FROM bug/705-fallback-searxng-from-exa TO development (#706)

* init

* feat: US-001 - Add exa-py dependency and EXA_API_KEY constant

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: ryaneggz <kre8mymedia@gmail.com>

* feat: US-002 through US-006 - Exa search integration with fallback chain

- Add _normalize_exa_results() for canonical result format
- Add _search_with_exa() async helper with asyncio.to_thread()
- Reorder fallback chain: Exa -> SearXNG -> Tavily
- Fix Tavily normalizer missing engines field
- Add defensive (engines ?? []) guard in SearchEngine.tsx
- Add prd.json and progress.txt for Ralph tracking

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: ryaneggz <kre8mymedia@gmail.com>

* chore: Mark US-002 through US-006 as passing in prd.json

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Signed-off-by: ryaneggz <kre8mymedia@gmail.com>

* Completed ralph loop

---------

Signed-off-by: ryaneggz <kre8mymedia@gmail.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

Author: ryaneggz

.claude/commands/team.md

@@ -1,9 +1,10 @@
-# Rummage: Multi-Agent Code Analysis & Proposal System
+# Team: Multi-Agent Code Analysis & Proposal System
 
 ## Arguments
+
 - `query`: The feature or task description to analyze
 - `subagents`: Number of elite coding agents to spawn (default: 3)
-- `output_dir`: Directory for spec artifacts (default: `.claude/specs/<auto-generated-name>/`)
+- `output_dir`: Directory for spec artifacts (default: `.specs/<auto-generated-name>/`)
 
 **Usage:** `/team query="<feature description>" subagents=<N> output_dir="<path>"`
 
@@ -16,10 +17,11 @@
 Store the contents as `INITIAL_REPORT`. This context MUST be provided to all subagents in Phase 1. Do NOT proceed without completing this step.
 
 ### Output Directory Setup
+
 - IF `$ARGUMENTS.output_dir` is provided, use that path
-- ELSE auto-generate: `.claude/specs/<feature-slug>/` (derive slug from query)
+- ELSE auto-generate: `.specs/<feature-slug>/` (derive slug from query)
 - RUN `mkdir -p $ARGUMENTS.output_dir` to ensure directory exists
-- All artifacts (PROPOSAL_*.md, REVIEW.md, TASKS.md) will be written to this directory
+- All artifacts (PROPOSAL\_\*.md, REVIEW.md, TASKS.md) will be written to this directory
 
 ---
 
@@ -32,71 +34,89 @@ Generate exactly `$ARGUMENTS.subagents` elite coding agent perspectives. Each ag
 For each agent (1 through N), adopt a distinct elite architect persona and execute:
 
 ```markdown
-You are AGENT_<N>: <PERSONA_NAME> — an elite software architect with deep expertise in <SPECIALIZATION>.
+You are AGENT\_<N>: <PERSONA_NAME> — an elite software architect with deep expertise in <SPECIALIZATION>.
 
 ## Your Mission
+
 Analyze the codebase and produce a comprehensive implementation proposal for:
 
 **Feature:** $ARGUMENTS.query
 
 ## Context Provided
+
 <INSERT INITIAL_REPORT HERE>
 
 ## Your Expertise Lens
+
 - <SPECIALIZATION_1>: How you approach this domain
-- <SPECIALIZATION_2>: Your design philosophy  
+- <SPECIALIZATION_2>: Your design philosophy
 - <SPECIALIZATION_3>: Your quality standards
 
 ## Research Phase
+
 1. Examine the codebase structure and patterns
 2. Identify integration points relevant to the feature
 3. Assess existing conventions and architectural decisions
 4. Note potential risks, edge cases, and dependencies
 
 ## Deliverable Requirements
+
 Produce a detailed proposal covering:
 
 ### 1. Executive Summary
+
 Brief overview of your recommended approach (2-3 sentences)
 
 ### 2. Architectural Analysis
+
 - Current state assessment
 - Proposed changes and their rationale
 - Integration points and dependencies
 
 ### 3. Implementation Strategy
+
 - Step-by-step implementation plan
 - File changes required (new files, modifications)
 - Key code patterns to follow
 
 ### 4. Design Decisions
+
 - Trade-offs considered
 - Why this approach over alternatives
 - Alignment with existing codebase patterns
 
 ### 5. Risk Assessment
+
 - Potential pitfalls
 - Edge cases to handle
 - Testing considerations
 
 ### 6. Estimated Complexity
+
 - Scope: [Small | Medium | Large]
 - Risk Level: [Low | Medium | High]
 - Suggested priority order for implementation
 
 ## Output
+
 Write your complete proposal to: `$ARGUMENTS.output_dir/PROPOSAL_<PERSONA_NAME>.md`
 ```
 
