and 16 (fix bot comments).
Security & Code Quality review for NeoHaskell. Use when reviewing code changes, PRs, or architectural decisions for security implications. Evaluates OWASP, NIST, EU compliance. Handles pipeline phases 2 (ADR review) and 10 (implementation review).
Testing QA Designer for NeoHaskell. Designs comprehensive test specifications with exhaustive edge cases, boundary conditions, and happy paths BEFORE implementation. Handles pipeline phase 6 (Test Spec Design). Outside-in TDD methodology.
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-build description: Runs cabal build all and streams the output to the pipeline log. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Build Runs `nix develop --command cabal build all` and writes the output to `.pipeline/build.log`. ## Inputs
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 05-performance-design description: Runs the performance review on the design draft, producing grounded findings against the 50k req/s target. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Performance design review Produces `.integration-pipe
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the DevEx review document for the feature's ADR. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce DevEx review Reads the ADR and the grounded security/performance findings, evaluates the API surface against the Jess p
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-init-pipeline description: Initialises the pipeline state machine for a NeoHaskell feature. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Init pipeline Bootstraps `.pipeline/state.json` for the feature so subsequent phases can drive against
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: dx-council-lang description: Language design expert panel. Use when user says 'ask the council' or needs expert opinions on language syntax, DX tradeoffs, or ergonomics questions for NeoHaskell. Selects 3 relevant experts from a 19-person roster and researches their
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: dx-council-cli description: CLI Developer Experience expert panel. Use when user says 'Ask the DX council' or needs expert opinions on CLI design, developer experience, error messages, command structure, or UX decisions. Spawns parallel research for multiple CLI des
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 07-architecture-design description: Runs the architecture design phase (produce → review-quality) without a human pause; advances when the rubric verdict is pass. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Architecture design Produces the
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 08-test-spec-design description: Runs the test spec design phase (produce → review-quality) without a human pause; advances when the rubric verdict is pass. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Test spec design Produces the test spe
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-record description: Persists grounded performance findings to the pipeline state and marks phase 5 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Performance Findings Reads grounded findings on stdin, writes `.integration-pi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-hot-path-analysis description: Analyses the design draft against the performance methodology to surface raw hot-path findings. kind: leaf executor: opus model: claude-opus-4-7 --- # Hot-Path Analysis Reads the design draft and asks every section of the performa
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the DevEx review against the rubric and gates phase 6 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review DevEx-review quality Reads the DevEx review and the design draft, applies eve
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the test specification against the rubric and gates phase 8 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review test-spec quality Reads the produced test spec and the architecture doc
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 08-test-spec-design description: Runs the test spec design phase (produce → review-quality) without a human pause; advances when the rubric verdict is pass. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Test spec design Produces the test spe
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-adr-draft description: Drafts the ADR (Status Proposed) and opens a draft PR so the maintainer can review the ADR on GitHub. kind: leaf executor: opus model: claude-opus-4-7 --- # ADR draft Produces `docs/decisions/NNNN-<slug>.md` matching the NeoHaskell ADR te
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-threat-model description: Produces raw security findings from the ADR using OWASP + STRIDE applied to a feature diff. kind: leaf executor: opus model: claude-opus-4-7 --- # Threat model Reads the proposed ADR and produces a raw findings array using the methodol
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-classify-feature description: Classifies the feature into a complexity tier via opus judgment, then persists it for the grounding loop. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Classify feature Decides the feature's tier (`trivial` /
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the implementation-ready architecture document for the feature. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce architecture doc Reads the ADR + DevEx review and writes `docs/architecture/<adr-number>-<slug>.md` with
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 06-devex-review description: Runs the DevEx review (produce → review-quality) without a human pause; advances automatically when the review passes the rubric. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # DevEx review Produces a DevEx review
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-persist description: Persists the classifier's JSON output to the pipeline state and marks phase 2 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Persist classification Reads the decide step's JSON output, calls `pipeline.py classi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-hlint description: Runs hlint on changed files and captures warnings for later surfacing in the PR body. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Hlint Runs `hlint` on the changed Haskell files and writes the output to `.pipeline/hlint
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 11-build-loop description: Iterates build and test until both are green, then captures hlint warnings without gating. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Build Loop Drives the implementation to a state where `cabal build all` and `
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-record description: Persists grounded security findings to the pipeline state and marks phase 4 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Security Findings Reads grounded findings on stdin, writes `.pipeline/findings-04
