peaks-loop 3.1.2 → 4.0.0-beta.10

package/CHANGELOG.md

@@ -1,24 +1,484 @@
 # Changelog
 
+## [Unreleased]
+
+## 4.0.0-beta.10 — 2026-07-16
+
+### Status: PRE-IMPLEMENTATION
+
+> **This CHANGELOG entry documents the CONTRACT for 4.0.0-beta.10.
+> The 3 slices below are NOT YET IMPLEMENTED in this branch.**
+> See `docs/release/4.0.0-beta.10.md` for the user-facing runbook
+> and `peaks workflow verify-pipeline --rid 2026-07-15-cli-surface-cleanup`
+> for the pre-implementation gate (9/9 gates PASS).
+>
+> **Ship path**: user implements Slices 1+2+3 per the runbook,
+> runs `pnpm test:full` + manual AC verification, then runs
+> `npm publish --tag beta`. No code change is included in the
+> sediment/version-bump commit that produced this CHANGELOG.
+
+### Removed — `peaks minimax-worker` (Slice 1)
+
+- **`peaks minimax-worker` CLI removed** — historical MiniMax
+  provider integration. `peaks minimax-worker --help` now exits
+  non-zero with `COMMAND_NOT_FOUND`.
+- **`peaks worker minimax` CLI removed** — same module.
+- **`peaks config provider minimax *` subtree removed** —
+  `peaks config provider minimax set --help` exits non-zero.
+- **6 source files deleted**:
+  - `src/cli/commands/worker-commands.ts`
+  - `src/services/providers/minimax-provider-service.ts`
+  - `src/services/providers/minimax-worker-service.ts`
+  - `src/services/config/provider-service.ts` (entire file; 100% MiniMax)
+  - `tests/unit/minimax-worker-service.test.ts`
+  - `tests/unit/minimax-provider-service.test.ts`
+- **~21 source files modified** (drop imports, drop function bodies,
+  drop registrations, replace `'minimax-2.7'` literal with
+  `'claude-opus-4-7'` in 5 files).
+- **~8 test files modified** (drop MiniMax cases; replace literal in 3).
+- **4 openspec files updated** (1-line wording fixes).
+- **Net security improvement**: smaller secret-handling surface
+  (5 MiniMax functions removed).
+
+### Changed — 10 role-skill CLI commands hidden (Slice 2)
+
+- **10 top-level role-skill CLI commands** now hidden from `peaks --help`
+  via `Commander.hidden()`. Still invokable for internal
+  `peaks sub-agent dispatch --role <role>` paths. Final canonical
+  list (per QA grep audit):
+  - `peaks prd`
+  - `peaks qa`
+  - `peaks sc`
+  - `peaks audit`
+  - `peaks code-review`
+  - `peaks perf-audit`
+  - `peaks security-audit`
+  - `peaks upgrade`
+  - `peaks agent`
+  - `peaks code`
+- **`peaks skill list`** now excludes skills with
+  `visibility: internal` frontmatter. Use `--include-internal` to opt in.
+- **`peaks skill search`** accepts `--include-internal` flag.
+- **8 SKILL.md files** updated to add/rename `visibility: internal`:
+  `peaks-prd`, `peaks-qa`, `peaks-rd`, `peaks-sc`, `peaks-ui`,
+  `peaks-txt`, `peaks-final-review`, `peaks-perf-audit`,
+  `peaks-security-audit`.
+- **Integration test rewritten**:
+  `tests/integration/skill-search-cli.test.ts:72-79` now asserts
+  the new `visibility:` frontmatter field instead of legacy
+  `userInvocable`.
+
+### Added — `peaks ecc install|status|ls|show` (Slice 3)
+
+> **PIVOT vs initial design**: Originally Slice 3 was supposed to
+> spawn `<cached>/ecc agent run <name> --json` as a subprocess.
+> RD sub-agent triggered **Gate S3-0**: affaan-m/everything-claude-code
+> has no `ecc` binary; its real structure is `agents/*.md` flat
+> + SKILL.md descriptors. User chose Option B — drop the subprocess
+> model entirely.
+
+- **`peaks ecc install`** — downloads affaan-m/everything-claude-code
+  from GitHub releases to `~/.peaks/cache/ecc-<sha>/`. Selective
+  tarball extraction: ONLY `agents/` subtree (skip `rules/`,
+  `commands/`, `settings/`, `docs/`, `README.md`).
+- **`peaks ecc status`** — reads cache manifest, reports version + sha
+  + agent count.
+- **`peaks ecc ls`** — lists cached agents by parsing `agents/*.md`
+  frontmatter.
+- **`peaks ecc show <name>`** — outputs the agent's SKILL.md body
+  to stdout. LLM-side consumers (Skill-first path) read this output
+  to apply the agent's instructions.
+- **3 source files deleted** (with Slice 1):
+  - `src/cli/commands/agent-commands.ts` (the old `peaks agent run/list`)
+  - `src/services/agent/ecc-agent-service.ts` (the spawn orchestrator)
+  - `tests/unit/services/agent/ecc-agent-service.test.ts`
+- **3 new source files**:
+  - `src/services/agent/ecc-cache-service.ts` — pure-IO cache module
+    (`downloadToCache`, `readCacheManifest`, `listCachedAgents`,
+    `readAgentSkill`, `cleanupStaleCache`)
+  - `src/cli/commands/ecc-commands.ts` — 4 subcommands
+  - `tests/unit/agent/ecc-cache-service.test.ts`
+- **Dead-probes removed**:
+  - `npx ecc --version` probe (was: never worked; affaan-m/ECC not on npm)
+  - `npx ecc-agentshield --version` probe in
+    `src/services/audit/static-service.ts:104` (parallel fix)
+- **7-day cache TTL** via `bootstrapLogger.applyRetention` sweep
+  (wired at `src/cli/program.ts:121`). NOT via
+  `peaks doctor --cleanup-stale` (which is bound to `dropStale()`
+  for the binding store).
+
+### Removed — `peaks agent run` and `peaks agent list`
+
+- **`peaks agent run <name>` exits non-zero with `COMMAND_NOT_FOUND`**.
+- **`peaks agent list` exits non-zero with `COMMAND_NOT_FOUND`**.
+- **No replacement CLI** — LLM-side consumers use
+  `peaks ecc show <name>` or read `<cache>/agents/<name>.md` directly.
+
+### Net security / performance impact
+
+- **Security**: 0 subprocess attack surface (was: 1 per call in beta.9);
+  smaller secret-handling surface (5 MiniMax functions removed).
+- **Performance**: cache-dir probe is pure read (<15ms vs ~500ms
+  for the npm-registry probe in beta.9); selective tarball extraction
+  saves ~95% disk footprint.
+
+### Pre-implementation gate
+
+`peaks workflow verify-pipeline --rid 2026-07-15-cli-surface-cleanup`
+returns `ok: true, complete: true, violations: []` (9/9 gates PASS).
+This CHANGELOG entry is the contract; the actual implementation
+follows the runbook in `docs/release/4.0.0-beta.10.md`.
+
+### Acceptance criteria
+
+27 ACs total: 7 (Slice 1) + 8 (Slice 2) + 11 (Slice 3 redesign + 1).
+Full list in `docs/release/4.0.0-beta.10.md` §Acceptance criteria.
+
+---
+
+## 4.0.0-beta.9 — 2026-07-15
+
+### Fixed — `npm install peaks-loop` runtime path layout
+
+`4.0.0-beta.8` shipped a tarball where `bin/peaks.js` imported
+`'../dist/src/cli/index.js'`, but tsc actually emitted the runtime
+files at `dist/cli/index.js` (rootDir walked past `src/` because of
+cross-folder imports like `src/services/hooks/output.ts → ../../cli/`).
+Downstream `npm i -g peaks-loop` then `peaks -v` failed with
+`ERR_MODULE_NOT_FOUND: .../dist/src/cli/index.js`. The 4 audit/business
+templates also failed to enter the tarball because `package.json#files[]`
+whitelisted `dist/src/**/*.js` / `.d.ts` but never `.md`.
+
+This release:
+
+- `bin/peaks.js` import: `'../dist/src/cli/index.js'` → `'../dist/cli/index.js'`.
+- `scripts/copy-templates.mjs` dest: `dist/src/services/...` →
+  `dist/services/...` so the bundled templates land where the runtime
+  `import.meta.url` resolver expects them.
+- `package.json#files[]`: prefixes shifted from `dist/src/` to `dist/`,
+  plus a new `dist/**/*.md` entry so the 4 templates bundled by
+  `copy-templates.mjs` actually ship.
+- `tsconfig.build.json` (introduced in beta.8) confirmed correct — it
+  is the build-time emit that controls the dist tree shape.
+
+End-to-end smoke after the fix (local):
+
+```
+$ npm pack                                  # 30 MB tarball
+$ npm install ./peaks-loop-4.0.0-beta.9.tgz --prefix /tmp/test
+$ peaks --version                           # → "4.0.0-beta.9"
+$ peaks workspace init --project /tmp/fresh  # → templatesBooted: 5
+```
+
+`.peaks/project-scan/{project-scan.md, business-knowledge.md,
+security-template.md, perf-template.md, audit-output-schema.md}`
+all present after a fresh workspace init.
+
+### Notes
+
+- 4.0.0-beta.8 is left in place on npm as a known-broken release.
+  Consumers who already pinned `peaks-loop@4.0.0-beta.8` should bump
+  to `4.0.0-beta.9`.
+- Beta.9 contains everything beta.8 advertised plus the path fix.
+
+## 4.0.0-beta.8 — 2026-07-15
+
+### Added — project-scan bootstrap (slice 2026-07-15)
+
+- **`peaks project context` writes `.peaks/project-scan/project-scan.md`**
+  alongside `.peaks/PROJECT.md`. The project-scan.md file is the new
+  canonical home for project archetype + libraryVersions + tech stack
+  (project-scoped, git-tracked, survives session rotation). The old
+  `.peaks/_runtime/<sessionId>/rd/project-scan.md` path is fully
+  migrated across SKILL.md / references / peaks-rd SKILL.md /
+  peaks-ui SKILL.md (14 occurrences in 8 files).
+- **`peaks workspace init` triggers `bootstrapProjectScan`** on its
+  main path. Idempotent: existing `schemaVersion: 1` files are
+  preserved unless `--force`. New CLI flags:
+  - `--no-project-scan-bootstrap` — opt out of the bootstrap call.
+  - `--force-project-scan-templates` — overwrite the 4 bundled
+    audit/business templates (default: skip if present).
+- **5-template boot** — every `peaks workspace init` materialises
+  `.peaks/project-scan/{project-scan.md, business-knowledge.md,
+  security-template.md, perf-template.md, audit-output-schema.md}`.
+  The 4 audit/business templates are bundled at
+  `src/services/workspace/templates/project-scan/*.md` and copied
+  verbatim by `scripts/copy-templates.mjs` (post-`tsc` step; tsc
+  does not emit `.md` files).
+- **`scripts/copy-templates.mjs`** — mirrors bundled `.md` templates
+  into `dist/`. Wired into `package.json#scripts.build` after `tsc`.
+  Required to keep `import.meta.url` template reads working for
+  downstream `npm install peaks-loop` consumers.
+- **`tests/unit/workspace/templates/template-integrity.test.ts`**
+  (5 cases) — byte-equality guard between bundled and canonical
+  templates; fails CI on drift.
+- **`tests/unit/services/prd/project-scan-bootstrap-service.test.ts`**
+  (13 cases) — 0-1 path, existing-project path, monorepo variants,
+  idempotency, force overrides, dual-write.
+- **`tests/unit/workspace/init-hooks-project-scan.test.ts`**
+  (4 cases) — `peaks workspace init` integration with all 3 CLI flag
+  combinations.
+
+### Fixed — monorepo 0-1 misjudgement + scanArchetype fall-through
+
+Real-world smoke against `C:\Users\smallMark\Desktop\peaksclaw\ice-cola`
+(pnpm-workspace monorepo with `packages/{server,client,admin,hermes-agent}`)
+exposed two follow-on bugs that the unit suite missed:
+
+- **0-1 detection now considers monorepo sentinels.**
+  `bootstrapProjectScan.isZeroToOneProject` previously only checked
+  `<root>/src/`, silently mis-classifying pnpm-workspace projects as
+  0-1 (the source lives at `packages/<pkg>/src/`). New sentinels:
+  `pnpm-workspace.yaml` / `turbo.json` / `nx.json` + the
+  `package.json#workspaces` field. Source-root candidates extended
+  to `packages/` / `apps/` / `libs/` / `services/` / `workspaces/`.
+- **`scanArchetype` hoists `hasMonorepoConfig` above the `hasBackend`
+  check.** A monorepo with a backend sub-package (`packages/server`)
+  used to fall through to `legacy-fullstack` because `hasBackend`
+  included `backendDirsPresent.length > 0`. New behaviour: any monorepo
+  resolves to `frontend-monorepo` (no backend pkg) or
+  `fullstack-monorepo` (with backend pkg).
+- **`ProjectArchetype` union gains `'fullstack-monorepo'`.**
+  Previously the union was 5 values; now 6. `ArchetypeService` and
+  `peaks-code` consumers updated accordingly.
+
+### Changed
+
+- `src/services/memory/project-context-service.ts` — `generateProjectContext`
+  is now `async` and calls `bootstrapProjectScan` after writing
+  `PROJECT.md`. The companion `peaks skill presence:clear` was missing
+  an `await`; fixed in `src/cli/commands/core/skill-command.ts`.
+- `src/cli/commands/workspace/init-command.ts` — adds
+  `--no-project-scan-bootstrap` and `--force-project-scan-templates`;
+  surfaces `projectScan` envelope field on the init response.
+
+### Verification
+
+- vitest: 22/22 pass across 3 slice-owned test files.
+- tsc: 0 errors in slice-owned `src/` files.
+- CLI: `./bin/peaks.js workspace init --project <tmp>` writes 5 files
+  in `templatesBooted:5, durationMs:23`; `--no-project-scan-bootstrap`
+  correctly skips; `--force-project-scan-templates` correctly
+  overrides user sediment.
+- Ice-cola: `peaks project context --project .` produces
+  `archetype: fullstack-monorepo` with 18 real library versions
+  (Docusaurus 3.9.2 / Camofox / @easyops-cn / etc.).
+
+### Notes
+
+- pre-existing `FILE_SIZE_VIOLATION` on
+  `src/services/standards/project-standards-service.ts` (837 lines,
+  in-flight `.claude/rules/` → `.peaks/standards/` migration) is out
+  of slice scope.
+
+## 4.0.0-beta.7 — 2026-07-09
+
+### Added — zcode adapter (9th IDE) + runtime model detection
+
+- **`peaks ide model --current` CLI** — `src/cli/commands/ide-commands.ts` (80 lines) + register in `src/cli/program.ts`. Outputs `{ modelId, detected, registeredAdapters }`. Available as a top-level primitive (not adapter-specific).
+- **`detectCurrentIdeModel()` service** — `src/services/ide/current-model-detector.ts` (43 lines). Walks the adapter registry in insertion order, calls each adapter's optional `detectCurrentModel()` with try/catch isolation — one bad adapter cannot poison the chain.
+- **zcode-adapter `detectCurrentModel()` method** — `src/services/ide/adapters/zcode-adapter.ts`. Reads `~/.zcode/v2/config.json`, resolves active provider via 4-tier priority chain (env override → non-`builtin:` provider → first enabled → first provider), returns the model's id. Live test: `peaks ide model --current` against a real z-code installation → `"modelId": "M3"`.
+- **`IdeAdapter` interface optional field** — `readonly detectCurrentModel?: () => Promise<string | undefined>` in `src/services/ide/ide-types.ts`. Back-compat: 8 existing adapters without the field are unaffected.
+- **`getStrongestModelIdAsync()` async variant** — `src/services/config/model-routing.ts`. Sync `getStrongestModelId()` is unchanged (rd-service compat); new async variant lets future async callers fall back to runtime probe instead of `'claude-opus-4-7'`.
+
+### Changed — install no longer writes hardcoded default model
+
+- **`scripts/install-skills.mjs`** — removed `model: 'sonnet'` default and `providers.minimax.model` default from `createConfigDefaults()`. After install, `~/.peaks/config.json` no longer contains a Claude model recommendation; users set their own via `peaks config set model <id>` (or leave unset and the back-compat fallback applies).
+- **`STRONGEST_MODEL_ID` constant removed** — replaced by `getStrongestModelId(config?)` function in `src/services/config/model-routing.ts`. Resolution order: `config.model` → `PEAKS_STRONGEST_MODEL_DEFAULT` env var → back-compat `'claude-opus-4-7'`.
+- **`workflow-router-service.ts` / `minimax-worker-service.ts` / `rd-service.ts`** — threaded dynamic modelId through 16 call sites in `workflow-router-service.ts`, replaced literal `'claude-opus-4-7'` in `minimax-worker-service.ts` (×2) and `rd-service.ts`.
+
+### Added — zcode IDE (peaks-loop's 9th)
+
+- **`IdeId` union type** — `'zcode'` added in `src/services/ide/ide-types.ts`.
+- **`ZCODE_ADAPTER`** — new `src/services/ide/adapters/zcode-adapter.ts` (98 lines). Template: `claude-code-adapter.ts` with `compact.compactCommand` / `hookEvent` / `toolMatcher` / `envVar` degraded (z-code is a desktop application, no CLI binary; UNVERIFIED sentinel strings used where the interface requires non-optional types).
+- **`IDE_DETECTION_DIRS` + `IDE_SKILL_INSTALL_PROFILES`** — `scripts/install-skills.mjs`. New `{ id: 'zcode', dir: '.zcode' }` detection entry + zcode profile (skillsDir `~/.zcode/skills`, env override `PEAKS_ZCODE_SKILLS_DIR`).
+- **`ide-registry.ts`** — registers `ZCODE_ADAPTER` alongside the existing 8 adapters (claude-code / trae / trae-cn / codex / cursor / qoder / tongyi-lingma / hermes / openclaw).
+- **`tests/unit/ide/zcode-adapter.test.ts`** — 10 cases (T-1 id, T-2 dirName, T-3 standards rootFile = `CLAUDE.md`, T-4 rulesDir = `.claude/rules`, T-5 skillsDir, T-6 compactCommand undefined, T-7 hookEvent is string, T-8 type test, T-9 IDE_DETECTION_DIRS contains, T-10 IDE_SKILL_INSTALL_PROFILES contains).
+- **`tests/unit/ide/zcode-adapter-detect-model.test.ts`** — 14 cases (pure resolver + env-fixture file IO + path helper).
+- **`tests/unit/cli/ide-commands.test.ts`** — 5 cases (CLI + top-level command registration).
+- **Fixture sync** — `tests/unit/install-skills-script.test.ts` + `tests/unit/cli-program.workflow.test.ts` (Strategy A: env-var override, no hardcoded literals) + `tests/unit/ide/ide-registry.test.ts` (whitelist 6 → 7 adapters).
+
+### Documentation — SKILL.md sync with 4.0.0-beta.6 CLI reality
+
+- **`skills/peaks-code/SKILL.md`** — added inline `> CLI reality check` blocks at §Step 0.8 (D-001 `peaks code detect-job` is `peaks job init`), §Step 0.8 third paragraph (D-003 `JOB_SHAPE_NOT_DECIDED` is now passive `done: 0` warning, not exception), §Step 2.5 (D-002 `peaks session title` takes positional sid, not `--session-id` flag), §Step 11c + 11d (D-010 `<!-- peaks-memory:start -->` block requires YAML frontmatter `title:` + `kind:` + `---` + closing `<!-- peaks-memory:end -->`, otherwise `extractedCount: 0` silently). New **§CLI Drift Index** section after Boundaries provides single landing page for D-001/002/003/010 with inline-anchor table.
+- **`skills/peaks-slice-decompose/SKILL.md`** — new **§SC template hard rules (sediment 2026-07-09)** section with 3 rules: Rule 1 ESM/CJS pre-flight before specifying export syntax (D-009c), Rule 2 whitelist fixture sync is a primary task not collateral (D-009a), Rule 3 adapter-interface optional fields > required (D-009b reciprocal).
+
+### Verification (4.0.0-beta.7)
+
+- `pnpm tsc -p tsconfig.json --noEmit` — 0 errors
+- `pnpm vitest run tests/unit/ide/ tests/unit/cli/ide-commands.test.ts` — 15 files / 211 tests / 0 failed
+- `peaks workflow verify-pipeline --rid 003-add-zcode-adapter` — `ok=true, complete=true`, 9/9 gate PASS
+- `node bin/peaks.js ide model --current` (against real z-code installation) — `modelId: "M3"`, `detected: true`
