nurosys-agents 2.0.0

package/.agent/frontend/workflows/build-feature-react.workflow.md

@@ -0,0 +1,82 @@
+---
+name: build-feature
+description: Strict end-to-end feature workflow with mandatory plan approval, implementation, quality review, and testing.
+strict_mode: true
+owner: agent
+references:
+  - project-memory/constitution.md
+  - project-memory/repo-map.md
+  - project-memory/auth-model.md
+  - project-memory/quality-playbook.md
+  - project-memory/core-memory.md
+  - .agent/skills/feature-workflow/SKILL.md
+  - .agent/skills/auth-and-permissions/SKILL.md
+  - .agent/skills/react-quality-review/SKILL.md
+  - .agent/templates/FEATURE_PLAN.md
+  - .agent/templates/REVIEW_REPORT.md
+  - .agent/templates/TEST_PLAN.md
+outputs:
+  - Documentation/plans/FEATURE_PLAN.md (or Documentation/features/<feature_name>/ when running a named feature)
+  - Documentation/reports/REVIEW_REPORT.md
+  - Documentation/plans/TEST_PLAN.md (or Documentation/features/<feature_name>/ when running a named feature)
+---
+
+# Build Feature Workflow (Strict)
+
+## Non-negotiable gate
+
+- Always load and follow `project-memory/constitution.md`.
+- Always read `project-memory/quality-playbook.md`.
+- Always read and maintain `project-memory/core-memory.md`.
+- **No coding before plan approval.**
+- Required artifacts:
+  - Feature plan: `Documentation/features/<feature_name>/FEATURE_PLAN.md` or `Documentation/plans/FEATURE_PLAN.md`
+  - Review report: `Documentation/reports/REVIEW_REPORT.md`
+  - Test plan: `Documentation/features/<feature_name>/TEST_PLAN.md` or `Documentation/plans/TEST_PLAN.md`
+
+## Steps
+
+1. **Load foundation docs**
+   - Read `project-memory/constitution.md`, `project-memory/repo-map.md`, and `project-memory/auth-model.md`.
+   - Confirm constraints relevant to the request.
+
+2. **Run feature workflow skill and create plan**
+   - Use `.agent/skills/feature-workflow/SKILL.md`.
+   - Generate the feature plan in `Documentation/features/<feature_name>/FEATURE_PLAN.md` or `Documentation/plans/FEATURE_PLAN.md` from `.agent/templates/FEATURE_PLAN.md`.
+
+3. **Approval checkpoint (hard stop)**
+   - Present plan and stop for explicit approval.
+   - Do not modify implementation files before approval.
+
+4. **Implement after approval**
+   - Execute the approved file touch list in order.
+   - Keep pages thin and move logic into `src/views/**`, hooks, and services.
+
+5. **Conditional auth-and-permissions pass**
+   - Run `.agent/skills/auth-and-permissions/SKILL.md` when touching:
+     - `src/contexts/authContext.tsx`
+     - `src/utils/auth.ts`
+     - `src/configs/authConfig.ts`
+     - `src/types/auth.ts`
+     - protected pages under `src/app/[lang]/(dashboard)/(private)/**`
+     - auth-sensitive API routes under `src/app/api/**`
+
+6. **Run quality review and produce report**
+   - Run `.agent/skills/react-quality-review/SKILL.md`.
+   - Create `Documentation/reports/REVIEW_REPORT.md` from `.agent/templates/REVIEW_REPORT.md`.
+   - Apply required refactors from findings.
+
+7. **Generate and execute test plan**
+   - Create the test plan in `Documentation/features/<feature_name>/TEST_PLAN.md` or `Documentation/plans/TEST_PLAN.md` from `.agent/templates/TEST_PLAN.md`.
+   - Add/update relevant tests (unit/integration/manual).
+
+8. **Run test commands**
+   - Run available scripts (`npm run test`, `npm run lint`, feature-specific tests).
+   - Record command outcomes in the test plan file (Documentation/plans or Documentation/features/<feature_name>/).
+
+9. **Update core memory**
+   - Append the completed module or feature summary to `project-memory/core-memory.md`.
+   - Include files changed, decisions made, tests run, and follow-up notes for future work.
+
+10. **Final output**
+   - Return summary of files changed, verification steps, and test results.

