compound-agent 1.6.0 → 1.6.1

package/CHANGELOG.md

@@ -9,6 +9,32 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [1.6.1] - 2026-03-05
+
+### Changed
+
+- **Renamed brainstorm phase to spec-dev**: The `/compound:brainstorm` slash command is now `/compound:spec-dev`. The phase focuses on structured specification development using EARS notation, Mermaid diagrams, and Socratic dialogue rather than open-ended brainstorming. Old `brainstorm.md` command files are auto-cleaned during upgrade.
+- **Integration test stability**: Reduced integration test parallelism (`maxForks: 1`) and increased timeouts to 60s to eliminate non-deterministic ETIMEDOUT failures under load.
+
+### Added
+
+- **Spec reference file**: `.claude/skills/compound/spec-dev/references/spec-guide.md` provides quick-reference material for EARS patterns, Mermaid diagram selection, NL ambiguity detection, and trade-off documentation frameworks. Installed automatically during `ca setup`.
+- **Hook error visibility**: Hook runners now log errors to stderr when `CA_DEBUG` environment variable is set, instead of silently swallowing all failures.
+- **check-plan stdin safety**: `ca check-plan` now enforces a 30-second timeout and 1MB size limit when reading from stdin, preventing hangs in CI/CD environments.
+- **Embed lock expiry**: Embedding lock files now expire after 1 hour as a safety valve against zombie processes holding locks indefinitely.
+- **Phase-state backward compatibility**: Legacy `lfg_active` field in phase state files is automatically migrated to `cookit_active` on read.
+- **clean-lessons scope messaging**: `ca clean-lessons` now reports when non-lesson items are excluded from analysis.
+
+### Fixed
+
+- **Missing spec-guide.md**: The reference file was declared in skill templates and CHANGELOG but never generated during setup. Now installed alongside phase skills.
+- **Upgrade cleanup for lfg.md**: Added `lfg.md` to deprecated commands list so `ca setup --update` removes stale lfg command files from upgraded repos.
+- **Docs template terminology**: WORKFLOW.md template now uses "Spec Dev" instead of "Brainstorm" for Phase 1.
+- **Test file naming**: Renamed `brainstorm-phase.test.ts` to `spec-dev-phase.test.ts` to match the refactored phase name.
+- **Library bundle cleanup**: Moved CLI-only re-exports (`registerWatchCommand`, `registerLoopCommands`) out of the library barrel to eliminate unused import warnings in `dist/index.js`.
+- **plan.test.ts embedding guard**: Added `skipIf(skipEmbedding)` to unguarded test that calls `retrieveForPlan` without mocking.
+- **Agent template test count**: Updated setup.test.ts to expect 9 agent templates (was 8), matching the actual AGENT_TEMPLATES count after `lessons-reviewer.md` was added.
+
 ## [1.6.0] - 2026-03-02
 
 ### Added

package/README.md

@@ -19,7 +19,7 @@ AI coding agents forget everything between sessions. Compound Agent fixes this w
 
 ```mermaid
 graph LR
-    B[BRAINSTORM] --> P[PLAN]
+    B[SPEC DEV] --> P[PLAN]
     P --> W[WORK]
     W --> R[REVIEW]
     R --> C[COMPOUND]
@@ -64,7 +64,7 @@ Preferences survive across sessions and projects. Once captured, they appear at
 `ca list` shows all captured knowledge. `ca stats` shows health. `ca wrong <id>` invalidates bad lessons. Everything is git-tracked JSONL -- you can read, diff, and audit it.
 
 **"I want structured phases, not just 'go build this'."**
-Five workflow phases (brainstorm, plan, work, review, compound) with mandatory gates between them. Each phase searches memory and docs for relevant context before starting.
+Five workflow phases (spec-dev, plan, work, review, compound) with mandatory gates between them. Each phase searches memory and docs for relevant context before starting.
 
 **"My agent doesn't read the project docs before making decisions."**
 `ca knowledge "auth flow"` runs hybrid search (vector + keyword) over your indexed docs. Agents query it automatically during planning -- ADRs, specs, and standards surface before code gets written.
@@ -109,7 +109,7 @@ Then run `pnpm install`.
 The five-phase workflow:
 
 ```
-1. /compound:brainstorm  -->  Explore the problem, clarify scope
+1. /compound:spec-dev    -->  Develop precise specifications
 2. /compound:plan        -->  Create tasks enriched by memory search
 3. /compound:work        -->  Execute with agent teams + TDD
 4. /compound:review      -->  Multi-agent review with inter-communication
@@ -119,7 +119,7 @@ The five-phase workflow:
 Or run all phases sequentially:
 
 ```
-/compound:lfg "Add auth to API"
+/compound:cook-it "Add auth to API"
 ```
 
 Each phase searches memory for relevant past knowledge and injects it into agent context. The compound phase captures new knowledge, closing the loop.
@@ -166,7 +166,7 @@ The CLI binary is `ca` (alias: `compound-agent`).
 | `ca import <file>` | Import items from JSONL file |
 | `ca prime` | Load workflow context (used by hooks) |
 | `ca verify-gates <epic-id>` | Verify review + compound tasks exist and are closed |
-| `ca phase-check` | Manage LFG phase state (init/status/clean/gate) |
+| `ca phase-check` | Manage cook-it phase state (init/status/clean/gate) |
 | `ca audit` | Run audit checks against the codebase |
 | `ca rules check` | Run repository-defined rule checks |
 | `ca test-summary` | Run tests and output a compact summary |