+
+### Sediment (Step 11) — 7 lessons in `.peaks/memory/`
+
+| File | Topic |
+|---|---|
+| `z-code-peaks-loop-9-ide-adapter-vendor-neutrality-adapter.md` | z-code 9th IDE + vendor-neutrality adapter pattern |
+| `peaks-loop-install-model-getstrongestmodelid-fallback.md` | install no longer hardcodes default model |
+| `desktop-application-ide-adapter-z-code-cli.md` | desktop-app IDE adapter field-degradation decision |
+| `peaks-code-runbook-4-0-0-beta-6-skill-md-cli-d-001-d-002-d-003-d-010.md` | 4 SKILL.md / CLI drift points |
+| `2026-07-09-zcode-adapter-overview.md` | RID 003 overall summary (project record) |
+| `peaks-ide-runtime-detect-zcode-only.md` | `peaks ide model --current` z-code 4-tier priority chain |
+| `ide-adapter-detectcurrentmodel-optional-interface-pattern.md` | optional interface field extension pattern |
+
+### Known follow-ups (S3-cleanup backlog)
+
+- **Slice C extension** — other 8 IDEs' `detectCurrentModel()` (claude-code / trae / cursor / codex / qoder / tongyi-lingma / hermes / openclaw). Currently only z-code has a working detector; all 8 still fall through to the back-compat `'claude-opus-4-7'`.
+- **zai / GLM provider entry** — `D-008`: peaks-loop has no built-in provider entry for z-code's default `builtin:zai` (GLM) or `builtin:bigmodel` providers. Users on those providers must add custom `provider.<uuid>` entries to their `~/.peaks/config.json` themselves.
+- **zcode-adapter UNVERIFIED sentinel values** — `hookEvent: 'PreToolUse'`, `toolMatcher: 'Bash'`, `envVar: 'ZCODE_PROJECT_DIR'` are placeholder guesses based on Anthropic-compatible protocol. Real z-code desktop application dogfood will reveal the actual values; replace once known.
+- **`IdeAdapter` interface optional fields** — `D-009b` proposes converting `hookEvent` / `toolMatcher` / `envVar` to optional types, or adding an `unverified: boolean` flag, so future adapters with unknown protocols don't need to fabricate sentinel values.
+
+### Breaking changes
+
+None. All 8 pre-existing adapters continue to load without changes. CLI subcommands `peaks ide *` and `peaks ide model --current` are additive.
+
+## 4.0.0-beta.5 — 2026-07-08
+
+### Added — peaks-solo dispatcher (分诊员)
+
+- **`peaks-solo` skill** — `skills/peaks-solo/SKILL.md` + 3 references (triage / fallback / sediment). Natural-language front door for the Peaks-Loop skill family. Use when the user describes a task in NL and does not know which peaks-* skill fits. 0 breaking change: 3.x / 4.x `/peaks-code` / `/peaks-content` / `/peaks-doctor` etc. continue to work.
+- **`peaks skill search` CLI** — `src/services/skill/skill-search-service.ts` + `src/cli/commands/skill-search-commands.ts`. Query / tag / domain filters; substring match; structured JSON output. Used by `peaks-solo` to find the right leaf skill. Available as a top-level primitive (not dispatcher-specific).
+- **Sub-skills unchanged** — `peaks-code / peaks-content / peaks-doctor / peaks-issue-fix-orchestrator / peaks-sop / etc.` are NOT modified. peaks-solo sits alongside, not on top.
+
+### Verification (4.0.0-beta.5)
+
+- `peaks skill list` shows `peaks-solo` first
+- `peaks skill search --query "code"` returns `peaks-code` with matchScore > 0
+- `peaks skill search --query "xxxxxxxxxxxxx"` returns `[]` (no error)
+- `pnpm vitest run` — full regression green; `peaks-code / peaks-content / peaks-doctor` tests unchanged
+- `peaks code --help` / `peaks content --help` / `peaks doctor --help` — exit 0, behavior unchanged
+
+## 4.0.0-beta.4 — 2026-07-08
+
+Loop Engineering crystallization is now the product surface. The 4.0.0-beta.3 line shipped the framework; this release ships the post-run crystallization engine, the Darwin-style ratchet, the bundle share / desktop extension surface, and the karpathy-engineered red-line set that locks them together. Every durable-change entry is gated on a real, completed run; every evolution round is gated on an independent-context evaluation.
+
+### Added — Loop Engineering Asset layer (§4.1, AC-1/AC-2/AC-3)
+
+- **`loop_release` table** — `src/services/skillhub/migrations/002-loop-release.sql`. Schema version pinned to `peaks.loop/1` via CHECK constraint. Indexes on `lifecycle_status` and `scenario`. Non-breaking: 4.x `bee_release` rows continue to read.
+- **`loop_bee_relation` table** — `src/services/skillhub/migrations/003-loop-bee-relation.sql`. Schema version `peaks.loop-bee-relation/1`. Roles: `main` / `supporting` / `candidate` / `retired`. Partial UNIQUE index `WHERE role='main'` enforces at most one main bee per loop at the storage layer (defense in depth; the service layer also enforces it).
+- **Schema for the four-loop + dual-asset model** — Zod schemas in `src/services/loop/loop-release-types.ts` and `src/services/loop/loop-bee-relation-types.ts` mirror the §4.1 / §4.6 row shapes and refuse to parse rows whose `schema_version` is not the fixed literal.
+
+### Added — Post-run crystallization flow (§5, AC-4 / AC-5 / AC-6 / AC-7)
+
+- **`crystallization_event` table** — `src/services/skillhub/migrations/006-crystallization-event.sql`. Schema version `peaks.crystallization/1`. Carries the 4-section evidence brief inline; FKs to optional created/updated loop_release / bee_release.
+- **`CrystallizationService`** — `src/services/crystallization/crystallization-service.ts`. Pre-run gate: `task_status='completed'` AND `gates_passed=true` AND `evidence_collected=true` (Zod literal enforcement + service-layer re-assertion). Atomic single-transaction write of loop_release + bee_release header + bee_manifest + loop_bee_relation(main) + crystallization_event(brief inline).
+- **`peaks asset crystallize`** CLI — `src/cli/commands/asset-commands.ts`. Required options: `--brief-what-happened`, `--brief-why-it-matters`, `--brief-what-learned`, `--brief-what-action`. A partial brief is rejected with `MISSING_BRIEF_SECTION` and exit 1 (RL-7).
+- **`peaks asset dispose` / `peaks asset status`** — cross-asset dispose (trace_only / retain / destroy) and lifecycle status dashboard.
+
+### Added — Darwin-style ratchet (§6, AC-8 / AC-9 / AC-10 / AC-11)
+
+- **`evolution_evaluation` table** — `src/services/skillhub/migrations/005-evolution-evaluation.sql`. Schema version `peaks.evolution/1`. Carries `target_kind`, `target_release_id`, `optimization_dimensions_json` (length-1 enforced at service layer), `target_count=1` (CHECK), `author_id`, `evaluator_id`, `skeptic_id` (three independent agents), `verdict` ∈ {`keep`, `revert`, `needs-user-decision`}, `user_confirmation_pointer`, `brief_pointer`, and the four-section brief projection.
+- **`EvolutionService`** — `src/services/evolution/evolution-service.ts`. Hard rules: single object (AC-8) + single optimization dimension (AC-8) + author ≠ evaluator ≠ skeptic (AC-10/AC-12/AC-14) + score_delta >= score_delta_min for `keep` (AC-11, default 1.0) + user_confirmation_pointer required for `keep` (AC-15).
+- **`peaks evolution propose / evaluate / revert / mark-keep / status`** CLI — `src/cli/commands/evolution-commands.ts`. Each error path emits a stable wire-format code (`EVOLUTION_MULTI_OBJECT` / `EVOLUTION_MULTI_DIMENSION` / `EVOLUTION_SELF_SCORE` / `EVOLUTION_DELTA_BELOW_THRESHOLD` / `EVOLUTION_MISSING_USER_CONFIRMATION`).
+- **Independent-evaluator runner** — `src/services/evolution/independent-evaluator-runner.ts`. Frozen `EvaluationPackage` containing only `target_kind / target_release_id / optimization_dimension / before_snapshot / after_snapshot / diff / rubric / red_lines / source_traces` — no author session, no author reasoning, no recommendation framing (AC-12 / AC-13).
+- **Regression-skeptic runner** — `src/services/evolution/regression-skeptic-runner.ts`. A SEPARATE agent that emits `driftRisks / overfitRisks / safetyRegressionRisks` plus an optional `blocker` (which forces `verdict='revert'`).
+
+### Added — Karpathy-engineered red-line set (AC-21 / AC-22 / AC-23)
+
+- **`.peaks/standards/loop-engineering-guidelines.md`** — the single source of truth for the 10 red lines RL-0..RL-9, each in the four-section karpathy form (Failure modes / Rewrite / Self-check / Out-of-scope). Co-equal karpathy × darwin layers (RL-0).
+- **`peaks standards lint --category loop-engineering`** — `src/services/standards/loop-engineering-lint.ts`. Parses the guideline file and asserts every red line has all four sections, fails closed otherwise.
+- **`peaks skill lint --category loop-engineering-readiness --path <skill-dir>`** — `src/services/standards/loop-engineering-readiness-lint.ts` + `src/cli/commands/skill-loop-engineering-readiness-commands.ts`. Asserts a peaks-* SKILL.md (a) references `.peaks/standards/loop-engineering-guidelines.md`, (b) does not introduce a CLI verb the user is meant to type, (c) does not introduce a JSON / manifest hand-authoring surface. Alias verb: `peaks skill ready --category loop-engineering-readiness --path <skill-dir>`.
+- **Unit guard** — `tests/unit/standards/loop-engineering-guidelines.test.ts` enforces the four-section shape for every red line in the file.
+- **peaks-code domain boundary (RL-8)** — `skills/peaks-code/SKILL.md` self-declares as the code-domain long-task loop engineering orchestrator and is NOT a general-purpose orchestrator. Cross-domain peaks-* skills (peaks-content, peaks-issue-fix-orchestrator) import the shared guideline file and pass the readiness lint.
+
+### Added — Bundle share + desktop extension surface (§7A, AC-24 / AC-25 / AC-26)
+
+- **Share / desktop fields** — `src/services/skillhub/migrations/004-loop-bee-extension.sql`. Adds `shareable / share_excluded_paths / desktop_visible / export_bundle_format` to `loop_release`, and `shareable / desktop_visible` to `bee_release`. `export_bundle_format` is pinned to `peaks.bundle/1` via CHECK constraint. Non-breaking: every new column has a DEFAULT.
+- **Bundle writer** — `src/services/share/bundle-writer.ts`. Writes a `peaks.bundle/1` tar.gz with `manifest.json`, `relations.json`, `evidence_briefs/*.json`, content-addressed `blobs/<sha256>`, and `EVALUATION_REQUIRED.md`. Hard block: `shareable=false` throws `BundleNotShareableError(SHARE_BUNDLE_ERROR_CODES.NOT_SHAREABLE)` BEFORE any tarball work, for both loop and bee kinds.
+- **Bundle reader** — `src/services/share/bundle-reader.ts`. Hard blocks: `format_version_major != 1` is rejected (AC-25); `schema_versions` mismatch is rejected; the imported release ALWAYS lands as `candidate` (AC-25 / §10 RL-9). The bundle carries an `EVALUATION_REQUIRED.md` marker telling the receiver to run an independent evaluation before any durable change.
+- **`peaks loop export` / `peaks loop import`** — `src/cli/commands/loop-commands.ts`. Existing 14.x `peaks loop *` subcommands are preserved; M7 only ADDS export / import.
+- **`peaks bee export` / `peaks bee import`** — `src/cli/commands/share-commands.ts`. Same hard-block semantics.
+- **Receiver-side integration test** — `tests/integration/share-bundle-roundtrip.test.ts` asserts the round-trip lands as `candidate` (AC-25), that `shareable=false` blocks the export at the CLI layer, and that without an `evolution_evaluation` row the receiver has no path to `stable` (AC-26).
+
+### Added — SkillHub expansion (5 new tables / non-breaking)
+
+The 6+ relation table layout is preserved. 5 new tables added under `.peaks/_runtime/<sessionId>/`:
+- `loop_release` (002)
+- `loop_bee_relation` (003)
+- loop↔bee share / desktop extension columns (004)
+- `evolution_evaluation` (005)
+- `crystallization_event` (006)
+
+Migration is non-breaking: every new column has a DEFAULT and the new tables sit alongside the existing 4.x schema.
+
+### Added — New CLI verbs (spec §7.4)
+
+- `peaks loop init / list / show / search / recent / export / import`
+- `peaks asset crystallize / dispose / status`
+- `peaks evolution propose / evaluate / revert / mark-keep / status`
+- `peaks skill lint --category loop-engineering-readiness --path <skill-dir>`
+- `peaks skill ready --category loop-engineering-readiness --path <skill-dir>` (alias)
+
+### Changed
+
+- **`peaks workflow *`** is reframed as the execution trace surface (per spec §7.6 demotion); it remains functional. The user-facing verb is "replay this run", not "create a new asset".
+- **peaks-maker skill** is narratively repositioned to "Loop crystallizer + Bee creator + Evolution gatekeeper" (per spec §7.5); only the SKILL.md and memory references are re-narrated, the `id` is preserved.
+
+### Verification (4.0.0-beta.4)
+
+- `tsc --noEmit` zero errors
+- `peaks standards lint --category loop-engineering` parses all 10 red lines + 4 sections each (lint passes)
+- `pnpm vitest run tests/unit/{standards,crystallization,evolution,share}/` — 13 files / 154 unit tests pass
+- `pnpm vitest run tests/integration/{share-bundle-roundtrip,dogfood-loop-engineering-crystallization,asset-crystallize-cli,evolution-cli,skill-loop-engineering-readiness-cli}.test.ts` — 5 files / 26 integration tests pass
+- M8 dogfood (`tests/integration/dogfood-loop-engineering-crystallization.test.ts`) crystallizes the M0..M7 work into a real loop + bee + relation + event in 644ms (defense in depth: the release crystallizes itself)
+
+### Migration
+
+- 4.x `bee_release` rows continue to read. No data migration required.
+- The 5 new tables are added via non-destructive `CREATE TABLE IF NOT EXISTS` migrations.
+- The `bundle-writer / bundle-reader` pair is additive; existing `peaks skill sediment export / import` remains as an alias for one release cycle, then deprecates.
+- The peaks-code SKILL.md is unchanged from 4.0.0-beta.3 — no `peaks-solo` migration needed for users already on beta.2 or later.
+
+## 4.0.0-beta.2 — 2026-07-07
+
+### Renamed
+- **`peaks-solo` → `peaks-code`** — the long-running code-domain orchestrator skill (PRD/RD/QA/UI/SC/TXT pipeline) has been renamed to communicate its scope more accurately. The on-disk directory `~/.peaks/skills/.system/bees/peaks-code/` is preserved as a stable install location; only the manifest `id` and `displayName` change (`peaks-code` / `Peaks Code`). All four sibling skills (`peaks-solo-resume` → `peaks-resume`, `peaks-solo-status` → `peaks-status`, `peaks-solo-test` → `peaks-test`) are now top-level primitives rather than child skills of `peaks-solo`. Migration: `peaks session migrate-skill-name --from peaks-solo --to peaks-code --apply`.
+
+### Changed (Breaking)
+- **`peaks skill presence:set peaks-solo` no longer recognized** — the canonical skill name is now `peaks-code`. Existing `.peaks/_runtime/*/active-skill.json` files carrying `skill: "peaks-solo"` are migrated by `peaks session migrate-skill-name`. Manual override: edit the file and replace `"skill": "peaks-solo"` with `"skill": "peaks-code"`.
+- **CLI surface unchanged** — `peaks code`, `peaks code --fast`, and the entire `peaks-code` skill runbook continue to function; only the skill-identification field (`id`, `displayName`, presence `skill` value) changes. The runbook name `peaks code` is the user-facing verb and remains as-is.
+
+### Fixed (post-rename polish)
+- **Rule name normalization** — `Code Code-Change Red Line` → `Code Commit Ban Red Line` (8 occurrences across `src/services/audit/enforcers/code-ban.ts`, `src/cli/commands/hook-handle.ts`, and the audit tests). Aligns with `red-line-catalog.ts` entry `rl-code-ban-001` / `Code Commit Ban`.
+- **Wire-format error code rename** — `SOLO_MODE_REQUIRES_SOLO_WORKFLOW` → `CODE_MODE_REQUIRES_CODE_WORKFLOW` and `UNSUPPORTED_SOLO_MODE` → `UNSUPPORTED_CODE_MODE` in `src/cli/commands/workflow-commands.ts` (4 test sites updated).
+- **Test local consts** — `soloResult/soloOutput` → `codeResult/codeOutput` in `tests/unit/cli-program.workflow.test.ts`; `SOLO_PATH/soloAbsPath` → `CODE_PATH/codeAbsPath` in `tests/unit/code/skills-subagent-scope-dir.test.ts`; `SOLO_REF/SOLO_FANOUT_REF` → `CODE_REF/CODE_FANOUT_REF` in `tests/unit/dispatch/dispatch-fanout-mandatory.test.ts`.
+- **CHANGELOG self-contradiction** — the rename entry previously read `peaks-code → peaks-code` (collapsed by the global `s/Solo/Code/g` substitution). Now correctly reads `peaks-solo → peaks-code`.
+- **`deriveRuleName` test expectation** — `tests/unit/services/audit/classifier.test.ts` expected 9 words but the function correctly truncates to 8. Updated to match the new word split introduced by `Code Commit Ban` (2 words) replacing `Code Code-Change` (1 hyphenated word).
+- **Output style mode label** — `.claude/output-styles/peaks-skill-swarm.md` now uses `Code` instead of `Solo` in the mode badge and prose (3 occurrences).
+- **Project history row** — `.peaks/PROJECT.md` session-history row `001-solo-memory-write-broken` → `001-code-memory-write-broken`.
+- **Video demo copy** — `examples/video-demo/src/copy.ts` `skillWas: 'peaks-solo'` → `skillWas: 'peaks-code'` (legacy label, both locales).
+
+### Added
+- **Manifest id field** — `.peaks/skills/.system/bees/peaks-code/manifest.json` now carries `{ "id": "peaks-code", "displayName": "Peaks Code" }`. The `migrate-skill-name` service explicitly skips this file (test: `session-migrate-skill-name.test.ts:69` "跳过 .peaks/skills/.system/bees/peaks-code/manifest.json") so the canonical id is only ever edited by hand or by an explicit `peaks skill sediment refine-bee` flow, never by bulk migration.
+- **Spec §4.1.1 rewrite** — `docs/superpowers/specs/2026-07-04-peaks-maker-dynamic-skill-sediment-design.md` §4.1.1 has been rewritten from "peaks-code as preserved alias" to "peaks-code as the system-stable code-domain bee (renamed from peaks-code)". The plan's spec-coverage table now references the renamed heading.
+- **Migration helper** — `peaks session migrate-skill-name --from <old> --to <new> [--apply]` is the supported path for renaming skill references in `.peaks/_runtime/`. Idempotent; skip-list includes `.peaks/memory/**` and `.peaks/skills/.system/bees/peaks-code/manifest.json`.
+
+### Verification
+- `tsc --noEmit` zero errors
+- Full `vitest run` exit 0 (all 30 touched test files / 265 directly-tested assertions pass)
+- `silent-warning-detector` OK on 481 files
+- All rename+polish changes committed as `b1dbc51 chore(rename): complete peaks-solo → peaks-code sweep + polish`
+
 ## 3.1.2 — 2026-07-04
 