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-hot-path-analysis description: Analyses the ADR against the performance methodology to surface raw hot-path findings. kind: leaf executor: opus model: claude-opus-4-7 --- # Hot-Path Analysis Reads the ADR draft and asks every section of the performance methodol
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-ground description: Applies the grounding loop to raw performance findings to filter overkill and surface real blockers. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Performance Findings Filters raw performance findings through the 4-questi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the architecture document against the rubric and gates phase 7 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review architecture-doc quality Reads the produced architecture doc and the
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the DevEx review against the rubric and gates phase 6 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review DevEx-review quality Reads the DevEx review and the ADR, applies every check
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-record description: Persists grounded performance findings to the pipeline state and marks phase 5 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Performance Findings Reads grounded findings on stdin, writes `.pipeline/findi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-test description: Runs cabal test with streaming details and writes the output to the pipeline log. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Test Runs `nix develop --command cabal test --test-show-details=streaming` and writes the outp
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-fix-iter description: Reads build and test logs, identifies the root cause, and edits implementation files to fix it. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Iter Reads the latest build and test logs and applies a targeted fix to the impl
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-ground description: Applies the grounding loop to raw security findings to filter overkill and surface real blockers. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Security Findings Filters raw security findings through the 4-question ground
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-security-design description: Runs the design-level security review for the outbound integration (threat-model → ground → record). kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Security review (design) Runs a threat-model pass, then the ma
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: feature-pipeline-preview description: Use when implementing a new NeoHaskell feature end-to-end — from ADR draft through PR merge — and the work needs the structured 17-phase pipeline with ADR, security and performance reviews (each grounded against feature complexi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-ground description: Applies the grounding loop to merged static and deep security findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Implementation Security Findings Filters merged static + deep findings through the 4-question grounding
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-static-scan description: Runs the static security checks script across changed files and emits raw findings JSON. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Static Security Scan Runs `sec-static-checks.py` against the implementation's ch
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-deep-audit description: Reads the implementation source and applies the security methodology to produce file-line findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Deep Security Audit Reads changed source files plus siblings and applies every
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-record description: Persists grounded implementation security findings and marks phase 12 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Implementation Security Findings Reads grounded findings on stdin, writes `.pipeline/fi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-deep-audit description: Reads the implementation source and applies the performance methodology to produce file-line findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Deep Performance Audit Reads changed source files plus siblings and applies
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-ground description: Applies the grounding loop to merged static and deep performance findings. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Implementation Performance Findings Filters merged static + deep performance findings through the 4-
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-review description: Opus reads the PR diff against the NeoHaskell quality rubric, writes findings to .pipeline/findings-17.json, and surfaces a digest in conversation. kind: leaf executor: opus model: claude-opus-4-7 --- # PR Review (Opus) Opus reads the open P
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-static-scan description: Runs the static performance checks script across changed files and emits raw findings JSON. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Static Performance Scan Runs `perf-static-checks.py` against the implementati
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the test specification against the rubric and gates phase 8 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review test-spec quality Reads the produced test spec and the architecture doc
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 04-record description: Persists grounded implementation performance findings and marks phase 13 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Implementation Performance Findings Reads grounded findings on stdin, writes `.pipel
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-init-pipeline description: Initialises the pipeline state machine for a NeoHaskell outbound integration. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Init pipeline Bootstraps `.integration-pipeline/state.json` for the integration so subseq
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-apply description: Applies the maintainer-approved findings-17 fixes, builds, tests, commits per finding, and pushes the branch. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Apply approved fixes Invoked AFTER `pipeline.py approve 17`. Reads `.pipe
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-build description: Runs cabal build all as a strict gate with no retry. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Build Runs `nix develop --command cabal build all` and writes the output to `.pipeline/final-build.log`. Non-zero ex
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 17-opus-pr-review description: Opus reviews the open PR locally, proposes fixes, pauses for maintainer approval; then applies the accepted fixes. kind: process executor: opus model: claude-opus-4-7 --- # Opus PR Review After the PR is opened (phase 16), an Opus re
