Brain plugin

.gitignore

@@ -60,3 +60,6 @@ coverage/
 # Auto-generated workflows (session artifacts)
 workflows/auto/
 workflows/auto/*.yaml
+
+# Index caches (large, not needed in repo)
+.indexes/

agents/backend-api.md

@@ -20,7 +20,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, context7, memory, grep, glob
+**Tools**: read, write, edit, bash, skill, lsp, context7, memory, grep, glob, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # ⚙️ Backend API Developer
 
@@ -63,3 +63,11 @@ You leverage the **`fullstack-dev`** OpenCode skill (found in `skills/fullstack-
 
 > [!NOTE]
 > Coordinate closely with the `Frontend-Engineer` to define API contracts before implementation. Use `lead-architect` for major schema changes.
+
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>

agents/backend-laravel.md

@@ -21,7 +21,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, context7, memory
+**Tools**: read, write, edit, bash, skill, lsp, context7, memory, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # 🎨 Backend Laravel Agent
 
@@ -210,3 +210,11 @@ class BlogPostForm extends Component
 2. **Task Decomposition**: Break into logical units (Migration → Model → Controller → Livewire)
 3. **Subagent Delegation**: Spawn core-builder for file modifications
 4. **Validation**: Use LSP to ensure no new diagnostics
+
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>

agents/backend-tauri.md

@@ -20,7 +20,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, context7, memory
+**Tools**: read, write, edit, bash, skill, lsp, context7, memory, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # backend-tauri
 
@@ -72,3 +72,11 @@ For each task:
 4. Set up proper permissions in `capabilities/`
 5. Test IPC communication end-to-end
 6. Ensure cross-platform compatibility (Windows, macOS, Linux)
+
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>

agents/core-factory.md

@@ -25,7 +25,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, skill, grep, glob, todowrite, memory, context7, sequential-thinking, lsp
+**Tools**: read, write, edit, skill, grep, glob, todowrite, memory, context7, sequential-thinking, lsp, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # Core Factory Agent
 
@@ -77,6 +77,14 @@ permission:
 </step_3>
 </process_flow>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
 - Always think in <thinking> blocks before actions.
 - WORKFLOW: Read-Edit-Validate. Use grep/glob for discovery, edit tool with oldString/newString.

agents/devops-engineer.md

@@ -22,7 +22,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, context7, memory
+**Tools**: read, write, edit, bash, skill, context7, memory, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # DevOps Engineer Agent
 
@@ -74,6 +74,14 @@ permission:
 </step_3>
 </process_flow>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
 - DEVOPS: Terminal execution, MCP management, system tasks.
 - Handle db:init, clean, process:check, backups.

agents/docs-curator.md

@@ -20,7 +20,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, codesearch, websearch, webfetch, todowrite, memory, context7, sequential-thinking
+**Tools**: read, write, edit, bash, skill, lsp, codesearch, websearch, webfetch, todowrite, memory, context7, sequential-thinking, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # Docs Curator Agent
 
@@ -72,6 +72,14 @@ permission:
 </step_3>
 </process_flow>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
 - KNOWLEDGE: Document, self-improve, and evolve the system.
 - Update rules/*.md, agents instructions, and project docs.

agents/frontend-ui-ux.md

@@ -19,7 +19,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, codesearch, task, mcp, memory, context7, sequential-thinking
+**Tools**: read, write, edit, bash, skill, lsp, codesearch, task, mcp, memory, context7, sequential-thinking, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # Frontend UI/UX Specialist
 
@@ -101,6 +101,14 @@ Design system reference (e.g., shadcn/ui, Tailwind UI, custom)
   </route>
 </routing_intelligence>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
   <must>Use only valid tools: read, write, edit, bash, websearch, codesearch, context7_resolve-library-id, context7_query-docs, skill_use</must>
   <must_not>Reference invalid tools like "ui", "ux", "lsp", or "file"</must_not>

agents/lead-architect.md

@@ -19,7 +19,7 @@ permission:
 ---
 
 
-**Tools**: read, write, edit, bash, skill, lsp, codesearch, task, mcp, memory, context7, sequential-thinking
+**Tools**: read, write, edit, bash, skill, lsp, codesearch, task, mcp, memory, context7, sequential-thinking, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # 🏛️ Lead Architect
 
@@ -48,3 +48,11 @@ You are the **Lead Architect**. You are responsible for the overall technical vi
 
 > [!TIP]
 > Prioritize simplicity and maintainability. Avoid "over-engineering" but ensure the foundation is solid.
+
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>

agents/lead-strategist.md

@@ -25,7 +25,7 @@ permission:
 ---
 
 
-**Tools**: skill, bash, read, lsp, codesearch, todowrite, task, memory, context7, sequential-thinking
+**Tools**: skill, bash, read, lsp, codesearch, todowrite, task, memory, context7, sequential-thinking, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # Lead Strategist Agent
 
@@ -85,6 +85,14 @@ permission:
 </step_3>
 </process_flow>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
 - LEAD: Orchestration, architecture, and roadmap. Use <thinking> for strategy.
 - You have access to skills, LSP, codesearch, and multi-agent task delegation.

agents/qa-guardian.md

@@ -19,7 +19,7 @@ permission:
 ---
 
 
-**Tools**: read, bash, skill, lsp, context7, memory, grep, glob
+**Tools**: read, bash, skill, lsp, context7, memory, grep, glob, brain_diagnostic, brain_sidecar_status, brain_status, brain_search, brain_embed_test, brain_index_project
 
 # QA Guardian Agent
 
@@ -79,6 +79,14 @@ permission:
 </step_4>
 </process_flow>
 
+<brain_plugin_workflow>
+- Check Brain health with brain_sidecar_status or brain_diagnostic before non-trivial debugging, feature work, refactors, architecture analysis, or documentation audits.
+- If the index is empty, stale, or missing expected results, run brain_index_project before relying on retrieval.
+- Use brain_search for semantic codebase discovery, then read the top matching files directly before making decisions or edits.
+- Use brain_embed_test when search quality matters or when choosing better query terms for a complex investigation.
+- After broad edits or generated files, confirm Brain can see the new context with brain_status or a targeted brain_search.
+</brain_plugin_workflow>
+
 <constraints>
 - QUALITY: Code review, testing, security scanning, and debugging.
 - Use qa-tester skill, security-scan skill, and debug utilities.