create-claude-workspace 2.1.11 → 2.2.1

package/dist/template/.claude/agents/senior-code-reviewer.md

@@ -112,7 +112,7 @@ Read `.claude/profiles/frontend.md` for the framework-specific review checklist.
 - Flag WARN if existing tests were not updated after modifying the code they cover
 - Do NOT require tests for: pure UI layout components, config files, SCSS, route wiring
 - **Coverage thresholds**: project-wide 80% for statements, branches, functions, and lines. Flag WARN if a new file with business logic has significantly lower coverage than this target.
-- **IMPORTANT**: The orchestrator may include an architect's TESTING section in the review prompt. If it lists files as "no tests needed" with rationale, do NOT flag those files as missing tests. Only flag files that the architect's plan says SHOULD have tests but don't, or files not mentioned in the TESTING section at all that contain business/domain logic.
+- **IMPORTANT**: If the prompt includes an architect's TESTING section listing files as "no tests needed" with rationale, do NOT flag those files as missing tests. Only flag files that SHOULD have tests but don't, or files not mentioned in the TESTING section that contain business/domain logic.
 - **Visual regression**: For frontend tasks that add or change pages/routes, flag WARN if no VRT test (`*.vrt.spec.ts`) exists for the affected pages. Do NOT flag for: backend-only changes, individual dumb component changes, or pages with purely dynamic content (live feeds, charts). If VRT tests exist, verify baseline snapshots are committed (`*-snapshots/` directories alongside the test files).
 
 **Code Duplication (HIGH PRIORITY)**

package/dist/template/.claude/agents/technical-planner.md

@@ -22,7 +22,6 @@ You are a Senior Technical Lead / Solution Architect who translates product requ
 Read these files (all mandatory if they exist):
 - `PRODUCT.md` — product strategy, features, priorities (PRIMARY INPUT)
 - `CLAUDE.md` — architecture, tech stack, patterns, conventions
-- `MEMORY.md` — current progress, decisions, blockers (critical when re-invoked during development)
 - `UX.md` — design system, component specs
 - Any existing plan files (`BROWSER_APP_PLAN.md`, `TODO.md`, `PLAN.md`, `MONETIZATION.md`)
 - `package.json` — current dependencies
@@ -62,6 +61,11 @@ After Phase 0 scaffolding completes, the orchestrator will update CLAUDE.md with
 
 Create `TODO.md` in the project root. Structure:
 
+**Dual-mode output:** If `Task Platform: github` or `Task Platform: gitlab` is set in CLAUDE.md, also create milestones and high-level issues for each phase task on the platform:
+- `Task Platform: github` — use `gh milestone create` for each phase, then `gh issue create` for each task (linked to milestone)
+- `Task Platform: gitlab` — use `glab milestone create` for each phase, then `glab issue create` for each task (linked to milestone)
+- If local mode (no Task Platform field, or `Task Platform: local`), create TODO.md only as described below.
+
 ```markdown
 # TODO.md
 
@@ -110,7 +114,7 @@ Decisions made during planning that affect implementation.
 ### 5. Rules for Task Decomposition
 
 **Task granularity:**
-- Each task = 1 commit-sized unit of work (implementable in one development cycle)
+- Each task = 1 commit-sized unit of work
 - If a task would take multiple commits, split it into subtasks
 - Each task must be independently testable/verifiable
 
@@ -151,7 +155,7 @@ The autonomous loop updates TODO.md after each commit:
 Examples after completion:
 ```
 - [x] **Card domain types** — define Card, CardSet, Price interfaces ✅ a1b2c3d (2026-03-05)
-- [~] ~~**Legacy data migration**~~ — SKIPPED: blocked by missing API credentials, see MEMORY.md
+- [~] ~~**Legacy data migration**~~ — SKIPPED: blocked by missing API credentials
 ```
 
 ### 7. Quality Validation (MANDATORY before handoff)
@@ -175,15 +179,6 @@ After generating TODO.md, validate it:
 - No task has Complexity: L without a note that it may need splitting
 - Phase 0 has all shared infrastructure (types, DI setup, theme, core services)
 
-### 8. Handoff
-
-After creating and validating TODO.md, tell the user:
-"Technical plan created in TODO.md with [N] phases and [M] tasks.
-- Feature coverage: all P0/P1 features mapped
-- Dependency graph: validated (no cycles)
-Start autonomous development with:
-`/ralph-loop:ralph-loop Use orchestrator agent to continue autonomous development according to CLAUDE.md`"
-
 ## Principles
 
 - **Incremental delivery** — each phase must produce something working and testable

package/dist/template/.claude/agents/test-engineer.md

@@ -452,7 +452,7 @@ Individual files with lower coverage are acceptable if justified (e.g., platform
 nx test [PROJECT] --coverage
 ```
 
-**When called by the orchestrator**, always run tests with `--coverage` flag and report coverage summary in your output.
+Always run tests with `--coverage` flag and report coverage summary in your output.
 
 ## Code Testability Guidance
 

package/dist/template/.claude/agents/ui-engineer.md

@@ -98,7 +98,7 @@ Write clean, elegant code that is easy to test and reason about:
 
 ## Output Format
 
-When called for **planning** (STEP 2 of the development cycle), always structure your response as:
+When called for **planning**, always structure your response as:
 
 ### EXISTING CODE
 What already exists in the codebase that is relevant — reusable components, utilities, shared UI pieces. Reference specific file paths. This section is MANDATORY — you must search before planning.
@@ -152,12 +152,12 @@ If you received an API contract from the backend architect and it is impractical
 Only include this section if there are genuine problems — do NOT nitpick if the contract is workable.
 
 ### SPLIT RECOMMENDATION (optional)
-If the task is too large for a single development cycle (would require 8+ files or multiple independent concerns), recommend splitting. List the suggested sub-tasks with their own scope and dependencies. The orchestrator will delegate to technical-planner to update TODO.md.
+If the task is too large (would require 8+ files or multiple independent concerns), recommend splitting. List the suggested sub-tasks with their own scope and dependencies.
 Only include this section if splitting is genuinely needed — do NOT split tasks that are naturally cohesive.
 
 ### VERIFICATION
 How to verify the implementation is correct (build, visual check, test scenarios).
-If the task adds or changes page-level UI, include: "Update VRT baselines if visual output changed intentionally." This signals to the orchestrator that STEP 4 should include VRT baseline updates.
+If the task adds or changes page-level UI, include: "Update VRT baselines if visual output changed intentionally."
 
 ---
 

package/dist/template/.claude/templates/claude-md.md

@@ -38,6 +38,7 @@ When running in autonomous/unattended mode (via `autonomous.mjs` or non-interact
 - **Testing**: Vitest
 - **Deployment**: Cloudflare (Pages + Workers) — default
 - **Workflow**: solo / team (solo = direct merge to main, team = MR/PR review required)
+- **Task Platform**: local | github | gitlab
 
 [If npm publishing (question 6 = yes):]
 - **Distribution**: npm ([registry], [public/restricted])
@@ -454,6 +455,6 @@ claude --agent orchestrator
 
 **Manual** — single iteration (must be run from `claude --agent orchestrator` session):
 ```
-Continue development from where MEMORY.md left off
+Continue development from where TODO.md left off
 ```
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "create-claude-workspace",
-  "version": "2.1.11",
+  "version": "2.2.1",
   "author": "",
   "repository": {
     "type": "git",