 ### Added — Mechanical Job-mode gates (Step 0.8 enforcement)
 
-The v3.1.1 recorder-only design was bypassed twice under load (2026-07-03 35-slice incident, 2026-07-04 3018-file incident): the LLM skipped `peaks solo detect-job` and ran fake-completion narratives instead of Job mode. The v3.1.2 patch converts Step 0.8 from "LLM should call" to "LLM cannot proceed without calling" through four mechanical gates — no LLM interpretation required.
+The v3.1.1 recorder-only design was bypassed twice under load (2026-07-03 35-slice incident, 2026-07-04 3018-file incident): the LLM skipped `peaks code detect-job` and ran fake-completion narratives instead of Job mode. The v3.1.2 patch converts Step 0.8 from "LLM should call" to "LLM cannot proceed without calling" through four mechanical gates — no LLM interpretation required.
 
-- **PreToolUse hook** — `peaks workspace init` now installs a `Bash` matcher in `.claude/settings.local.json` (TEMPLATE_VERSION bumped 1.2.0 → 1.3.0) that runs `peaks solo gate-step-08 --project .` before every Bash call. Exit 0 = allow, exit 2 = block. Existing `Write|Edit|MultiEdit` matcher preserved. Re-running `peaks workspace init` is idempotent (`templateContentMatches`).
-- **`peaks solo gate-step-08`** — new service `src/services/solo/step-08-gate.ts` (~120 LOC) + CLI. Reads `.peaks/_runtime/<sessionId>/job-shape.json`. When `decision.isJob=true` AND `job/<jid>/progress.json` exists, prints `Next: slice #<done+1> of <total> (<currentSlice>)` so the LLM cannot wake up cold. When the file is missing AND the user's prompt (from `--prompt` / `last-prompt.txt`) matches the fail-closed backup regex `/(until|全部|until all done|disavow cost|不用考虑费用|all of them)/i`, exits 2 with `BLOCKED:` on stderr.
-- **Size-fear ban** — `peaks solo emit-handoff`. New service `src/services/solo/emit-handoff.ts` (~70 LOC) + CLI. Under Job mode (`isJob=true`), refuses to emit a final handoff while `remaining > 0`. Codes: `JOB_NOT_INITIALIZED` (no state.json), `JOB_REMAINING_BLOCKED` (remaining > 0), allow when `remaining === 0`, allow under `--force-under-job` override.
-- **Forced auto-compact** — extended `peaks solo context-now` with `--enforce-job-mode` (auto-enabled when `job-shape.json` says `isJob=true`). At `ratio >= 0.85` returns `action: 'auto-compact-now'` (MANDATORY, not advisory); at `ratio >= 0.95` returns `action: 'red-line'`. SKILL.md Step N+2 prose updated: "in Job mode, ≥ 0.85 is MANDATORY auto-compact."
+- **PreToolUse hook** — `peaks workspace init` now installs a `Bash` matcher in `.claude/settings.local.json` (TEMPLATE_VERSION bumped 1.2.0 → 1.3.0) that runs `peaks code gate-step-08 --project .` before every Bash call. Exit 0 = allow, exit 2 = block. Existing `Write|Edit|MultiEdit` matcher preserved. Re-running `peaks workspace init` is idempotent (`templateContentMatches`).
+- **`peaks code gate-step-08`** — new service `src/services/code/step-08-gate.ts` (~120 LOC) + CLI. Reads `.peaks/_runtime/<sessionId>/job-shape.json`. When `decision.isJob=true` AND `job/<jid>/progress.json` exists, prints `Next: slice #<done+1> of <total> (<currentSlice>)` so the LLM cannot wake up cold. When the file is missing AND the user's prompt (from `--prompt` / `last-prompt.txt`) matches the fail-closed backup regex `/(until|全部|until all done|disavow cost|不用考虑费用|all of them)/i`, exits 2 with `BLOCKED:` on stderr.
+- **Size-fear ban** — `peaks code emit-handoff`. New service `src/services/code/emit-handoff.ts` (~70 LOC) + CLI. Under Job mode (`isJob=true`), refuses to emit a final handoff while `remaining > 0`. Codes: `JOB_NOT_INITIALIZED` (no state.json), `JOB_REMAINING_BLOCKED` (remaining > 0), allow when `remaining === 0`, allow under `--force-under-job` override.
+- **Forced auto-compact** — extended `peaks code context-now` with `--enforce-job-mode` (auto-enabled when `job-shape.json` says `isJob=true`). At `ratio >= 0.85` returns `action: 'auto-compact-now'` (MANDATORY, not advisory); at `ratio >= 0.95` returns `action: 'red-line'`. SKILL.md Step N+2 prose updated: "in Job mode, ≥ 0.85 is MANDATORY auto-compact."
 - **On-disk slice progress** — `peaks job checkpoint --state done` now writes `.peaks/_runtime/<sessionId>/job/<jid>/progress.json` after updating state.json. New `peaks job progress --job-id <jid>` reader. `gate-step-08` reads the same file in its allow-job path so the LLM gets resume context BEFORE any Bash call. SKILL.md Step 0.7 updated: "if progress.json exists, read it FIRST and surface `Next: slice #N of M (<currentSlice>)`."
 
 ### Tests (v3.1.2)
 
-- `tests/unit/solo/gate-step-08.test.ts` — 9 unit tests covering the 4 paths + Next: slice context injection + backup-regex sanity + prompt-source fallback.
-- `tests/unit/solo/emit-handoff.test.ts` — 10 unit tests covering the 4 paths + `--force-under-job` override + `JOB_NOT_INITIALIZED` + skipped-slices edge case + `--job-id` override.
-- `tests/integration/solo-gate-step-08-hook.test.ts` — 5 integration tests spawning `node bin/peaks.js solo gate-step-08` as a real child process (the hook protocol itself), asserting exit 0 / exit 2 + stderr BLOCKED line + JSON envelope shape.
-- Extended `tests/unit/solo/solo-step-08-block-guard.test.ts` to assert SKILL.md + runbook.md reference `peaks solo gate-step-08`, `peaks solo emit-handoff`, `peaks job progress`, and the Job-mode MANDATORY auto-compact prose.
-- Extended `tests/unit/workspace/workspace-init-claude-hooks.test.ts` case-A to assert both PreToolUse matchers (`Write|Edit|MultiEdit` AND `Bash`) are emitted with the new Bash command invoking `peaks solo gate-step-08`.
+- `tests/unit/code/gate-step-08.test.ts` — 9 unit tests covering the 4 paths + Next: slice context injection + backup-regex sanity + prompt-source fallback.
+- `tests/unit/code/emit-handoff.test.ts` — 10 unit tests covering the 4 paths + `--force-under-job` override + `JOB_NOT_INITIALIZED` + skipped-slices edge case + `--job-id` override.
+- `tests/integration/code-gate-step-08-hook.test.ts` — 5 integration tests spawning `node bin/peaks.js code gate-step-08` as a real child process (the hook protocol itself), asserting exit 0 / exit 2 + stderr BLOCKED line + JSON envelope shape.
+- Extended `tests/unit/code/code-step-08-block-guard.test.ts` to assert SKILL.md + runbook.md reference `peaks code gate-step-08`, `peaks code emit-handoff`, `peaks job progress`, and the Job-mode MANDATORY auto-compact prose.
+- Extended `tests/unit/workspace/workspace-init-claude-hooks.test.ts` case-A to assert both PreToolUse matchers (`Write|Edit|MultiEdit` AND `Bash`) are emitted with the new Bash command invoking `peaks code gate-step-08`.
 
 ### Memory
 
@@ -28,17 +488,17 @@ Two consecutive ship-day incidents motivated this release: `2026-07-03-v3-1-0-jo
 
 ### Added — Step 0.8 detector-as-recorder (LLM-judged, CLI-validated)
 
-`peaks-solo` Step 0.8 is no longer prose-only: the LLM makes the Job-shape judgement, the CLI records it. The CLI is a **recorder and gate**, not a detector — no keyword regex anywhere in the service or CLI (the LLM is the source of truth). Downstream steps refuse to proceed until the decision file exists.
+`peaks-code` Step 0.8 is no longer prose-only: the LLM makes the Job-shape judgement, the CLI records it. The CLI is a **recorder and gate**, not a detector — no keyword regex anywhere in the service or CLI (the LLM is the source of truth). Downstream steps refuse to proceed until the decision file exists.
 
-- New service: `src/services/solo/job-shape-decision.ts` — `readJobShapeDecision` / `writeJobShapeDecision` / `validateJobShapeDecision`. Persists to `.peaks/_runtime/<sessionId>/job-shape.json`; server-side stamps `decidedAt` so the LLM cannot back-date; throws `JOB_SHAPE_NOT_DECIDED` on missing / unreadable / malformed.
+- New service: `src/services/code/job-shape-decision.ts` — `readJobShapeDecision` / `writeJobShapeDecision` / `validateJobShapeDecision`. Persists to `.peaks/_runtime/<sessionId>/job-shape.json`; server-side stamps `decidedAt` so the LLM cannot back-date; throws `JOB_SHAPE_NOT_DECIDED` on missing / unreadable / malformed.
 - New CLI subcommands:
-  - `peaks solo detect-job --is-job <bool> --rationale <text> --suggested-job-id <jid> [--suggested-strategy single|rotating] [--confidence high|medium|low] [--force]` — recorder. Validates `suggestedJobId` against `/^[a-z0-9][a-z0-9-]{2,40}$/`, hashes the user prompt, writes the file. Refuses overwrite without `--force`.
-  - `peaks solo read-job-shape` — downstream gate. Returns the current record or `JOB_SHAPE_NOT_DECIDED`.
-- New tests: `tests/unit/solo/job-shape-decision.test.ts` (validate happy/sad, server-stamp `decidedAt`, force semantics, missing/unreadable/malformed, round-trip, promptHash determinism), `tests/integration/solo-detect-job-command.test.ts` (CLI spawn happy / bad-flag / read-back / fresh / force), `tests/unit/solo/solo-step-08-block-guard.test.ts` (locks the SKILL.md / runbook contract: `BLOCKING on LLM judgement`, `peaks solo detect-job`, `peaks solo read-job-shape`, and runbook ordering before `# After Step 7`).
+  - `peaks code detect-job --is-job <bool> --rationale <text> --suggested-job-id <jid> [--suggested-strategy single|rotating] [--confidence high|medium|low] [--force]` — recorder. Validates `suggestedJobId` against `/^[a-z0-9][a-z0-9-]{2,40}$/`, hashes the user prompt, writes the file. Refuses overwrite without `--force`.
+  - `peaks code read-job-shape` — downstream gate. Returns the current record or `JOB_SHAPE_NOT_DECIDED`.
+- New tests: `tests/unit/code/job-shape-decision.test.ts` (validate happy/sad, server-stamp `decidedAt`, force semantics, missing/unreadable/malformed, round-trip, promptHash determinism), `tests/integration/code-detect-job-command.test.ts` (CLI spawn happy / bad-flag / read-back / fresh / force), `tests/unit/code/code-step-08-block-guard.test.ts` (locks the SKILL.md / runbook contract: `BLOCKING on LLM judgement`, `peaks code detect-job`, `peaks code read-job-shape`, and runbook ordering before `# After Step 7`).
 
 ### Hard red line #10 (v3.1.1)
 
-The LLM MUST NOT skip `peaks solo detect-job` even when the trigger is obvious from context. If the next step's `read-job-shape` throws `JOB_SHAPE_NOT_DECIDED`, Solo MUST record a decision before proceeding. Keyword-based "I already know it's a Job" is not a substitute — the LLM is the only authority for the Job-shape judgement, and the CLI is the only place that records it.
+The LLM MUST NOT skip `peaks code detect-job` even when the trigger is obvious from context. If the next step's `read-job-shape` throws `JOB_SHAPE_NOT_DECIDED`, Code MUST record a decision before proceeding. Keyword-based "I already know it's a Job" is not a substitute — the LLM is the only authority for the Job-shape judgement, and the CLI is the only place that records it.
 
 ### Memory
 
@@ -46,9 +506,9 @@ Incident + rationale: `.peaks/memory/2026-07-03-v3-1-0-job-trigger-miss.md`. Thi
 
 ## 3.1.0 — 2026-07-03
 
-### Fixed — peaks-solo external-references authority declaration
+### Fixed — peaks-code external-references authority declaration
 
-- `skills/peaks-solo/SKILL.md` External-references paragraph now ends with `Peaks-Loop Solo gates and artifacts remain authoritative.` to satisfy `tests/unit/skill-external-invocation.test.ts:63` (`PEAKS_AUTHORITATIVE_PATTERN`). One-line surgical patch; no behaviour change.
+- `skills/peaks-code/SKILL.md` External-references paragraph now ends with `Peaks-Loop Code gates and artifacts remain authoritative.` to satisfy `tests/unit/skill-external-invocation.test.ts:63` (`PEAKS_AUTHORITATIVE_PATTERN`). One-line surgical patch; no behaviour change.
 
 ### Added — Peaks-Loop Job
 
@@ -56,7 +516,7 @@ Incident + rationale: `.peaks/memory/2026-07-03-v3-1-0-job-trigger-miss.md`. Thi
 - New CLI flag `--main-loop-strategy single|rotating` on `job init`, with rotating-mode hard default for ≥3 slices and `rotateEvery=3` cadence.
 - New `--watch` poll mode + statusline event stub (`emitJobEvent`) for ambient progress visibility. Real statusline wire-up deferred to a follow-up slice.
 - Sub-agent wrapper `SubAgentJobWrapper` enforces `--budget-mb 512` default + cleanup gate before slice checkpoint.
-- New peaks-solo SKILL.md Steps 0.8 / 0.81 / 0.85 / 0.86 / 0.87 wrapping the existing single-rid runbook for multi-slice jobs.
+- New peaks-code SKILL.md Steps 0.8 / 0.81 / 0.85 / 0.86 / 0.87 wrapping the existing single-rid runbook for multi-slice jobs.
 - 9 hard-red-line rules embedded in the skill prose (cost re-ask ban, slice-coalesce ban, fake-completion ban, detached-mode ban, cleanup-skip ban, rotate-skip ban, etc).
 