package/.agent/frontend/workflows/feature-module-runner.md

@@ -0,0 +1,101 @@
+---
+description: Executes a SINGLE module from a feature folder. Implements, tests, reviews, fixes, and updates docs for that one module only. Invoked per-module by the orchestrator script (scripts/run-feature-modules.sh).
+---
+
+You are a single-module runner. You execute **exactly one module** of a multi-module feature. The orchestrator script (`scripts/run-feature-modules.sh`) handles sequencing — your job is the full lifecycle for the module number you are given.
+
+**Orchestrator gate:** After each module’s Cursor agent session completes successfully, the script runs `npm run build` and `npm run test` before launching the next module. If that gate fails, fix the tree and resume from the same module number. Each agent invocation uses `--model auto` (do not change the script to a fixed model for those sessions).
+
+## Input (provided in prompt)
+
+- **Feature folder** — e.g. `documentation/features/categories_and_amazon_catalog`
+- **Feature name** — derived from folder, e.g. `categories_and_amazon_catalog`
+- **Module number** — e.g. `1`
+- **Module file** — path to the module prompt, e.g. `documentation/features/.../..._MODULE_1_MODEL_LAYER.md`
+
+---
+
+## Setup — Read foundational docs
+
+Read these files first. They provide the rules and patterns you must follow.
+
+| File | Why |
+|------|-----|
+| `project-memory/constitution.md` | Non-negotiable safety rules |
+| `project-memory/quality-playbook.md` | Review checklist categories |
+| `project-memory/core-memory.md` | Feature history and prior decisions |
+| `.agent/workflows/build-feature-backend.workflow.md` | Implementation lifecycle phases |
+| `.agent/workflows/add-new-api-feature-module.md` | Module placement rules |
+
+---
+
+## Step A — Implement
+
+1. **Read the module prompt file** (the path is given in your input).
+2. The prompt contains curated repo-map context and the full module specification. Follow it.
+3. **Implement.** The master plan was already user-approved. Proceed directly without stopping for approval.
+4. Follow `add-new-api-feature-module.md` for module placement and Do/Don't rules.
+
+## Step B — Test & Review (Auto-Approve)
+
+Run tests and quality review. **Auto-approve all internal decisions** — no human gates.
+
+1. **Run tests**
+   - Scope to the feature area if possible (e.g. `npm run test -- --testPathPattern=<feature-name>`).
+   - Otherwise run `npm run test`.
+   - Fix all failures before proceeding.
+
+2. **Code review** — invoke `/react-quality-review` skill
+   - The skill audits React components and patterns against quality standards.
+   - Auto-approve its findings and fix any required issues identified.
+   - No human intervention — this is part of the module's autonomous workflow.
+
+3. **Fix and re-verify**
+   - If review found issues or tests failed, implement fixes.
+   - Re-run tests. If fixes touched reviewed code, re-check via `/react-quality-review` again.
+   - Repeat until tests pass and all required findings are resolved.
+   - If a fix cannot be resolved after two attempts, exit with failure — the orchestrator will pause.
+
+## Step C — Commit & Push
+
+Once all tests pass and review is clean:
+
+1. **Create or update MODULE.md**
+   - In the feature module directory (`src/components/<feature-name>/MODULE.md` or similar), using `.agent/templates/MODULE.md`.
+
+2. **Update core-memory**
+   - Update `project-memory/core-memory.md` feature progress table:
+     - `<feature_name>: Module <N> completed (<brief description>); next: Module <N+1>.`
+
+3. **Commit the module**
+   - `git add .` (all changes from this module)
+   - `git commit -m "feat(MODULE_<N>): <brief description of what this module adds>"`
+   - Example: `git commit -m "feat(MODULE_2): Add PDF export dialog with queue management"`
+
+4. **Push to remote**
+   - `git push origin HEAD` (push to your current branch)
+   - If pushing to a protected branch, ensure your branch is up to date first
+
+5. **Update repo-map** (if module added new routes/components/services)
+   - Add/update the module entry in `project-memory/repo-map.md` (relevant sections).
+   - If a new reusable component, hook, or service was introduced, add it to the Reusable Components Registry.
+
+---
+
+## Output
+
+When done, print a summary with:
+
+- What was implemented (key files created/modified)
+- Test results (pass/fail, coverage)
+- Review outcomes (findings, fixes applied)
+- Any remaining issues or deferred work
+
+---
+
+## Do not
+
+- Implement more than one module. You handle exactly the module number given.
+- Skip Step B (verify). Always test, review, fix, and update docs.
+- Stop for human confirmation.
+- Create unnecessary files outside the scope of this module.