-### Suggested Agent Personas (scale to $ARGUMENTS.subagents)
+### Dynamic Persona Generation (REQUIRED)
+
+Generate `$ARGUMENTS.subagents` personas **derived from `$ARGUMENTS.query`**. Store as `AGENT_PERSONAS`.
+
+| Agent | Persona           | Query-Specific Focus                   |
+| ----- | ----------------- | -------------------------------------- |
+| 1-N   | `<DOMAIN>_<ROLE>` | Critical dimension of **this feature** |
+
+**Format:** Compound names only (e.g., `AUTH_SENTINEL`, `CACHE_STRATEGIST`). No generic roles.
 
-| Agent | Persona | Specialization |
-|-------|---------|----------------|
-| 1 | ARCHITECT | System design, scalability, architectural patterns |
-| 2 | CRAFTSMAN | Clean code, maintainability, SOLID principles |
-| 3 | GUARDIAN | Security, error handling, edge cases, testing |
-| 4 | OPTIMIZER | Performance, efficiency, resource management |
-| 5 | INTEGRATOR | APIs, interfaces, system boundaries |
+**Inject per agent:**
+
+```
+You are <PERSONA> — specialist in <FOCUS>. Analyze "$ARGUMENTS.query" through YOUR lens.
+```
 
 ---
 
@@ -108,54 +128,65 @@ After all `$ARGUMENTS.subagents` proposals are generated, convene the AI Council
 You are the ELITE COUNCIL — a panel of distinguished software architects reviewing implementation proposals.
 
 ## Context
+
 **Feature Under Review:** $ARGUMENTS.query
 
 **Initial Project Context:**
 <INSERT INITIAL_REPORT HERE>
 
 ## Proposals to Review
-You have received $ARGUMENTS.subagents independent proposals from `$ARGUMENTS.output_dir/`:
-<LIST ALL PROPOSAL_<NAME>.md FILES>
+
+You have received $ARGUMENTS.subagents independent proposals from `$ARGUMENTS.output*dir/`:
+<LIST ALL PROPOSAL*<NAME>.md FILES>
 
 ## Council Directive
+
 Synthesize all proposals into a unified, actionable implementation plan. Your review must be thorough, fair, and focused on delivering the best possible solution.
 
 ## Review Framework
 
 ### 1. Proposal Comparison Matrix
-| Aspect | Agent 1 | Agent 2 | Agent N | Council Verdict |
-|--------|---------|---------|---------|-----------------|
-| Architecture | | | | |
-| Maintainability | | | | |
-| Risk Level | | | | |
-| Completeness | | | | |
+
+| Aspect          | Agent 1 | Agent 2 | Agent N | Council Verdict |
+| --------------- | ------- | ------- | ------- | --------------- |
+| Architecture    |         |         |         |                 |
+| Maintainability |         |         |         |                 |
+| Risk Level      |         |         |         |                 |
+| Completeness    |         |         |         |                 |
 
 ### 2. Consensus Points
+
 - What do all/most proposals agree on?
 - Which patterns are universally recommended?
 
 ### 3. Divergence Analysis
+
 - Where do proposals conflict?
 - What are the trade-offs between approaches?
 - Council's reasoned decision on each conflict
 
 ### 4. Unified Implementation Plan
+
 Synthesize the best elements into a single coherent plan:
+
 - Recommended architecture
 - Implementation sequence
 - Critical path items
 - Non-negotiable requirements
 
 ### 5. Risk Consolidation
+
 - Combined risk assessment from all perspectives
 - Mitigation strategies
 
 ### 6. Final Verdict
+
 - GO / NO-GO / CONDITIONAL recommendation
 - Required conditions if CONDITIONAL
 - Confidence level: [High | Medium | Low]
 