 ### New services
@@ -71,7 +531,7 @@ Incident + rationale: `.peaks/memory/2026-07-03-v3-1-0-job-trigger-miss.md`. Thi
 
 ### Migration
 
-- Existing single-rid flows are unchanged. The Job is opt-in via peaks-solo Step 0.8 detection; users who do not invoke multi-slice requests see no behavior difference.
+- Existing single-rid flows are unchanged. The Job is opt-in via peaks-code Step 0.8 detection; users who do not invoke multi-slice requests see no behavior difference.
 - The `peaks session rotate` and `peaks session cycle-summary` subcommands referenced in the spec are NOT shipped in v1; the rotation path uses constructor-injected stub callbacks. M6.5 follow-up will add the real subcommands.
 
 ### Spec / plan
@@ -90,19 +550,19 @@ The v2.13.0 auto-compact design shipped `shell-exec` as the only pathway, which
 - **New `ide-native` pathway** — `src/services/context/auto-compact-dispatcher.ts:138-155` routes main-session compacts through `installAutoCompactHook`, which writes a PreToolUse hook into `.claude/settings.local.json` (matcher `Bash|Task`, command `peaks session auto-compact-hook`).
 - **New `peaks session auto-compact-hook` CLI** — `src/cli/commands/session-auto-compact-hook-command.ts`. Reads `CLAUDE_CONTEXT_USAGE_PERCENT`; below 0.95 exits 0 silent; at ratio ≥ 0.95 spawns `claude --compact` in-band against the **current** runner (detached + unref, never blocks the runner's tool call). ENOENT-safe: if `claude` is not on PATH (e.g. user runs Claude Code as an MCP), the hook logs a stderr hint and exits 0 silent rather than crashing the runner.
 - **New `installAutoCompactHook` / `removeAutoCompactHook` service** — `src/services/hooks/auto-compact-hook-install.ts`. Idempotent: re-install returns `already-installed` and does NOT re-write the file; remove preserves other PreToolUse entries (the existing fact-forcing bypass hook on every peaks-loop consumer project survives).
-- **Lazy install** — hook is installed on first `peaks solo auto-compact` invocation, NOT on `peaks workspace init`. User has opted in by running the command; no zero-touch surprise.
+- **Lazy install** — hook is installed on first `peaks code auto-compact` invocation, NOT on `peaks workspace init`. User has opted in by running the command; no zero-touch surprise.
 - **Claude Code adapter** — `compactPathway` changed from `shell-exec` to `ide-native` for `target='main'`. Sub-agent shells still get the legacy `shell-exec` pathway.
 
-### Fixed — Solo Step N+2 prose (was 75% / "do NOT auto-execute")
+### Fixed — Code Step N+2 prose (was 75% / "do NOT auto-execute")
 
-- `skills/peaks-solo/SKILL.md` Step N+2 paragraph rewritten: thresholds now 0.85 pre-compact / 0.95 red-line (matches v2.13.0 auto-compact), explicit Karpathy §4 compact-red-line exception cited, `peaks solo context-now --json` is the canonical probe primitive (replaces `peaks context check --prompt-size <bytes>` hand-pass).
+- `skills/peaks-code/SKILL.md` Step N+2 paragraph rewritten: thresholds now 0.85 pre-compact / 0.95 red-line (matches v2.13.0 auto-compact), explicit Karpathy §4 compact-red-line exception cited, `peaks code context-now --json` is the canonical probe primitive (replaces `peaks context check --prompt-size <bytes>` hand-pass).
 - `src/services/context/main-session-monitor.ts` — `@deprecated` JSDoc on `evaluateMainSessionThreshold` cross-links to `evaluateCompactTrigger` (0.85/0.95); legacy 4-tier envelope retained for statusline callers (migration to v2.15.0).
 
 ### Tests — 4 new suites + 3 modified files
 
 - `tests/unit/services/hooks/auto-compact-hook-install.test.ts` (7 cases) — install / idempotent / remove / preservation of unrelated entries.
 - `tests/unit/services/context/auto-compact-dispatcher-ide-native.test.ts` (9 cases) — `dispatchIdeCompact({ target: 'main' })` routes to `ide-native`; round-trip install / remove.
-- `tests/unit/skills/solo-step-n-plus-2-prose.test.ts` (6 cases) — pins the 0.85 / 0.95 / Karpathy §4 contract; asserts `75%` is NOT in the paragraph.
+- `tests/unit/skills/code-step-n-plus-2-prose.test.ts` (6 cases) — pins the 0.85 / 0.95 / Karpathy §4 contract; asserts `75%` is NOT in the paragraph.
 - `tests/unit/cli/session-auto-compact-hook-command.test.ts` (3 cases) — dogfood surfacing: below-threshold silent, missing env-var silent, red-line + ENOENT exits 0 with stderr hint.
 - `tests/unit/services/context/main-session-monitor.test.ts` (+2 AC-4 cases), `tests/unit/context/auto-compact-main-target.test.ts` (pathway assertion updated) — coverage.
 
@@ -196,7 +656,7 @@ The v2.19.0 work shipped as code commits (`f91d71c`, `5186d3d`, `b598612`, `ced8
 
 The Understand Anything (UA) Claude Code plugin and `@colbymchenry/codegraph` are both complementary project-analysis tools. UA provides a high-level knowledge graph (tours, layers, expert prompts); codegraph provides low-level affected-file analysis. Until 3.0.2, peaks-loop consumers had to choose one or the other at the LLM level — there was no service-layer enforcement of either.
 
-This slice adds a single `peaks understand context` subcommand that encapsulates the routing decision in the service layer (per the peaks-loop "skill-first / CLI-auxiliary" architecture — see `peaks-solo` SKILL.md). The envelope carries a `source` tag so downstream consumers can audit which evidence contributed to the context.
+This slice adds a single `peaks understand context` subcommand that encapsulates the routing decision in the service layer (per the peaks-loop "skill-first / CLI-auxiliary" architecture — see `peaks-code` SKILL.md). The envelope carries a `source` tag so downstream consumers can audit which evidence contributed to the context.
 
 **Source routing** (deterministic, decided in `buildUnderstandContext`):
 - `ua-only` — UA knowledge graph present, no codegraph
@@ -217,41 +677,41 @@ The peak-loop skill-external-invocation audit pattern (capability discovery + re
 
 ### Removed — change-id shim (cleanup post v2.17.0 hard-kill, rolled up from v2.19.0)
 
-**PATCH bump from 2.18.3**. Resolves the user-reported "新项目第一次使用 peaks-solo 时初始化完不按流程走、开发效果差、再开 session 后才正常" defect. Two surgical fixes to the first-run Solo gates (3 source files, 1 new test file, 2 test-regression updates; no schema change, no new dependencies, no public API change).
+**PATCH bump from 2.18.3**. Resolves the user-reported "新项目第一次使用 peaks-code 时初始化完不按流程走、开发效果差、再开 session 后才正常" defect. Two surgical fixes to the first-run Code gates (3 source files, 1 new test file, 2 test-regression updates; no schema change, no new dependencies, no public API change).
 
 ### Bugfix — `step-0.55-1x-upgrade` always pauses, even in `full-auto` (P0)
 
-The 1.x → 2.0 upgrade is an irreversible external side effect (rewrites `~/.peaks/config.json` + the on-disk cache schema). Pre-fix, `peaks solo should-pause --step step-0.55-1x-upgrade --mode full-auto` returned `shouldPause: false` because the step was declared in `GATED_STEPS` but missing from the `HARD_PAUSE_STEPS` `Set` that already covered `step-0.5-openspec-opt-in` and `step-0.7-resume-detection`. In `full-auto` mode, the `shouldAutoProceed` branch silently auto-proceeded the upgrade without user consent. The result: 1.x → 2.0 downgrade (or stale 1.x standards/config residue) on the very first Solo run of a fresh project. Post-fix, the step pauses with `gateKind: 'mode-selection-itself'` and the LLM-side caller presents an `AskUserQuestion` to the user.
+The 1.x → 2.0 upgrade is an irreversible external side effect (rewrites `~/.peaks/config.json` + the on-disk cache schema). Pre-fix, `peaks code should-pause --step step-0.55-1x-upgrade --mode full-auto` returned `shouldPause: false` because the step was declared in `GATED_STEPS` but missing from the `HARD_PAUSE_STEPS` `Set` that already covered `step-0.5-openspec-opt-in` and `step-0.7-resume-detection`. In `full-auto` mode, the `shouldAutoProceed` branch silently auto-proceeded the upgrade without user consent. The result: 1.x → 2.0 downgrade (or stale 1.x standards/config residue) on the very first Code run of a fresh project. Post-fix, the step pauses with `gateKind: 'mode-selection-itself'` and the LLM-side caller presents an `AskUserQuestion` to the user.
 
-- **Modified (2):** `src/services/solo/mode-gate.ts` (added 1 entry to `HARD_PAUSE_STEPS`), `src/services/solo/user-touchpoint-classifier.ts` (reclassified the row from `tech/business-only` with `fullAutoCanProceed: true` to `commit-floor/always` with `fullAutoCanProceed: false`).
+- **Modified (2):** `src/services/code/mode-gate.ts` (added 1 entry to `HARD_PAUSE_STEPS`), `src/services/code/user-touchpoint-classifier.ts` (reclassified the row from `tech/business-only` with `fullAutoCanProceed: true` to `commit-floor/always` with `fullAutoCanProceed: false`).
 - **User-observed symptom (now fixed):** "首次启动后开发效果差" — caused by 1.x standards/config residue silently surviving `full-auto` and contaminating subsequent RD/QA artifact contracts.
 
-### Bugfix — `--mode` is optional on `peaks solo should-pause` (P1, Step 1 chicken-and-egg)
+### Bugfix — `--mode` is optional on `peaks code should-pause` (P1, Step 1 chicken-and-egg)
 
-`peaks solo should-pause --step step-1-mode-select` (no `--mode` flag) failed pre-fix with `error: required option --mode <mode> not specified`. Step 1's SEMANTIC is "ask the user what mode to use" — requiring `--mode` to ask mode is a chicken-and-egg. Post-fix, `--mode` is `.option()` (not `.requiredOption()`); the action handler resolves `const mode = opts.mode ?? 'full-auto'` once at the top. All 5 downstream `opts.mode` references inside the same action body now use the local `mode` const. Existing `--mode <x>` callers are unaffected (backward-compatible). The service-layer `shouldPauseAtGate` still requires a resolved mode — defaulting is at the CLI boundary, which is the correct layer for ergonomics.
+`peaks code should-pause --step step-1-mode-select` (no `--mode` flag) failed pre-fix with `error: required option --mode <mode> not specified`. Step 1's SEMANTIC is "ask the user what mode to use" — requiring `--mode` to ask mode is a chicken-and-egg. Post-fix, `--mode` is `.option()` (not `.requiredOption()`); the action handler resolves `const mode = opts.mode ?? 'full-auto'` once at the top. All 5 downstream `opts.mode` references inside the same action body now use the local `mode` const. Existing `--mode <x>` callers are unaffected (backward-compatible). The service-layer `shouldPauseAtGate` still requires a resolved mode — defaulting is at the CLI boundary, which is the correct layer for ergonomics.
 
-- **Modified (1):** `src/cli/commands/solo-commands.ts` (option() + `mode` const + 5 reference rewrites inside the same action handler).
+- **Modified (1):** `src/cli/commands/code-commands.ts` (option() + `mode` const + 5 reference rewrites inside the same action handler).
 - **User-observed symptom (now fixed):** "再开 session 后才正常" — the first session's failed Step 1 caused the LLM-side caller to fall back to a sticky default mode; the second session's retry landed in the "right" mode by accident.
 
 ### Test surface
 
-- **Created (1):** `tests/unit/services/solo/first-run-step-gates.test.ts` — 20 vitest cases (4 modes × `step-0.55-1x-upgrade` pause guard; default-mode vs explicit-mode Step 1; AC-3 backward-compat byte-equality; 8 regression guards on the other 2 hard-pause steps; cross-check that non-hard-pause steps still auto-proceed).
-- **Modified (2, regression-data extension):** `tests/unit/services/solo/mode-gate.test.ts` (added `step-0.55-1x-upgrade` to the local `HARD_PAUSE_STEP_SET` test fixture), `tests/unit/services/solo/stale-presence-detection.test.ts` (added the new step to the case-#5 non-Step-1 filter — these existing tests encoded the BUG behavior, so leaving them stale would have caused a false-positive green).
+- **Created (1):** `tests/unit/services/code/first-run-step-gates.test.ts` — 20 vitest cases (4 modes × `step-0.55-1x-upgrade` pause guard; default-mode vs explicit-mode Step 1; AC-3 backward-compat byte-equality; 8 regression guards on the other 2 hard-pause steps; cross-check that non-hard-pause steps still auto-proceed).
+- **Modified (2, regression-data extension):** `tests/unit/services/code/mode-gate.test.ts` (added `step-0.55-1x-upgrade` to the local `HARD_PAUSE_STEP_SET` test fixture), `tests/unit/services/code/stale-presence-detection.test.ts` (added the new step to the case-#5 non-Step-1 filter — these existing tests encoded the BUG behavior, so leaving them stale would have caused a false-positive green).
 - **Updated (1):** `package.json` + `src/shared/version.ts` (2.18.3 → 2.18.4).
 
 ### Why this is the simplest fix
 
 - 1-line addition to `HARD_PAUSE_STEPS` + 5-line row replacement in the classifier + 5-line command-arg change with 5 mechanical reference rewrites = 3 source edits. No new abstractions, no new env vars, no new schemas, no CLI surface change beyond the `--mode` defaulting.
-- 800-line file cap respected: largest modified file is `solo-commands.ts` (well below the 800 cap).
+- 800-line file cap respected: largest modified file is `code-commands.ts` (well below the 800 cap).
 - Karpathy #3 surgical: 6 files touched, all justified (3 source + 1 new test + 2 test-regression updates because the existing tests encoded the BUG behavior).
 
 ### Verification (QA cycle 1, verdict pass)
 
-- AC-1: `peaks solo should-pause --step step-0.55-1x-upgrade --mode full-auto` → `ok: true, shouldPause: true, gateKind: "mode-selection-itself"` ✅
-- AC-2: `peaks solo should-pause --step step-1-mode-select` (no `--mode`) → `ok: true, shouldPause: true` (no "required option" error) ✅
-- AC-3: `peaks solo should-pause --step step-1-mode-select --mode full-auto` → byte-identical to AC-2 ✅
+- AC-1: `peaks code should-pause --step step-0.55-1x-upgrade --mode full-auto` → `ok: true, shouldPause: true, gateKind: "mode-selection-itself"` ✅
+- AC-2: `peaks code should-pause --step step-1-mode-select` (no `--mode`) → `ok: true, shouldPause: true` (no "required option" error) ✅
+- AC-3: `peaks code should-pause --step step-1-mode-select --mode full-auto` → byte-identical to AC-2 ✅
 - Cross-mode 4×2 matrix (full-auto / assisted / strict / swarm × step-0.55 / step-1): 8/8 pause ✅
-- `peaks-solo` first-run sequence (skill presence:check-stale + upgrade --detect-1x + 4 should-pause steps): 6/6 clean ✅
+- `peaks-code` first-run sequence (skill presence:check-stale + upgrade --detect-1x + 4 should-pause steps): 6/6 clean ✅
 - vitest 527/527 passed (27 files, 0 failed, 22.61s wallclock)
 - 5-way fanout audit: code-reviewer PASS / karpathy-reviewer PASS (4 guidelines) / security-reviewer PASS / perf-baseline-reviewer PASS / qa-test-cases-writer PASS (21 TCs)
 - QA verdict: `pass` (`verdict-2026-06-29-fix-first-run-step-gates.md`)
@@ -486,9 +946,9 @@ Each original file gets a sibling `import { ... } from './X-helpers.js'` + `expo
 
 ### Behavior changes (user-visible)
 
-- **None at the CLI surface.** No new commands, no removed commands, no flag changes. Existing `peaks solo`, `peaks doctor`, `peaks request lint`, `peaks scan file-size` all behave identically.
+- **None at the CLI surface.** No new commands, no removed commands, no flag changes. Existing `peaks code`, `peaks doctor`, `peaks request lint`, `peaks scan file-size` all behave identically.
 - **Binding file format change (additive only):** on-disk binding files now carry a `#<pid>` suffix in `callerId` and `ownerHint`. Old v2.16.0 / v2.17.0 files without the suffix are still readable via `BindingSchema.safeParse` (v2.18.0 read path is forward-compatible); no automatic rewrite (write-amplification risk deferred to a future `peaks doctor --rebuild-binding` slice).
-- **Multi-Claude parallel fix (user-visible in the dogfood scenarios):** opening 2 Claude Code windows in the same project and running `peaks solo` in each now yields 2 distinct sids (was: 1 collided sid, per the v2.17.0 changelog dogfood scenario 1).
+- **Multi-Claude parallel fix (user-visible in the dogfood scenarios):** opening 2 Claude Code windows in the same project and running `peaks code` in each now yields 2 distinct sids (was: 1 collided sid, per the v2.17.0 changelog dogfood scenario 1).
 - **`/compact` resume (user-visible, no change):** same outer-session-id still reuses the same sid. Verified by dogfood.
 
 ### Verification
@@ -505,7 +965,7 @@ Each original file gets a sibling `import { ... } from './X-helpers.js'` + `expo
 |---|---|---|---|
 | A | P0 fix verification (binding file format) | PASS | `callerId: 5fa8a7d8-...#13472` matches shell pid 13472; pre-existing instance pid 3876 ≠ current shell — fix works in production-shape data |
 | B | peaks doctor stale-binding scan | PASS | 0 stale bindings, ttl=300000ms, v2.16.0 schema surface intact |
-| C | peaks solo smoke (4 sub-commands) | PARTIAL | 3/4 pass; `peaks request lint --project .` is invalid syntax (needs `<rid>` + `--role`); not a v2.18.0 regression |
+| C | peaks code smoke (4 sub-commands) | PARTIAL | 3/4 pass; `peaks request lint --project .` is invalid syntax (needs `<rid>` + `--role`); not a v2.18.0 regression |
 | D | env-collision simulation (single shell, 2 envs) | PASS | `fake-session-A#20984` + `fake-session-B#20984` → 2 distinct sids (fupv8b / xnntb1); transposed axis (same pid, different env) verified; primary 2-Claude-windows scenario needs CI integration test with 2 child node processes |
 
 **Dogfood verdict:** 3.5 / 4 pass; `recommend_v2.18.0_commit: true`.
@@ -557,7 +1017,7 @@ The first v2.18.0 release deferred two open items to a redo slice; both are now
 
 ### Feature — D2 Claude-instance-level hard-exclusive sid
 
-- Same Claude instance keeps one sid across multiple peaks-* skill activations (peaks-solo → peaks-rd → peaks-qa).
+- Same Claude instance keeps one sid across multiple peaks-* skill activations (peaks-code → peaks-rd → peaks-qa).
 - `/compact` resume: same outer-session-id reuses the same sid (post-compact-resume v2.11.0 D7 alignment).
 