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 13-performance-impl description: Runs the implementation-phase performance review with a static scan, deep audit, grounding, and record. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Performance Review (Implementation) Produces `.pipeline/fi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 14-fix-findings description: Addresses every blocker finding from phases 12 and 13 and re-runs build and tests. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Findings Reads the implementation security and performance findings and resolves every bl
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-hlint description: Runs hlint as a final pass that captures warnings for the PR body without gating the pipeline. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Hlint Runs `hlint` against the changed Haskell files one last time and wri
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-wait-checks description: Watches GitHub PR checks until they pass or fail. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Wait Checks Runs `gh pr checks <pr_number> --watch` and exits 0 only when every check passes. ## Inputs - `pipeline.p
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-record description: Persists grounded security findings to the pipeline state and marks phase 4 complete. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Record Security Findings Reads grounded findings on stdin, writes `.integration-pipeline
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-await-merge description: Prints maintainer merge instructions and closes the pipeline. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Await Merge Prints merge instructions to the maintainer and marks phase 18 complete. Approval is reserved f
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-fix-bot-comments description: Reads CodeRabbit and other bot review comments, addresses them, and pushes fix-up commits. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Fix Bot Comments Reads CodeRabbit / bot comments on the PR, addresses each, and p
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 12-security-impl description: Runs the implementation-phase security review with a static scan, deep audit, grounding, and record. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Security Review (Implementation) Produces `.pipeline/findings-12
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 17-ci-cycle description: Waits on CI, addresses bot comments, and waits for maintainer merge. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # CI Cycle Waits on CI checks, addresses bot review comments, and waits for the maintainer to merge. #
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-test description: Runs cabal test as a strict gate with no retry. kind: leaf executor: script model: claude-haiku-4-5-20251001 --- # Final Test Runs `nix develop --command cabal test` and writes the output to `.pipeline/final-test.log`. Non-zero exit aborts the
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the implementation-ready architecture document for the integration. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce architecture doc Reads the design draft + DevEx review and writes `.integration-pipeline/integration
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-ground description: Applies the grounding loop to raw performance findings to filter overkill and surface real blockers. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Ground Performance Findings Filters raw performance findings through the 4-questi
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 03-design-draft description: Drafts the outbound integration's design document at .integration-pipeline/integration-design.md in Status Proposed, then opens a draft PR carrying the design in its body so the maintainer can review it on GitHub. kind: leaf executor: op
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-classify-integration description: Classifies the integration into a complexity tier via opus judgment, then persists it for the grounding loop. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Classify integration Decides the integration's t
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-threat-model description: Produces raw security findings from the design draft using OWASP + STRIDE applied to a feature diff. kind: leaf executor: opus model: claude-opus-4-7 --- # Threat model Reads the proposed design draft and produces a raw findings array
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-decide description: Judges the integration's complexity tier from the design draft, issue, and diff scope, with explicit rationale. kind: leaf executor: opus model: claude-opus-4-7 --- # Decide complexity tier Returns the single tier that best describes the int
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 02-review-quality description: Independently validates the architecture document against the rubric and gates phase 7 completion. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Review architecture-doc quality Reads the produced architecture doc and the
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 06-devex-review description: Runs the DevEx review (produce → review-quality) without a human pause; advances automatically when the review passes the rubric. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # DevEx review Produces a DevEx review
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the DevEx review document for the integration design draft. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce DevEx review Reads the design draft and the grounded security/performance findings, evaluates the API surfac
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 07-architecture-design description: Runs the architecture design phase (produce → review-quality) without a human pause; advances when the rubric verdict is pass. kind: process executor: haiku model: claude-haiku-4-5-20251001 --- # Architecture design Produces the
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 09-test-writing description: Translates the test spec into compiling hspec tests that all fail against stub implementations. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Test Writing Translates the test spec into Haskell `hspec` tests and stub implem
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 01-produce description: Produces the test specification covering happy paths, edges, errors, and properties. kind: leaf executor: opus model: claude-opus-4-7 --- # Produce test spec Reads the architecture doc and writes `.integration-pipeline/integration-tests.md`
> **ARCHIVAL KNOWLEDGE — DO NOT USE OR REFERENCE.** > Superseded by `codemap/` + root `AGENTS.md`. Scheduled for deletion once the pipeline is verified (Phase 6). > Manifest: `docs/archive/2026-07-ai-artifacts/MANIFEST.md` --- name: 10-implementation description: Implements the integration under the `integrations/Integration/[Name]/` module tree per the architecture doc using nhcore-first, NeoHaskell style. kind: leaf executor: sonnet model: claude-sonnet-4-6 --- # Implementation Replaces the