 ## Output
+
 Write the complete council review to: `$ARGUMENTS.output_dir/REVIEW.md`
 ```
 
@@ -165,15 +196,18 @@ Write the complete council review to: `$ARGUMENTS.output_dir/REVIEW.md`
 
 After council review, generate the implementation contract:
 
-```markdown
+````markdown
 You are the TASK ARCHITECT — responsible for converting the council's unified plan into an actionable task checklist.
 
 ## Input
+
 - Council Review: `$ARGUMENTS.output_dir/REVIEW.md`
 - Original Feature: $ARGUMENTS.query
 
 ## Directive
+
 Create a precise, atomic task list that serves as the binding contract for implementation. Each task must be:
+
 - **Binary**: Clearly done or not done
 - **Atomic**: Single responsibility
 - **Ordered**: Correct dependency sequence
@@ -185,47 +219,57 @@ Create a precise, atomic task list that serves as the binding contract for imple
 # Implementation Tasks: <Feature Title>
 
 ## Pre-Implementation
+
 - [ ] Verify development environment setup
 - [ ] Create feature branch: `feature/<short-name>`
 - [ ] Review REVIEW.md council decisions
 
 ## Core Implementation
+
 - [ ] Task 1: <specific action>
-  - Files: `path/to/file.ext`
-  - Acceptance: <how to verify>
+    - Files: `path/to/file.ext`
+    - Acceptance: <how to verify>
 - [ ] Task 2: <specific action>
-  ...
+      ...
 
 ## Integration
+
 - [ ] Wire up components
 - [ ] Update configuration files
 - [ ] Add necessary imports/exports
 
 ## Testing
+
 - [ ] Unit tests for <component>
 - [ ] Integration tests for <flow>
 - [ ] Edge case coverage per REVIEW.md
 
 ## Documentation
+
 - [ ] Update relevant documentation
 - [ ] Add inline code comments for complex logic
 - [ ] Update CHANGELOG if applicable
 
 ## Verification
+
 - [ ] All tests passing
 - [ ] Linting/formatting clean
 - [ ] Self-review against REVIEW.md
 - [ ] Ready for PR
 
 ## Completion Signature
+
 - Total Tasks: <N>
 - Estimated Effort: <T>
 - Dependencies: <list>
 ```
+````
 
 ## Output
+
 Write the complete task contract to: `$ARGUMENTS.output_dir/TASKS.md`
-```
+
+````
 
 ---
 
@@ -258,15 +302,18 @@ After each task:
 ## Progress Log
 - [x] Task N: Completed <timestamp>
   - Notes: <any relevant observations>
-```
+````
 
 ## Completion
+
 When all tasks are checked:
+
 1. Final verification against `$ARGUMENTS.output_dir/REVIEW.md`
 2. Run full test suite
 3. Update `$ARGUMENTS.output_dir/TASKS.md` with completion status
 4. Report ready for review
-```
+
+````
 
 ---
 
@@ -306,17 +353,17 @@ Generate a brief validation summary:
 
 ## Output
 Append validation results to: `$ARGUMENTS.output_dir/TASKS.md` under `## Validation Results`
-```
+````
 
 ---
 
 ## Quick Reference
 
-| Phase | Input | Output | Key Action |
-|-------|-------|--------|------------|
-| 0 | — | INITIAL_REPORT + output_dir setup | Read CLAUDE.md, create output dir |
-| 1 | INITIAL_REPORT + query | `output_dir/PROPOSAL_*.md` | Generate N agent proposals |
-| 2 | All proposals | `output_dir/REVIEW.md` | Council synthesis |
-| 3 | REVIEW.md | `output_dir/TASKS.md` | Create task contract |
-| 4 | TASKS.md | Code changes | Implement features |
-| 5 | Implementation | Validation in TASKS.md | Verify & document |
+| Phase | Input                  | Output                            | Key Action                        |
+| ----- | ---------------------- | --------------------------------- | --------------------------------- |
+| 0     | —                      | INITIAL_REPORT + output_dir setup | Read CLAUDE.md, create output dir |
+| 1     | INITIAL_REPORT + query | `output_dir/PROPOSAL_*.md`        | Generate N agent proposals        |
+| 2     | All proposals          | `output_dir/REVIEW.md`            | Council synthesis                 |
+| 3     | REVIEW.md              | `output_dir/TASKS.md`             | Create task contract              |
+| 4     | TASKS.md               | Code changes                      | Implement features                |
+| 5     | Implementation         | Validation in TASKS.md            | Verify & document                 |