 ### Feature — Doctor stale-binding scan (AC-10)
@@ -576,7 +1036,7 @@ The first v2.18.0 release deferred two open items to a redo slice; both are now
 ### Behavior changes (user-visible)
 
 - **`--change-id <slug>`** is now a metadata-only slug, not a filesystem scope identifier. Existing commands that accept `--change-id` continue to work; the slug appears in envelope `data.changeId` for traceability.
-- **On-disk artifact paths** unchanged from a CLI user's perspective — reviewable artifacts still appear at the locations `peaks solo` / `peaks qa` write them. Internally they key off session id, not change-id.
+- **On-disk artifact paths** unchanged from a CLI user's perspective — reviewable artifacts still appear at the locations `peaks code` / `peaks qa` write them. Internally they key off session id, not change-id.
 - **`peaks doctor` exit code**: now exits 1 when stale bindings are present (in addition to its existing failure conditions).
 
 ### Verification
@@ -590,9 +1050,9 @@ The first v2.18.0 release deferred two open items to a redo slice; both are now
 
 Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-session-88411f/txt/handoff-2026-06-29-v2-17-0.md` for details):
 
-1. **Multi-Claude parallel提速**: open 2 Claude instances in the same project, run `peaks solo <different goal>` in each. Verify both bindings coexist with distinct sids.
+1. **Multi-Claude parallel提速**: open 2 Claude instances in the same project, run `peaks code <different goal>` in each. Verify both bindings coexist with distinct sids.
 2. **D1 conflict detection**: 2 instances both targeting the same source file. Verify each writes to its own session-scoped artifact directory.
-3. **`/compact` resume**: run `peaks solo <goal>`, `/compact`, continue. Verify same sid reused.
+3. **`/compact` resume**: run `peaks code <goal>`, `/compact`, continue. Verify same sid reused.
 4. **Doctor stale**: kill -9 a Claude instance, run `peaks doctor`, verify stale binding listed + `--cleanup-stale` works.
 
 ### Commits
@@ -605,19 +1065,19 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 ### Out of scope (deferred)
 
 - Runner wiring for D1/D2 in `peaks workspace init` + `peaks audit-goal` (modules exist, not yet called).
-- 800-line cap backlog (`session-manager.ts` 710, `solo/` and `verdict/` modules) — unchanged from v2.15.x.
+- 800-line cap backlog (`session-manager.ts` 710, `code/` and `verdict/` modules) — unchanged from v2.15.x.
 - Final removal of `change-id.ts` shims (would require rewriting the 11 src services that still reference them).
 
 ---
 
 ## [2.15.1] — 2026-06-28 — 12 Gaps CLI 全套落地 + ice-cola dogfood 验证
 
-**PATCH bump from 2.15.0**. Follow-up ship for the 12 Gaps positioning memory (peaks-loop 真实定位 + 全套 15 个 Gap CLI 落地). No breaking changes. All new commands are additive top-level (no conflict with existing `peaks solo` / `peaks qa` / `peaks slice` role commands).
+**PATCH bump from 2.15.0**. Follow-up ship for the 12 Gaps positioning memory (peaks-loop 真实定位 + 全套 15 个 Gap CLI 落地). No breaking changes. All new commands are additive top-level (no conflict with existing `peaks code` / `peaks qa` / `peaks slice` role commands).
 
 ### Feature — slice DAG layered parallelism + foundation/upstreamSync/complexity 字段
 
 - `src/services/dispatch/slice-dag.ts` — `SliceNode` 加 3 optional 字段 (`foundation?: boolean` / `upstreamSync?: boolean` / `complexity?: 'trivial'|'simple'|'complex'`). `validateDag` 加字段合法性校验 + foundation-only-depends-on-foundation 防御性规则. `topologicalLevels` 同层内 priority 排序 (foundation > upstreamSync > id asc). 老 DAG hash 稳定.
-- `src/services/solo/dag-orchestrator.ts` — 新增 `runLayeredDag` (业务 slice 不等所有 foundation,只等其 dependsOn 子集). cancel-on-fail 保留.
+- `src/services/code/dag-orchestrator.ts` — 新增 `runLayeredDag` (业务 slice 不等所有 foundation,只等其 dependsOn 子集). cancel-on-fail 保留.
 - `src/cli/commands/dispatch-from-dag.ts` — 切到 `runLayeredDag` + envelope 加 `sliceMeta` 字段.
 
 ### Feature — G11/13/14/15 CLI 全套落地 (10 commands)
@@ -631,7 +1091,7 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 
 - **G1 slice review** (4 cmds): `peaks slice-review` / `slice-score` / `slice-accept` / `slice-reject` — 4 项业务清单 + 12 Gaps 阈值 (avg >= 3 + no item <= 2)
 - **G3 prd blocks** (1 cmd): `peaks prd check-blocks` — 4 必填块校验 (业务场景 / 边界 / UI 装配 / 上游基线) + 业务禁区子节
-- **G4 user touchpoint** (3 cmds): `peaks gate-classify` / `user-touchpoints` / `commit-boundary-actions` — 14 Solo gate 静态分类 (business / tech / mode-selection / commit-boundary / commit-floor)
+- **G4 user touchpoint** (3 cmds): `peaks gate-classify` / `user-touchpoints` / `commit-boundary-actions` — 14 Code gate 静态分类 (business / tech / mode-selection / commit-boundary / commit-floor)
 - **G5 qa business** (4 cmds): `peaks qa-business-review` / `-score` / `-accept` / `-reject` — 6 项业务清单 + 同一阈值
 
 ### Feature — G6/G7/G8/G9/G10 CLI (10 commands, 30+ tests)
@@ -645,7 +1105,7 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 ### Documentation
 
 - 6 new memory files: `peaks-loop-24h-ai-programmer-positioning` / `user-role-and-tech-decision` / `prd-template-design` / `slice-review-and-qa-perspective` / `fork-sync-and-layered-parallel` / `fast-iteration-quality-loop`
-- 4 SKILL.md 校准注: peaks-solo / peaks-prd / peaks-rd / peaks-qa 加 2026-06-28 校准 anchor
+- 4 SKILL.md 校准注: peaks-code / peaks-prd / peaks-rd / peaks-qa 加 2026-06-28 校准 anchor
 - 12 Gaps 完整定位: 24h 程序员场景 / 唯一蜂群 / 反伪选择 / 业务审阅 / 轻量回归 / 上线观察期等
 
 ### Fix
@@ -687,9 +1147,9 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 - **`peaks workspace init`** (MODIFIED) — When an outer-session-mismatch rotation fires, the CLI now calls `clearStalePresenceOnRotation` to clear the stale presence. Two guards prevent accidental destruction of user-explicit mode choices:
   - **Reconnect guard** — recorded outer id matches the NEW outer id → do NOT clear (reconnect).
   - **Live-different-outer guard** — recorded outer id belongs to a different LIVE outer session → do NOT clear (would destroy another user's mode).
-- **`peaks solo should-pause --step step-1-mode-select`** (MODIFIED) — Now consults `presence:check-stale` automatically. When the presence is stale, returns `shouldPause: true, reason: 'stale-presence — re-ask Step 1'`. The hard-pause on Step 1 itself is preserved (defect #1 from slice 2026-06-28-solo-mode-bypass-fix).
-- **`skills/peaks-solo/SKILL.md` Step 1** (MODIFIED) — Wording changed from "if user did not name a profile, AskUserQuestion" to "if user did not name a profile OR presence is stale, AskUserQuestion". Cross-references the new `references/mode-selection-with-stale-presence.md`.
-- **`skills/peaks-solo/references/mode-selection-with-stale-presence.md`** (NEW) — Detection protocol + worked example (88b27d defect) + ACL.
+- **`peaks code should-pause --step step-1-mode-select`** (MODIFIED) — Now consults `presence:check-stale` automatically. When the presence is stale, returns `shouldPause: true, reason: 'stale-presence — re-ask Step 1'`. The hard-pause on Step 1 itself is preserved (defect #1 from slice 2026-06-28-code-mode-bypass-fix).
+- **`skills/peaks-code/SKILL.md` Step 1** (MODIFIED) — Wording changed from "if user did not name a profile, AskUserQuestion" to "if user did not name a profile OR presence is stale, AskUserQuestion". Cross-references the new `references/mode-selection-with-stale-presence.md`.
+- **`skills/peaks-code/references/mode-selection-with-stale-presence.md`** (NEW) — Detection protocol + worked example (88b27d defect) + ACL.
 - **`src/services/skills/skill-presence-service.ts`** — New exports: `checkStalePresence`, `clearStalePresenceOnRotation`, types `StalenessCheck`, `StaleReason`.
 
 ### Feature — User-feedback → peaks-loop enforcement (defect B fix)
@@ -702,12 +1162,12 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 
 ### Feature — Commit-boundary hard-floor (full-auto boundary = commit only)
 
-- **`src/services/solo/mode-gate.ts`** — New `HardFloorCategory` value: `'commit-boundary-side-effect'`. New `CommitBoundaryActionId` union with 5 actions: `git-push`, `git-tag`, `npm-publish`, `npm-install-global`, `peaks-global-install`. New function `detectCommitBoundaryAction(command)` matches the patterns. New `shouldPauseAtGate({ commitBoundaryAction: true })` flag — when true, ALWAYS pauses regardless of mode (overrides full-auto / swarm auto-proceed).
+- **`src/services/code/mode-gate.ts`** — New `HardFloorCategory` value: `'commit-boundary-side-effect'`. New `CommitBoundaryActionId` union with 5 actions: `git-push`, `git-tag`, `npm-publish`, `npm-install-global`, `peaks-global-install`. New function `detectCommitBoundaryAction(command)` matches the patterns. New `shouldPauseAtGate({ commitBoundaryAction: true })` flag — when true, ALWAYS pauses regardless of mode (overrides full-auto / swarm auto-proceed).
 - **Per the user-given rule** `.peaks/memory/2026-06-28-full-auto-boundary.md`: "full-auto 只做到 commit 就是，push 不用". The commit-boundary hard-floor is the machine enforcement of that advisory rule.
 
 ### Test results
 - 4 new test files: `presence-staleness.test.ts` (12), `stale-presence-detection.test.ts` (9), `feedback-promotion.test.ts` (18), `commit-boundary-hard-floor.test.ts` (247). Total new cases: **286**.
-- Existing solo tests (mode-gate × 81, post-compact × 11) pass unchanged.
+- Existing code tests (mode-gate × 81, post-compact × 11) pass unchanged.
 - Full unit suite baseline 4394 → 4680 passing (286 added). 0 new failures; pre-existing 7 unrelated failures unchanged.
 
 ### Out-of-scope
@@ -720,10 +1180,10 @@ Recommended scenarios before next release (see `peaks/_runtime/2026-06-29-sessio
 slice-2026-06-28-layered-dag PRD: 大需求(1 周内做不完)= 基础先行 + 业务并行(节省 2-3 天 wall time);fork 场景 = 上游 tag 断点同步;复杂度分流 = user-attended vs overnight 排程。
 
 - **`src/services/dispatch/slice-dag.ts`** — `SliceNode` 加 3 可选字段(`foundation?: boolean` / `upstreamSync?: boolean` / `complexity?: 'trivial'|'simple'|'complex'`)。`validateDag` 加新字段合法性校验 + 防御性规则(foundation slice 不可 dependsOn 非 foundation)。`topologicalLevels` 同层内 priority 排序: foundation > upstreamSync > id asc。`serializeDag` / `hashDag` 含新字段,**老 DAG hash 稳定**。
-- **`src/services/solo/dag-orchestrator.ts`** — 新增 `runLayeredDag` 函数。同 `runDag` 语义 + 业务 slice 不等所有 foundation,只等其 `dependsOn` 子集。cancel-on-fail 保留。`runDag` 保留(向后兼容,内部走 priority-sorted levels)。
+- **`src/services/code/dag-orchestrator.ts`** — 新增 `runLayeredDag` 函数。同 `runDag` 语义 + 业务 slice 不等所有 foundation,只等其 `dependsOn` 子集。cancel-on-fail 保留。`runDag` 保留(向后兼容,内部走 priority-sorted levels)。
 - **`src/cli/commands/dispatch-from-dag.ts`** — 切到 `runLayeredDag`。envelope 加 `sliceMeta` 字段(per-slice foundation/upstreamSync/complexity)。
-- **2 new test files** — `tests/unit/dispatch/slice-dag-foundation.test.ts` (19 cases) + `tests/unit/solo/dag-orchestrator-layered.test.ts` (5 cases) = **24 new tests, 0 regression**。
-- **dispatch + solo tests**: 215/215 通过(单跑)。
+- **2 new test files** — `tests/unit/dispatch/slice-dag-foundation.test.ts` (19 cases) + `tests/unit/code/dag-orchestrator-layered.test.ts` (5 cases) = **24 new tests, 0 regression**。
+- **dispatch + code tests**: 215/215 通过(单跑)。
 - **全量 vitest** 4843 cases: 4824 passed / 2 failed(并发 race,pre-existing)。
 
 **不触动:** transition gates / hard contracts / Karpathy 4 / sub-agent 协议 / 老 DAG 兼容性。
@@ -777,8 +1237,8 @@ slice-2026-06-28-layered-dag PRD: 大需求(1 周内做不完)= 基础先行 +
 - 8 tests pass
 
 **G4 user touchpoint classifier**(slice-2026-06-28-user-touchpoints):
-- 3 commands: `peaks solo gate-classify` / `peaks solo user-touchpoints` / `peaks solo commit-boundary-actions`
-- 14 个 Solo gate 静态分类: business / tech / mode-selection / commit-boundary / commit-floor
+- 3 commands: `peaks code gate-classify` / `peaks code user-touchpoints` / `peaks code commit-boundary-actions`
+- 14 个 Code gate 静态分类: business / tech / mode-selection / commit-boundary / commit-floor
 - `userShouldReview`: always / business-only / never
 - 7 tests pass
 
@@ -838,7 +1298,7 @@ slice-2026-06-28-layered-dag PRD: 大需求(1 周内做不完)= 基础先行 +
 会话期间从资深前端 + 后端半盲 + 极致工期 + 24h AI 程序员 + 存量无 UT 的真实场景,沉淀 6 个 memory 文件 + 索引 + 4 个 SKILL.md 校准注。**核心叙事:** `90% 效率 + 80% 质量` > `80% 效率 + 90% 质量`;user 在循环里 = 业务/产品审阅者不参与技术决策;主路径 = 唯一蜂群模式;prd 质量前置 = 4 必填块;QA = 业务视角 + 轻量回归 + 上线观察期。
 
 - **6 new memory files** — `.peaks/memory/peaks-loop-{24h-ai-programmer-positioning, user-role-and-tech-decision, prd-template-design, slice-review-and-qa-perspective, fork-sync-and-layered-parallel, fast-iteration-quality-loop}.md` (warm.project index 18 → 24)
-- **4 SKILL.md 校准注** — peaks-solo / peaks-prd / peaks-rd / peaks-qa 各加精简 anchor(均通过 25KB cap,只引用 memory 不重复内容)
+- **4 SKILL.md 校准注** — peaks-code / peaks-prd / peaks-rd / peaks-qa 各加精简 anchor(均通过 25KB cap,只引用 memory 不重复内容)
 - **不触动** transition gates / hard contracts / Karpathy 4 / 模式枚举 / mode-gate.ts / sub-agent 协议
 
 ### Fix — 2 pre-existing bugs (2026-06-28 follow-up)
@@ -966,26 +1426,26 @@ Slice B of the v2.14.0 anti-fake-green hardening PRD (`v2-14-0-anti-fake-green-h
 
 ---
 
-## [2.13.4] — 2026-06-28 — Solo mode gate + verify-pipeline canonical path + auto-compact main target
+## [2.13.4] — 2026-06-28 — Code mode gate + verify-pipeline canonical path + auto-compact main target
 
-**PATCH bump from 2.13.3** (slice `2026-06-28-solo-mode-bypass-fix`, 4 production defects reported by user in solo session 2026-06-28).
+**PATCH bump from 2.13.3** (slice `2026-06-28-code-mode-bypass-fix`, 4 production defects reported by user in code session 2026-06-28).
 
 The four defects all stem from the v2.13.0 two-axis convention landing debt: the canonical evidence location is `.peaks/_runtime/change/<changeId>/<role>/...` (per `change-scope-service.ts`), but v2.13.0's mode-gate, verify-pipeline, and auto-compact dispatcher all referenced the pre-1.3.0 sibling-of-`_runtime/` form. v2.13.4 also adds the user-requested economy-vs-concurrency separation (per direction 2026-06-28: "效率比省钱更重要，是在效率达到最大值的时候，再去考虑经济问题").
 
 ### Bug fixes
 
-- **Step 1 AskUserQuestion is no longer auto-defaulted to `mode: full-auto`** (defect #1) — `src/services/solo/mode-gate.ts:104-196` now treats `step-1-mode-select` (and `step-0.5-openspec-opt-in`, `step-0.7-resume-detection`) as a `HARD_PAUSE_STEPS` set: even `full-auto` mode pauses for the user to pick the mode. A new `gateKind: 'mode-selection-itself' | 'mode-driven' | 'hard-floor'` discriminator lets the LLM-side runner distinguish "you paused because the user must choose" from "you paused because the user already chose assisted/strict" from "you paused because a hard-floor category always wins". Dogfood-verified: a new session no longer writes `mode: "full-auto"` to `.peaks/_runtime/active-skill.json` on the first tool call without surfacing the Step 1 AskUserQuestion. 14 new test cases in `tests/unit/solo/mode-gate-step-1-hard-pause.test.ts` cover the four-mode matrix + hard-floor precedence. The pre-existing `tests/unit/services/solo/mode-gate.test.ts` (77 cases) was also updated for the new `gateKind` field — 77/77 still pass.
+- **Step 1 AskUserQuestion is no longer auto-defaulted to `mode: full-auto`** (defect #1) — `src/services/code/mode-gate.ts:104-196` now treats `step-1-mode-select` (and `step-0.5-openspec-opt-in`, `step-0.7-resume-detection`) as a `HARD_PAUSE_STEPS` set: even `full-auto` mode pauses for the user to pick the mode. A new `gateKind: 'mode-selection-itself' | 'mode-driven' | 'hard-floor'` discriminator lets the LLM-side runner distinguish "you paused because the user must choose" from "you paused because the user already chose assisted/strict" from "you paused because a hard-floor category always wins". Dogfood-verified: a new session no longer writes `mode: "full-auto"` to `.peaks/_runtime/active-skill.json` on the first tool call without surfacing the Step 1 AskUserQuestion. 14 new test cases in `tests/unit/code/mode-gate-step-1-hard-pause.test.ts` cover the four-mode matrix + hard-floor precedence. The pre-existing `tests/unit/services/code/mode-gate.test.ts` (77 cases) was also updated for the new `gateKind` field — 77/77 still pass.
 - **`peaks workflow verify-pipeline` now resolves the canonical evidence path** (defect #3) — `src/services/workflow/pipeline-verify-service.ts:216,219,260,288,295` rebuilds evidence paths as `.peaks/_runtime/change/<changeId>/<role>/...` (was `.peaks/<changeId>/<role>/...`, the SKILL.md 2.8.3 hard-ban shape). `src/services/workflow/artifact-paths.ts:67,148` gets the same fix in the security/performance findings resolver. A 1-minor-release deprecation window accepts the legacy `.peaks/<changeId>/...` and `.peaks/_runtime/<changeId>/...` forms with a `DEPRECATION_LEGACY_PATH_USED` warning so un-migrated workspaces still resolve. `PipelineVerification.usedCanonicalPath: boolean` is added to the return envelope so QA / TXT can surface the deprecation state. The CLI help-text at `src/cli/commands/workflow-commands.ts:448` is updated to cite the canonical path. 4 new test cases in `tests/unit/workflow/pipeline-verify-canonical-path.test.ts` cover canonical / legacy misplaced / top-level fallback / absent. The pre-existing `tests/unit/pipeline-verify-service.test.ts` (60 cases) was updated to write evidence at both paths so the deprecation contract is exercised end-to-end — 60/60 still pass.
-- **`auto-compact` now targets the main-session context, not a sub-agent shell** (defect #4) — `src/services/context/auto-compact-dispatcher.ts` and `src/services/solo/auto-compact-orchestrator.ts` accept a new `target: 'main' | 'sub-agent'` parameter (default `'main'`). For `target='main' + ide='claude-code'`, the dispatcher returns the `llm-self-compress` pathway and the orchestrator writes `.peaks/_runtime/<sessionId>/txt/auto-compact-pending.json` (with `pending: true, target: 'main', ratio, redLine`) so the next main-session LLM turn fires `/compact` in-band rather than spawning a detached `sh -c /compact` (which previously only compressed the sub-agent shell). For `target='sub-agent'`, the legacy shell-spawn behavior is preserved. Non-claude-code IDEs + `target='main'` return `noop` with a "main-session target unsupported" reason. 6 new test cases in `tests/unit/context/auto-compact-main-target.test.ts` cover the dispatch matrix + the orchestrator's intent-file write.
+- **`auto-compact` now targets the main-session context, not a sub-agent shell** (defect #4) — `src/services/context/auto-compact-dispatcher.ts` and `src/services/code/auto-compact-orchestrator.ts` accept a new `target: 'main' | 'sub-agent'` parameter (default `'main'`). For `target='main' + ide='claude-code'`, the dispatcher returns the `llm-self-compress` pathway and the orchestrator writes `.peaks/_runtime/<sessionId>/txt/auto-compact-pending.json` (with `pending: true, target: 'main', ratio, redLine`) so the next main-session LLM turn fires `/compact` in-band rather than spawning a detached `sh -c /compact` (which previously only compressed the sub-agent shell). For `target='sub-agent'`, the legacy shell-spawn behavior is preserved. Non-claude-code IDEs + `target='main'` return `noop` with a "main-session target unsupported" reason. 6 new test cases in `tests/unit/context/auto-compact-main-target.test.ts` cover the dispatch matrix + the orchestrator's intent-file write.
 
 ### Features
 
-- **`peaks workspace migrate-change-scope --project <path> [--apply] [--json]`** — slice 2026-06-28-solo-mode-bypass-fix migration tool. Dry-run by default; `--apply` atomically renames misplaced `.peaks/_runtime/<changeId>/` (and `.peaks/<changeId>/`) entries into the canonical `.peaks/_runtime/change/<changeId>/` location, writes a `.peaks-migration.json` marker (with `from:`, `slice:`, `tool:`, `migratedAt:`) for audit. **Refusal conditions** (defense-in-depth, both pre-slice audit + new): entries that look like date-stamped session ids (`YYYY-MM-DD-session-X`) are refused to avoid destroying the session workspace (`MIGRATION_REFUSED_SESSION_ID_COLLISION`); entries whose target dir exists with non-byte-equal contents are refused to avoid clobbering (`MIGRATION_REFUSED_TARGET_NOT_EMPTY`); a hard-coded `PEAKS_TOP_LEVEL_DENY` / `RUNTIME_DENY` whitelist + `looksLikeChangeScopeId` structural check ensures `.peaks/memory`, `.peaks/standards`, `.peaks/retrospective`, `.peaks/sc`, `.peaks/sops`, `.peaks/project-scan`, `.peaks/_sub_agents` are **never** treated as misplaced change-ids. **Idempotent**: re-running on a clean workspace reports no work. Dogfood-verified: the actual misplaced `.peaks/_runtime/2026-06-27-verdict-aggregator-v2-12-debt/` (8 files) was migrated end-to-end; subsequent `peaks workflow verify-pipeline --rid 2026-06-27-verdict-aggregator-v2-12-debt --change-id 2026-06-27-verdict-aggregator-v2-12-debt --project .` reports `usedCanonicalPath: true` (was `false`) and zero `DEPRECATION_LEGACY_PATH_USED` warnings. 5 new test cases in `tests/unit/workspace/migrate-change-scope.test.ts` cover dry-run, apply, idempotency, session-id refusal, and target-not-empty refusal; a 6th case locks the `.peaks/<project-data>` whitelist contract.
+- **`peaks workspace migrate-change-scope --project <path> [--apply] [--json]`** — slice 2026-06-28-code-mode-bypass-fix migration tool. Dry-run by default; `--apply` atomically renames misplaced `.peaks/_runtime/<changeId>/` (and `.peaks/<changeId>/`) entries into the canonical `.peaks/_runtime/change/<changeId>/` location, writes a `.peaks-migration.json` marker (with `from:`, `slice:`, `tool:`, `migratedAt:`) for audit. **Refusal conditions** (defense-in-depth, both pre-slice audit + new): entries that look like date-stamped session ids (`YYYY-MM-DD-session-X`) are refused to avoid destroying the session workspace (`MIGRATION_REFUSED_SESSION_ID_COLLISION`); entries whose target dir exists with non-byte-equal contents are refused to avoid clobbering (`MIGRATION_REFUSED_TARGET_NOT_EMPTY`); a hard-coded `PEAKS_TOP_LEVEL_DENY` / `RUNTIME_DENY` whitelist + `looksLikeChangeScopeId` structural check ensures `.peaks/memory`, `.peaks/standards`, `.peaks/retrospective`, `.peaks/sc`, `.peaks/sops`, `.peaks/project-scan`, `.peaks/_sub_agents` are **never** treated as misplaced change-ids. **Idempotent**: re-running on a clean workspace reports no work. Dogfood-verified: the actual misplaced `.peaks/_runtime/2026-06-27-verdict-aggregator-v2-12-debt/` (8 files) was migrated end-to-end; subsequent `peaks workflow verify-pipeline --rid 2026-06-27-verdict-aggregator-v2-12-debt --change-id 2026-06-27-verdict-aggregator-v2-12-debt --project .` reports `usedCanonicalPath: true` (was `false`) and zero `DEPRECATION_LEGACY_PATH_USED` warnings. 5 new test cases in `tests/unit/workspace/migrate-change-scope.test.ts` cover dry-run, apply, idempotency, session-id refusal, and target-not-empty refusal; a 6th case locks the `.peaks/<project-data>` whitelist contract.
 
 ### Internal
 
-- `src/services/solo/mode-gate.ts` (+28/-4 lines) — `HARD_PAUSE_STEPS` set + `GateKind` union + `gateKind` field on `GateDecision`.
-- `src/services/solo/auto-compact-orchestrator.ts` (+24/-2 lines) — `target` parameter, `writeMainSessionCompactIntent` helper, surface `target` in return envelope.
+- `src/services/code/mode-gate.ts` (+28/-4 lines) — `HARD_PAUSE_STEPS` set + `GateKind` union + `gateKind` field on `GateDecision`.
+- `src/services/code/auto-compact-orchestrator.ts` (+24/-2 lines) — `target` parameter, `writeMainSessionCompactIntent` helper, surface `target` in return envelope.
 - `src/services/context/auto-compact-dispatcher.ts` (+38/-5 lines) — `CompactTarget` type, `target` parameter, reordered non-claude-code refusal, `shell-exec + main` → `llm-self-compress` rewrite.
 - `src/services/context/auto-compact-types.ts` (+6 lines) — `target?: 'main' | 'sub-agent'` on `AutoCompactResult.data`.
 - `src/services/preferences/preferences-types.ts` (+13 lines) — explicit doc-comments on `economyMode` (model selection only, NOT concurrency) and `swarmMode` (controls subgraph shape, NOT fan-out). Fan-out is governed by `fanout.defaultMode: 'fan-out'` (hard constraint per slice 2026-06-24-audit-5th-p2).
@@ -996,12 +1456,12 @@ The four defects all stem from the v2.13.0 two-axis convention landing debt: the
 - `src/cli/commands/workspace/migrate-change-scope-command.ts` (NEW, 230 lines) — `migrateChangeScope()` core + `migrateOne()` per-entry handler with the deny-list + `shallowContentEqual` (1-level recursion) + 3 refusal conditions.
 - `tests/unit/workspace/banned-path-directive-guard.test.ts` (+11 lines) — added `KEEP_DESCRIPTIONS` anchor for the new help-text so the AC-2.2 banned-path-guard still passes.
 - `tests/unit/pipeline-verify-service.test.ts` (+24/-12 lines) — `writeRdEvidence` / `writeQaEvidence` now write at canonical + legacy paths; `isResolvedChangeId` updated for the bare-id contract.
-- `tests/unit/services/solo/mode-gate.test.ts` (linter-updated) — assertions for the new `gateKind` field on every `GateDecision` return.
-- 5 new test files: `tests/unit/solo/mode-gate-step-1-hard-pause.test.ts` (14 cases), `tests/unit/workflow/pipeline-verify-canonical-path.test.ts` (4 cases), `tests/unit/workflow/artifact-paths-canonical.test.ts` (4 cases), `tests/unit/context/auto-compact-main-target.test.ts` (6 cases), `tests/unit/workspace/migrate-change-scope.test.ts` (6 cases) — 34 new cases total, all pass.
+- `tests/unit/services/code/mode-gate.test.ts` (linter-updated) — assertions for the new `gateKind` field on every `GateDecision` return.
+- 5 new test files: `tests/unit/code/mode-gate-step-1-hard-pause.test.ts` (14 cases), `tests/unit/workflow/pipeline-verify-canonical-path.test.ts` (4 cases), `tests/unit/workflow/artifact-paths-canonical.test.ts` (4 cases), `tests/unit/context/auto-compact-main-target.test.ts` (6 cases), `tests/unit/workspace/migrate-change-scope.test.ts` (6 cases) — 34 new cases total, all pass.
 
 ### Test results
 
-- `pnpm vitest run` on the 4 affected module areas (solo/workflow/context/workspace): **181/181 pass** (5 new test files + pre-existing suites).
+- `pnpm vitest run` on the 4 affected module areas (code/workflow/context/workspace): **181/181 pass** (5 new test files + pre-existing suites).
 - Full unit suite: 4394/4418 pass (7 pre-existing failures unrelated to this slice — `doctor.test.ts` version-mismatch × 5, `tokenizer.test.ts` time-decay flake × 1, `35-checks-aggregate.test.ts` × 1).
 - `pnpm tsc --noEmit`: clean.
 
@@ -1051,7 +1511,7 @@ The four defects all stem from the v2.13.0 two-axis convention landing debt: the
 |---|---|
 | v2.13.3 | 4 bug fixes (parser / publish pipeline / CLI warnings / handoff sha256) + 3 new scripts + 4 new test cases + package.json prepublishOnly + README publish note + CHANGELOG + version bump + ship-state memory |
 
-### Verified (peaks solo dogfood + QA on this session)
+### Verified (peaks code dogfood + QA on this session)
 
 - AC-1 (parser fix): `tests/unit/services/verdict/envelopes.test.ts` → **11/11 pass** (was 7/7 in v2.13.2, +4 markdown fallback cases H/I/J/K). Dogfood script confirms: real `audit/security.md` with HIGH violation → `parseSecurityEnvelope` returns non-null envelope; `peaks verdict aggregate` returns `reasons: [{severity: HIGH, file: src/auth.ts:42}]`.
 - AC-2 (publish pipeline): `scripts/prepublish-build.sh` end-to-end via git-bash: `[prepublish-build] build OK — proceeding to publish` (exit 0). The `prepublishOnly` hook in `package.json` (line 47) is wired to `node scripts/prepublish-build.mjs`.
@@ -1129,7 +1589,7 @@ v2.13.1 shipped with a BLOCKER bug found via post-release dogfood: `aggregateVer
 |---|---|
 | v2.13.2 | `aggregateVerdict()` dedup bug fix + `peaks verdict aggregate` CLI + `envelopes.ts` unification + `prd/handoff.md` auto-regen + `MUT_REPORT` soft-block window + 4 new test files + 4 updated test files + CHANGELOG + version bump + ship-state memory |
 
-### Verified (peaks solo dogfood + QA on this session)
+### Verified (peaks code dogfood + QA on this session)
 
 - AC-1 (BLOCKER fix): `tests/unit/services/verdict/verdict-aggregator.test.ts` → **16/16 pass** (was 13/13 in v2.13.1, +3 cross-source cases). Dogfood script confirms: `reasons.length = 1, sources = ["security-audit","perf-audit"], verdict = warn`.
 - AC-2 (CLI surface): `tests/unit/cli/commands/verdict-aggregate-command.test.ts` → **4/4 pass**. Real CLI: `peaks verdict aggregate --help` shows `--from-rid/--sid/--project/--json` options correctly.
@@ -1160,11 +1620,11 @@ v2.13.1 shipped with a BLOCKER bug found via post-release dogfood: `aggregateVer
 
 ---
 
-## [2.13.1] — 2026-06-27 — Verdict reasoning layer (multi-signal convergence for peaks-solo)
+## [2.13.1] — 2026-06-27 — Verdict reasoning layer (multi-signal convergence for peaks-code)
 
 **PATCH bump from 2.13.0** (slice `2026-06-27-verdict-aggregator`, red-line scope 3 source files + 4 new test files + 2 updated test files).
 
-peaks-solo previously received 5 heterogeneous signals (security-audit, perf-audit, karpathy-reviewer, peaks-mut, peaks-qa) but had no convergence layer. The v2.12.0 audit-output schema documented 4 aggregation rules (`.peaks/project-scan/audit-output-schema.md:66-78`) but they were never implemented; the `mut-report.json` was consumed by peaks-qa internally with `loadMutReport() === null → gate=skipped` (soft consumption), and `micro-cycle.md` had no verdict-reasoning surface. v2.13.1 fills the gap without unifying envelope schemas (deferred to v2.14) and without changing any verdict string.
+peaks-code previously received 5 heterogeneous signals (security-audit, perf-audit, karpathy-reviewer, peaks-mut, peaks-qa) but had no convergence layer. The v2.12.0 audit-output schema documented 4 aggregation rules (`.peaks/project-scan/audit-output-schema.md:66-78`) but they were never implemented; the `mut-report.json` was consumed by peaks-qa internally with `loadMutReport() === null → gate=skipped` (soft consumption), and `micro-cycle.md` had no verdict-reasoning surface. v2.13.1 fills the gap without unifying envelope schemas (deferred to v2.14) and without changing any verdict string.
 
 ### Features
 
@@ -1176,10 +1636,10 @@ peaks-solo previously received 5 heterogeneous signals (security-audit, perf-aud
 
 - `src/services/verdict/verdict-aggregator.ts` (NEW, 223 lines) — pure `aggregateVerdict()` + locally-defined `KarpathyEnvelope` / `MutEnvelope` / `QaEnvelope` types (surgeon scope; v2.14 will move them to shared if a unification pass lands).
 - `src/services/artifacts/artifact-prerequisites.ts` — added `MUT_REPORT` constant (32 lines) + wired into `FEATURE_TABLE['rd:qa-handoff']` (line 276) and `BUGFIX_TABLE['rd:qa-handoff']` (line 303); `REFACTOR_TABLE` inherits via reference (line 312); `MINIMAL_TABLE` / `CONFIG_TABLE` exempt.
-- `skills/peaks-solo/references/micro-cycle.md` — added `## Verdict reasoning (v2.13.1)` section (91 lines) after the unchanged repair-cycle cap rule.
+- `skills/peaks-code/references/micro-cycle.md` — added `## Verdict reasoning (v2.13.1)` section (91 lines) after the unchanged repair-cycle cap rule.
 - `tests/unit/artifact-prerequisites/mut-report-prereq.test.ts` (NEW, 4 cases).
 - `tests/unit/services/verdict/verdict-aggregator.test.ts` (NEW, 13 cases).
-- `tests/unit/skills/solo/micro-cycle-verdict-reasoning.test.ts` (NEW, 4 cases).
+- `tests/unit/skills/code/micro-cycle-verdict-reasoning.test.ts` (NEW, 4 cases).
 - `tests/unit/artifact-prerequisites.test.ts` (UPDATED, +25 lines) — seeded `mut-report.json` in 3 pass-path tests; added to negative-path missing-list.
 - `tests/unit/artifact-prerequisites-typed.test.ts` (UPDATED, +20 lines) — same across bugfix + feature + refactor.
 
@@ -1193,11 +1653,11 @@ peaks-solo previously received 5 heterogeneous signals (security-audit, perf-aud
 |---|---|
 | v2.13.1 | MUT_REPORT prereq + `aggregateVerdict()` service + `## Verdict reasoning` section + 4 new test files + 2 updated test files + CHANGELOG + version bump + ship-state memory |
 
-### Verified (peaks solo dogfood on this session)
+### Verified (peaks code dogfood on this session)
 
 - AC-1 (MUT_REPORT prereq): `tests/unit/artifact-prerequisites/mut-report-prereq.test.ts` → 4/4 pass; `tests/unit/artifact-prerequisites.test.ts` → 9/9 pass; `tests/unit/artifact-prerequisites-typed.test.ts` → 19/19 pass.
 - AC-2 (verdict-aggregator): `tests/unit/services/verdict/verdict-aggregator.test.ts` → 13/13 pass; 8 AC-2 behaviors (A all-pass, B security-block, C mut-block, D qa-return-to-rd, E mixed-warn, F all-empty, G precedence block-dominant, H CRITICAL accumulation) all asserted.
-- AC-3 (micro-cycle reasoning): `tests/unit/skills/solo/micro-cycle-verdict-reasoning.test.ts` → 4/4 pass; 6-step cycle body byte-stable.
+- AC-3 (micro-cycle reasoning): `tests/unit/skills/code/micro-cycle-verdict-reasoning.test.ts` → 4/4 pass; 6-step cycle body byte-stable.
 - AC-4 (零回归): `tests/unit/parallel-fan-out.test.ts` → 18/18 pass (v2.12.0 stability pin); `tests/unit/rd/karpathy-skip-on-config-docs-chore.test.ts` → 11/11 pass; `tests/unit/rd/deprecated-reviewer-back-compat.test.ts` → 12/12 pass.
 - Total: 8 test files, **90/90** tests pass, duration 1.27s.
 - `./node_modules/.bin/tsc --noEmit` → 0 errors.
@@ -1224,7 +1684,7 @@ peaks-solo previously received 5 heterogeneous signals (security-audit, perf-aud
 
 **MINOR bump from 2.12.0** (slice `v2-13-0-auto-compact-protocol`, 5-sub-task plan AC-1..AC-5, red-line scope ~6 source files + 2 IDE adapter fields).
 
-peaks-solo now autonomously drives context compaction so the LLM-runner stays alive with context < 95% on **any AI CLI**, with **zero human / zero LLM intervention**. Two-tier threshold model:
+peaks-code now autonomously drives context compaction so the LLM-runner stays alive with context < 95% on **any AI CLI**, with **zero human / zero LLM intervention**. Two-tier threshold model:
 
 - **85% pre-compact zone** — peaks-loop writes a pre-compact checkpoint + convergence plan + auto-decisions log + IDE-side compact dispatch (async-friendly).
 - **95% RED LINE** — peaks-loop refuses sub-agent dispatch and forces synchronous IDE compact; mandatory, LLM cannot opt out.
@@ -1233,9 +1693,9 @@ Adapter-driven protocol (no hard-coded IDE names): `IdeAdapter.compact?: IdeComp
 
 ### Features
 
-- **`peaks solo context-now`** (AC-1) — auto-probes the active IDE adapter's context-fill % without requiring the LLM to pass `--prompt-size <bytes>` manually. Adapter-driven: reads `IdeAdapter.compact.envVarForContextPercent` (Claude Code MVP: `CLAUDE_CONTEXT_USAGE_PERCENT`). Returns a verdict (`ok` / `soft-warn` / `pre-compact` / `red-line`) plus source-tagged probe (`claude-code-env` / `statusline-poll` / `conservative-fallback`). When no IDE-specific signal is available, returns `ratio: 0` with `source: 'conservative-fallback'` so the orchestrator never auto-fires on a missing signal.
-- **`peaks solo auto-compact`** (AC-4) — 0-intervention loop. Honors D6.e in-flight-batch deferral for the pre-compact zone; forces synchronous dispatch at red-line. `--force` and `--bypass-red-line` are test seams (never `true` in production).
-- **Convergence toolkit** (AC-2) — `src/services/solo/auto-compact-orchestrator.ts` `evaluateCompactTrigger` (pure) + `runAutoCompact` (side effects: `writePreCompactCheckpoint` + `appendAutoDecisionLog` + `dispatchIdeCompact`). Checkpoints land at `.peaks/_runtime/<sessionId>/checkpoints/{pre-compact,red-line}-<ISO>.json`; the LLM-readable decision log lands at `.peaks/_runtime/<sessionId>/txt/auto-decisions.md` so D7's post-compact-detect picks it up unchanged.
+- **`peaks code context-now`** (AC-1) — auto-probes the active IDE adapter's context-fill % without requiring the LLM to pass `--prompt-size <bytes>` manually. Adapter-driven: reads `IdeAdapter.compact.envVarForContextPercent` (Claude Code MVP: `CLAUDE_CONTEXT_USAGE_PERCENT`). Returns a verdict (`ok` / `soft-warn` / `pre-compact` / `red-line`) plus source-tagged probe (`claude-code-env` / `statusline-poll` / `conservative-fallback`). When no IDE-specific signal is available, returns `ratio: 0` with `source: 'conservative-fallback'` so the orchestrator never auto-fires on a missing signal.
+- **`peaks code auto-compact`** (AC-4) — 0-intervention loop. Honors D6.e in-flight-batch deferral for the pre-compact zone; forces synchronous dispatch at red-line. `--force` and `--bypass-red-line` are test seams (never `true` in production).
+- **Convergence toolkit** (AC-2) — `src/services/code/auto-compact-orchestrator.ts` `evaluateCompactTrigger` (pure) + `runAutoCompact` (side effects: `writePreCompactCheckpoint` + `appendAutoDecisionLog` + `dispatchIdeCompact`). Checkpoints land at `.peaks/_runtime/<sessionId>/checkpoints/{pre-compact,red-line}-<ISO>.json`; the LLM-readable decision log lands at `.peaks/_runtime/<sessionId>/txt/auto-decisions.md` so D7's post-compact-detect picks it up unchanged.
 - **IDE-aware compact dispatcher** (AC-3) — `src/services/context/auto-compact-dispatcher.ts` reads `IdeAdapter.compact` and dispatches via the adapter-declared pathway. `shell-exec` (Claude Code MVP) spawns the compact command via `child_process.spawn`. `ide-native` is reserved for a future slice. `llm-self-compress` returns success + instructs the LLM to summarize on next turn. `noop` returns explicit failure for legacy adapters.
 
 ### Internal
@@ -1243,10 +1703,10 @@ Adapter-driven protocol (no hard-coded IDE names): `IdeAdapter.compact?: IdeComp
 - `src/services/context/auto-compact-types.ts` (NEW) — types + 3 constants: `AUTO_COMPACT_SOFT_WARN_RATIO = 0.5`, `AUTO_COMPACT_PRE_COMPACT_RATIO = 0.85`, `AUTO_COMPACT_RED_LINE_RATIO = 0.95`.
 - `src/services/context/auto-compact-reader.ts` (NEW) — `readContextPercent` (AC-1 probe).
 - `src/services/context/auto-compact-dispatcher.ts` (NEW) — `dispatchIdeCompact` (AC-3 IDE dispatch).
-- `src/services/solo/auto-compact-orchestrator.ts` (NEW) — `evaluateCompactTrigger` + `runAutoCompact` (AC-2 + AC-4 core).
+- `src/services/code/auto-compact-orchestrator.ts` (NEW) — `evaluateCompactTrigger` + `runAutoCompact` (AC-2 + AC-4 core).
 - `src/services/ide/ide-types.ts` — `IdeAdapter.compact?: IdeCompactProfile` + `IdeCompactProfile` interface.
 - `src/services/ide/adapters/claude-code-adapter.ts` — MVP `compact` profile: `CLAUDE_CONTEXT_USAGE_PERCENT` + `claude --compact` + `shell-exec`.
-- `src/cli/commands/solo-commands.ts` — `peaks solo context-now` + `peaks solo auto-compact` subcommands.
+- `src/cli/commands/code-commands.ts` — `peaks code context-now` + `peaks code auto-compact` subcommands.
 - `package.json` + `src/shared/version.ts` — `2.12.0 → 2.13.0`.
 
 ### Decision records
@@ -1257,10 +1717,10 @@ Adapter-driven protocol (no hard-coded IDE names): `IdeAdapter.compact?: IdeComp
 
 | Commit tag | Scope |
 |---|---|
-| v2.13.0-alpha.1 (`edffc33`) | Two-tier threshold + auto-compact types + reader + dispatcher + orchestrator + Claude Code MVP adapter + `peaks solo context-now` + `peaks solo auto-compact` CLI |
+| v2.13.0-alpha.1 (`edffc33`) | Two-tier threshold + auto-compact types + reader + dispatcher + orchestrator + Claude Code MVP adapter + `peaks code context-now` + `peaks code auto-compact` CLI |
 | `a8b9804` | In-session dogfood limitation documented (current ad-hoc Claude Code session cannot be externally compacted — reserved for follow-up PreToolUse-hook slice) |
 
-### Verified (peaks solo dogfood on this session)
+### Verified (peaks code dogfood on this session)
 
 - `context-now` boundary tests: `0.30 = ok` / `0.50 = soft-warn` / `0.84 = soft-warn` / `0.85 = pre-compact` / `0.949 = pre-compact` / `0.95 = red-line` / `1.0 = red-line` ✓
 - `auto-compact @ 1.0` (red-line): dispatched (shell-exec ok), `red-line` checkpoint written at `.peaks/_runtime/<sessionId>/checkpoints/red-line-<ISO>.json`, convergence plan + auto-decisions.md appended, `redLineGated: true` ✓
@@ -1273,12 +1733,12 @@ Adapter-driven protocol (no hard-coded IDE names): `IdeAdapter.compact?: IdeComp
 
 ### Out-of-scope (NOT changed)
 
-- `src/services/code-review/ecc-bridge.ts` + `src/services/dispatch/sub-agent-dispatcher.ts` + `src/services/agent/ecc-agent-service.ts` + `src/services/prd/handoff-service.ts` + `project-scan-reader.ts` + `src/services/rd/{strategic,tactical,strategy,impl,ast-gate,types}.ts` + `peaks-qa/` + `peaks-solo/SKILL.md` main flow + `peaks-prd/SKILL.md` main body — all untouched per the v2.13.0 red-line scope.
+- `src/services/code-review/ecc-bridge.ts` + `src/services/dispatch/sub-agent-dispatcher.ts` + `src/services/agent/ecc-agent-service.ts` + `src/services/prd/handoff-service.ts` + `project-scan-reader.ts` + `src/services/rd/{strategic,tactical,strategy,impl,ast-gate,types}.ts` + `peaks-qa/` + `peaks-code/SKILL.md` main flow + `peaks-prd/SKILL.md` main body — all untouched per the v2.13.0 red-line scope.
 
 ### Known limitations (carry-forward to v2.13.1)
 
 - **Ad-hoc Claude Code runner cannot be externally compacted.** The v2.13.0-alpha.1 shell-exec pathway spawns the IDE's compact command via `child_process.spawn` — a separate child process. The current Claude Code runner that invoked `auto-compact` is unaffected; its own context window stays at 100% until that runner's own compact logic kicks in (Claude Code's own auto-compact or a user-issued `/compact` slash command). Follow-up slice: register a PreToolUse hook via `peaks hooks install` that intercepts the next Bash call and writes a stderr hint when ratio ≥ 0.95 — fills the already-reserved `ide-native` compact pathway.
-- **`peaks-solo` Step N+2 prose update** — `skills/peaks-solo/SKILL.md` should mention `peaks solo context-now` + `auto-compact` so LLM sessions invoke the autonomous loop instead of `--prompt-size` hand-passing.
+- **`peaks-code` Step N+2 prose update** — `skills/peaks-code/SKILL.md` should mention `peaks code context-now` + `auto-compact` so LLM sessions invoke the autonomous loop instead of `--prompt-size` hand-passing.
 
 ---
 
@@ -1340,7 +1800,7 @@ Each Group A→D ran the full RD→QA loop independently. Group C final QA: 14/1
 
 ### Out-of-scope (NOT changed)
 
-`src/services/code-review/ecc-bridge.ts` + `src/services/dispatch/sub-agent-dispatcher.ts` + `src/services/agent/ecc-agent-service.ts` + `src/services/prd/handoff-service.ts` + `project-scan-reader.ts` + `src/services/rd/{strategic,tactical,strategy,impl,ast-gate,types}.ts` + `peaks-qa/` + `peaks-solo/SKILL.md` main flow + `peaks-prd/SKILL.md` main body — all untouched per the v2.12.0 red-line scope.
+`src/services/code-review/ecc-bridge.ts` + `src/services/dispatch/sub-agent-dispatcher.ts` + `src/services/agent/ecc-agent-service.ts` + `src/services/prd/handoff-service.ts` + `project-scan-reader.ts` + `src/services/rd/{strategic,tactical,strategy,impl,ast-gate,types}.ts` + `peaks-qa/` + `peaks-code/SKILL.md` main flow + `peaks-prd/SKILL.md` main body — all untouched per the v2.12.0 red-line scope.
 
 ---
 
@@ -1355,7 +1815,7 @@ Read-only observability layer on top of the v2.11.0 slice topology + 10/90 parad
 - **`peaks observability status`** (AC-1) — aggregate metrics for active session: total slices, success count, fail count, fanout cost total, repair-cycle peak.
 - **`peaks observability slices`** (AC-2) — per-slice list with rid, state, fanout count, repair-cycle count, duration (ms).
 - **`peaks observability fanout`** (AC-3) — fanout cost breakdown per sub-agent role (rd / qa / code-reviewer / security-reviewer / karpathy-reviewer).
-- **`peaks observability repair-cycles`** (AC-4) — RD→QA repair-cycle count per slice; cap = 3 (peaks-solo repair-loop contract); capHit flag.
+- **`peaks observability repair-cycles`** (AC-4) — RD→QA repair-cycle count per slice; cap = 3 (peaks-code repair-loop contract); capHit flag.
 - **`peaks observability report --period day|week|month`** (AC-5) — markdown summary (header + status + slice table + fanout table + repair-cycle table + top-5 slowest) suitable for paste into PR descriptions or `.peaks/PROJECT.md` timeline entries.
 - **JSONL persistence** (AC-6) — append-only `.peaks/_runtime/<sessionId>/metrics/slices.jsonl`; zod schema v1; cross-session prune to last 10 session files by mtime.
 - **Hook integration** (AC-7) — metrics emitted from 7 sites: `peaks request transition` (Slice A), `peaks sub-agent dispatch`, `peaks session checkpoint`, D5 mode-gate, D6 context-trigger, D7 post-compact, `peaks request transition` RD→QA prereq (Slice C). All emits fire-and-forget per PRD Q4 (full-auto must never fail-loud).
@@ -1373,9 +1833,9 @@ Read-only observability layer on top of the v2.11.0 slice topology + 10/90 parad
 - Modified: `src/services/artifacts/request-artifact-service.ts` (hook #1/7).
 - Modified: `src/cli/commands/dispatch-commands.ts` (hook #2/7).
 - Modified: `src/services/session/session-checkpoint-service.ts` (hook #3/7).
-- Modified: `src/cli/commands/solo-commands.ts` (hook #4/7).
+- Modified: `src/cli/commands/code-commands.ts` (hook #4/7).
 - Modified: `src/cli/commands/context-commands.ts` (hook #5/7).
-- Modified: `src/services/solo/post-compact-detector.ts` (hook #6/7).
+- Modified: `src/services/code/post-compact-detector.ts` (hook #6/7).
 - Modified: `src/services/artifacts/artifact-prerequisites.ts` (hook #7/7).
 - Modified: `src/cli/program.ts` (+1 line: registerObservabilityCommands).
 - Modified: `src/shared/version.ts` (CLI_VERSION 2.11.0 → 2.11.1).
@@ -1387,7 +1847,7 @@ Read-only observability layer on top of the v2.11.0 slice topology + 10/90 parad
 
 **MINOR bump from 2.10.0** (slice `v2-11-rm-rd-techdoc-immutable-handoff`, 6 multi-CC groups A-F, plan at `.peaks/memory/2026-06-26-v2-11-rm-rd-techdoc-immutable-handoff.md`).
 
-Implements the "metering is value" + "10/90 paradigm" alignment: peaks-prd produces a single immutable handoff that all downstream consumers share; peaks-rd's parallel audit fan-out (now 5-way with ECC code-review + karpathy-reviewer hard gate) owns security/perf review; peaks-qa is trimmed to business-test-only; peaks-txt sediments business knowledge; peaks-solo removes runtime friction (auto-proceed, context monitor, post-compact resume).
+Implements the "metering is value" + "10/90 paradigm" alignment: peaks-prd produces a single immutable handoff that all downstream consumers share; peaks-rd's parallel audit fan-out (now 5-way with ECC code-review + karpathy-reviewer hard gate) owns security/perf review; peaks-qa is trimmed to business-test-only; peaks-txt sediments business knowledge; peaks-code removes runtime friction (auto-proceed, context monitor, post-compact resume).
 
 ### Features
 
@@ -1396,7 +1856,7 @@ Implements the "metering is value" + "10/90 paradigm" alignment: peaks-prd produ
 - **Tier 5+6 (Group C — `9fea8eb`)** — peaks-txt sediment step (`appendBusinessConcept`, idempotent on (concept, sourceRid), 7 tests); peaks-qa trim (removed `qa/security-findings.md` + `qa/performance-findings.md` from Gate D prerequisites).
 - **Tier 7 (Group D — `cd427f6`)** — ECC code-review bridge (`src/services/code-review/ecc-bridge.ts`): envelope validator `isEccEnvelope` + `adaptEccEnvelopeToRdCodeReview` + 5-state `detectEcc` + `runEccCodeReview` aggregator; 17 tests.
 - **Tier 8 (Group E — this CC)** — migration codemod `peaks migrate v2-10-to-v2-11` (deprecates historical `rd/tech-doc.md` files with a YAML banner frontmatter; text-only, idempotent).
-- **Tier 9 (Group F — commit `9e3ef49`)** — D5 self-decision (`src/services/solo/mode-gate.ts` + `peaks solo should-pause`); D6 context monitor (`src/services/context/main-session-monitor.ts` + `peaks context check`); D7 post-compact resume (`src/services/solo/post-compact-detector.ts` + `peaks solo post-compact-detect`); SKILL.md new Step N+2 + Step 0.7 D7 branch. +111 tests.
+- **Tier 9 (Group F — commit `9e3ef49`)** — D5 self-decision (`src/services/code/mode-gate.ts` + `peaks code should-pause`); D6 context monitor (`src/services/context/main-session-monitor.ts` + `peaks context check`); D7 post-compact resume (`src/services/code/post-compact-detector.ts` + `peaks code post-compact-detect`); SKILL.md new Step N+2 + Step 0.7 D7 branch. +111 tests.
 
 ### Migration
 
@@ -1414,7 +1874,7 @@ Implements the "metering is value" + "10/90 paradigm" alignment: peaks-prd produ
 
 - **ECC envelope validation** assumes the `everything-claude-code` plugin is installed and exposes a `code-review` agent with the `{ passed, violations[], gateAction }` shape. If the plugin is absent, peaks-rd falls back to inline review (5-state detector + `code-review-ecc-degraded-to-inline` TXT note).
 - **D5 hard-floor categories** (irreversible external side effects / auth-credential / multi-day investment) still pause for AskUserQuestion regardless of mode.
-- **D7 post-compact resume** requires re-invoking `/peaks-solo` in fresh context (Option A — no SessionStart hook in v2.11.0; Option B deferred).
+- **D7 post-compact resume** requires re-invoking `/peaks-code` in fresh context (Option A — no SessionStart hook in v2.11.0; Option B deferred).
 - **Migration scope** is text-only — historical `rd/tech-doc.md` files coexist with the new `prd/handoff.md`; prune is a future slice.
 
 ---
@@ -1427,7 +1887,7 @@ Implements the 10/90 paradigm foundation: 10% human / 90% LLM autonomous workflo
 
 ### Features
 
-- **Multi-pass slice decomposition** (`peaks slice decompose --granularity=service|file|both|auto`, slice W2/W3): produces a v2 hierarchical topology (`DecompositionResultV2` with `passes[].slices[]` and `granularity` / `parentSliceId` fields) that supports peaks-solo fan-out RD. v1 envelopes continue to read via `SchemaRouter` (`src/services/slice/schema-router.ts`).
+- **Multi-pass slice decomposition** (`peaks slice decompose --granularity=service|file|both|auto`, slice W2/W3): produces a v2 hierarchical topology (`DecompositionResultV2` with `passes[].slices[]` and `granularity` / `parentSliceId` fields) that supports peaks-code fan-out RD. v1 envelopes continue to read via `SchemaRouter` (`src/services/slice/schema-router.ts`).
 - **LLMArbitrator** (W2 T5): content-hash SHA-256 cache (`<cacheDir>/<hash>.json`), budget cap (`resetArbitratorBudget()`), live/cache/failure-path callId routing.
 - **GranularityDecider** (W2 T6): stop-condition + tie-break for file-vs-service subdivision.
 - **CrossPassEdgeMerger** (W2 T7): static detection (type-shares / fixture-shares / re-exports / import-binding) + LLM fallback; static detection runs UNCONDITIONALLY (W6 fix #1) so it works without an `llmRunner`.
@@ -1437,7 +1897,7 @@ Implements the 10/90 paradigm foundation: 10% human / 90% LLM autonomous workflo
 - **peaks slice plan PickedFileRouter** (W6 CC-β): `parsePickedFile()` helper replaces raw `JSON.parse(...)` for `-picked.json` envelope; new `PICKED_ENVELOPE_INVALID` error code on validation failure.
 - **3 new skills** (W4): `peaks-slice-decompose` (v2 schema references), `peaks-audit` (6-dim reference), `peaks-final-review` (4-dim reference).
 - **5 existing skill updates** (W7 CC-α):
-  - `peaks-solo` — new Step 0.6 (Audit + Goal gate) and Step N+1 (Final Review gate) between Step 0.5 and Step 0.7.
+  - `peaks-code` — new Step 0.6 (Audit + Goal gate) and Step N+1 (Final Review gate) between Step 0.5 and Step 0.7.
   - `peaks-rd` — new references: `reading-v2-slice-results.md` (SchemaRouter dual-read), `writing-handoff-frontmatter.md` (frontmatter fields + canonical path).
   - `peaks-qa` — new reference: `reading-handoff-frontmatter.md` (cross-check `decisions[]` vs `tests/`, `risks[]` vs `tests/unit/security/`).
   - `peaks-prd` — new reference: `prd-for-multi-pass.md` (AC tagging `[pass-1]` / `[pass-2]`).
@@ -1459,7 +1919,7 @@ Implements the 10/90 paradigm foundation: 10% human / 90% LLM autonomous workflo
 ### Bugfixes — W8 / W8-b stabilization (slice `add-slice-topology-multipass`, post-W7 follow-up)
 
 - **W8 CC-α** (commit `56a9d9e`): stabilize 3 pre-existing flaky tests — `tests/unit/cli-program.workflow.test.ts` per-file timeout raised from vitest default 5000ms → 10000ms (`vi.setConfig({ testTimeout: 10000 })`); `tests/unit/dispatch-cli-latency-benchmark.test.ts` 250ms → 300ms threshold (median + min) with description + inline-comment updates for Karpathy #3 honesty.
-- **W8-b CC-α** (commits `e17f868` + `30f9b51`): fix 3 newly-surfaced state-bound pre-existing failures — `tests/unit/package.test.ts` `beforeAll` runs `scripts/sync-version.mjs` so the version-source assertion is deterministic without `pnpm build`; `tests/unit/cli-program.core.test.ts` + `tests/unit/project-commands.test.ts` use `vi.hoisted` + `vi.mock` to isolate from real-project filesystem state (synthetic passing doctor report + `vi.importActual` passthrough with default `doctorReport`/`runbookHealth` injection); `src/shared/version.ts` synced to `2.10.0` (W7-solo T22 missed this; W8-b surfaced it via the `beforeAll` sync-version run).
+- **W8-b CC-α** (commits `e17f868` + `30f9b51`): fix 3 newly-surfaced state-bound pre-existing failures — `tests/unit/package.test.ts` `beforeAll` runs `scripts/sync-version.mjs` so the version-source assertion is deterministic without `pnpm build`; `tests/unit/cli-program.core.test.ts` + `tests/unit/project-commands.test.ts` use `vi.hoisted` + `vi.mock` to isolate from real-project filesystem state (synthetic passing doctor report + `vi.importActual` passthrough with default `doctorReport`/`runbookHealth` injection); `src/shared/version.ts` synced to `2.10.0` (W7-code T22 missed this; W8-b surfaced it via the `beforeAll` sync-version run).
 - Net effect: full suite `3974 / 0 failed / 17 skipped` (was `3974 / 3 failed / 17 skipped` after W7 due to the 3 state-bound failures, and 3974 / 0 / 17 was timing-flaky due to the 3 W8 targets).
 
 ### Risks / gaps carried forward
@@ -1477,7 +1937,7 @@ Implements the 10/90 paradigm foundation: 10% human / 90% LLM autonomous workflo
 
 - **Sub-agent fan-out is mandatory** (slice `2026-06-24-audit-5th-p2`): `preferences.fanout.defaultMode = 'serial'` opt-out removed. When the slice DAG has ≥ 2 leaves at one topological level, the orchestrator MUST use `peaks sub-agent dispatch --from-dag <dag-file> --batch-id <id>`. No preference, env-var, or CLI flag overrides this. `FANOUT_MODES = ['fan-out']` only; legacy `serial` values auto-migrate via `peaks preferences migrate --write`.
 - **4-policy bundle (slice `2026-06-24-efficiency-4p-bundle`)**: (a) default fan-out via `--from-dag`; (b) periodic checkpoint frequency locked at 20 tool calls (no `~` approximation, no `--periodic-every` override); (c) Karpathy reviewer skipped for `config | docs | chore` request types (5-way review otherwise); (d) `swarmSpeculative.maxConcurrent` default bumped 2 → 3.
-- **Test-tool-detection block** (slice `2026-06-24-test-tool-detection-injection`): Every sub-agent dispatched by Solo (`peaks-rd`, `peaks-qa`, `peaks-ui`, `peaks-txt`, `peaks-sc`, `peaks-general-purpose`, single + DAG) receives a hard "Test Tool Detection (mandatory)" block at the top of its prompt: read `package.json#scripts.test`, use project-local runner (`./node_modules/.bin/<runner>` or `pnpm test --`), NEVER `npx <runner>`. Wired at both `dispatch-commands.ts` and `dag-orchestrator.ts` chokepoints; envelope version bumped to `2.2.0`.
+- **Test-tool-detection block** (slice `2026-06-24-test-tool-detection-injection`): Every sub-agent dispatched by Code (`peaks-rd`, `peaks-qa`, `peaks-ui`, `peaks-txt`, `peaks-sc`, `peaks-general-purpose`, single + DAG) receives a hard "Test Tool Detection (mandatory)" block at the top of its prompt: read `package.json#scripts.test`, use project-local runner (`./node_modules/.bin/<runner>` or `pnpm test --`), NEVER `npx <runner>`. Wired at both `dispatch-commands.ts` and `dag-orchestrator.ts` chokepoints; envelope version bumped to `2.2.0`.
 
 ### Bugfixes — handoff path canonicalization (v1 + v2 + v3 + v4 + v5 + v6 + v7)
 
@@ -1544,7 +2004,7 @@ Replaced all 20 with a 4-helper API at `src/services/artifacts/artifact-template
 **Added.** Sub-agent dispatch (both single + DAG paths, all roles rd/qa/ui/txt/sc/general-purpose) now prepends a `## Test Tool Detection (mandatory)` block to every sub-agent prompt. The block tells the sub-agent to read `package.json#scripts.test` first and use the project-local runner (`./node_modules/.bin/<runner>` or `pnpm test -- <file>`) — never `npx <runner>`. Runtime introspection: `peaks test --json`.
 
 - New module: `src/services/dispatch/test-tool-detection.ts` (47 lines; exports `TEST_TOOL_DETECTION_BLOCK` + `formatTestToolDetection()`)
-- Dispatch chokepoints updated: `src/cli/commands/dispatch-commands.ts:187`, `src/services/solo/dag-orchestrator.ts:157,182-183`
+- Dispatch chokepoints updated: `src/cli/commands/dispatch-commands.ts:187`, `src/services/code/dag-orchestrator.ts:157,182-183`
 - Dispatch envelope bumped: `envelopeVersion: '2.1.0'` → `'2.2.0'` (consumers can detect the new prompt shape)
 - New tests: `tests/unit/dispatch/test-tool-detection.test.ts` (6), `tests/unit/dispatch/test-tool-detection-injection.test.ts` (9), `tests/unit/skills/test-tool-detection-docs.test.ts` (4) — 19 new assertions
 - Block size: 749 bytes UTF-8 (≤800 cap)
@@ -1558,9 +2018,9 @@ Replaced all 20 with a 4-helper API at `src/services/artifacts/artifact-template
 ### Changed
 
 - **`FanoutMode` closed set narrowed to `['fan-out']`.** The `'serial'` member is gone from `src/services/preferences/preferences-types.ts`; any preferences.json file carrying `defaultMode = 'serial'` (or any non-fan-out value) now throws `PREFERENCES_FANOUT_INVALID` at load. The `migratePreferences` path silently rewrites legacy `'serial'` → `'fan-out'` for 2.8.3-era files and surfaces the change in the migration envelope's `changes[]`.
-- **SKILL.md contract flipped.** `peaks-solo` SKILL.md now states "Hard constraint: fan-out is mandatory"; the previous "Fan-out opt-out" subsection is removed.
+- **SKILL.md contract flipped.** `peaks-code` SKILL.md now states "Hard constraint: fan-out is mandatory"; the previous "Fan-out opt-out" subsection is removed.
 - **Reference docs reorganized.** `references/fanout-opt-out.md` (escape hatch) → `references/fanout-mandatory.md` (hard constraint + migration contract). `references/swarm-dispatch-contract.md` opt-out callout removed.
-- **Test surface refreshed.** `tests/unit/solo/skills-solo-fanout-opt-out.test.ts` → `tests/unit/solo/skills-solo-fanout-mandatory.test.ts`; pins the new hard constraint and verifies the opt-out file is deleted.
+- **Test surface refreshed.** `tests/unit/code/skills-code-fanout-opt-out.test.ts` → `tests/unit/code/skills-code-fanout-mandatory.test.ts`; pins the new hard constraint and verifies the opt-out file is deleted.
 
 ### Migration
 
@@ -1889,11 +2349,11 @@ are unchanged by this release), `pnpm build` clean.
   `.peaks/_runtime/<sessionId>/dispatch/contracts/<slice-id>.json`. The B/C/D
   dispatch prompts auto-inject A's contract under a `slice A contract:`
   segment so downstream slices see the dependency without re-reading source.
-- **Solo DAG orchestrator with cancel-on-fail (slice 1.2)** —
-  `src/services/solo/dag-orchestrator.ts` exports `runDag(dag, opts)`:
+- **Code DAG orchestrator with cancel-on-fail (slice 1.2)** —
+  `src/services/code/dag-orchestrator.ts` exports `runDag(dag, opts)`:
   topological-layer fan-out, per-layer join barrier, 任一叶子失败整组回退
   (any leaf failure → in-flight sub-agents receive cancel signal → RD
-  returns to repair). peaks-solo SKILL.md references the orchestrator.
+  returns to repair). peaks-code SKILL.md references the orchestrator.
 - **5-IDE `awaitBatch` real implementation (slices 1.2 + 1.3 + 1.4)** —
   `SubAgentDispatcher.awaitBatch(batchId, opts): Promise<BatchResult[]>`
   is now implemented across **all 5 IDEs**: claude-code (1.2 MVP) +
@@ -1953,7 +2413,7 @@ are unchanged by this release), `pnpm build` clean.
 - **`runDag` cancel-on-fail correctness (slice 1.2.c)** — when any leaf
   fails, in-flight sub-agents in the same batch receive a cancel signal
   at the envelope level; the orchestrator no longer waits for them to
-  finish naturally. Pinned by `tests/unit/solo/dag-orchestrator.test.ts`.
+  finish naturally. Pinned by `tests/unit/code/dag-orchestrator.test.ts`.
 
 ### Security
 
@@ -2407,10 +2867,10 @@ are unchanged by this release), `pnpm build` clean.
   skillsActive, todoState, reason, createdAt). Max 10 retained, oldest auto-pruned.
 - **`peaks session resume`** (slice 011) — reads checkpoint JSON, emits structured
   markdown block for skill to prepend on session restart.
-- **peaks-solo Step 0.5** (slice 011) — cross-date session check.
-  IDE-agnostic; lives in `skills/peaks-solo/references/cross-date-session-check.md`.
-- **peaks-solo Step 0.75** (slice 011) — checkpoint resume probe.
-- **peaks-solo Step N** (slice 011) — periodic checkpoint guidance.
+- **peaks-code Step 0.5** (slice 011) — cross-date session check.
+  IDE-agnostic; lives in `skills/peaks-code/references/cross-date-session-check.md`.
+- **peaks-code Step 0.75** (slice 011) — checkpoint resume probe.
+- **peaks-code Step N** (slice 011) — periodic checkpoint guidance.
 
 ### Security
 
@@ -2599,12 +3059,12 @@ Full suite: **2957 passed, 12 skipped, 0 failed**.
   5+ historical slice records exist; v1.1 will switch to percentile
   lookup.
 
-- **`peaks-solo` Step 0.6** — pre-mode-selection slice decomposition.
-  Solo runs the algorithm automatically after Step 0.55 (1.x detection)
+- **`peaks-code` Step 0.6** — pre-mode-selection slice decomposition.
+  Code runs the algorithm automatically after Step 0.55 (1.x detection)
   returns "fresh". The user picks a profile informed by the
   decomposition's parallel structure.
 
-- **3 new `peaks-solo/references/*.md`** — `slice-algorithm.md`
+- **3 new `peaks-code/references/*.md`** — `slice-algorithm.md`
   (algorithm spec), `understand-anything-integration.md` (KG consumer
   contract), `fzf-integration.md` (operator-facing fzf usage).
 
@@ -2618,7 +3078,7 @@ Full suite: **2957 passed, 12 skipped, 0 failed**.
   now derive from `parallelBatches` (with legacy `--type` lookup as
   the fallback path).
 
-- **Extended `peaks-solo/SKILL.md`** — adds "Peaks-Loop Slice
+- **Extended `peaks-code/SKILL.md`** — adds "Peaks-Loop Slice
   Decomposition (Step 0.6 — pre-mode-selection)" section.
 
 - **Extended `runbook.md`** — adds "Step 2.5: Slice Decomposition"
@@ -3092,7 +3552,7 @@ subprocess ignores it when peaks-loop's env vars are present).
   write. Soft-fail per sub-step; never blocks the whole upgrade.
 
 - **`peaks upgrade --detect-1x`** — read-only probe returning a JSON
-  envelope the peaks-solo skill consumes to gate the AskUserQuestion
+  envelope the peaks-code skill consumes to gate the AskUserQuestion
   in Step 0.55.
 
 - **`peaks standards migrate --from-claude-rules`** — thins `.claude/rules/`