Flow Swarm

OpenClaw → exec (background) → Claude Code
                                    ↓
                    MCP Server starts (autoStart: true)
                    150+ tools available via @claude-flow/cli
                                    ↓
                    SWARM MODE prompt → swarm_init tool called
                    agent_spawn × N → task_create → execute
                                    ↓
                    memory_store (findings) → task_complete
                                    ↓
                              Output + persisted state

ruflo --version    # 3.5.x+
claude --version   # Claude Code CLI

# Full setup: install RuFlo + register MCP + init project
./scripts/setup-flow-swarm.sh /path/to/project

# Verify
./scripts/setup-flow-swarm.sh --verify /path/to/project

npm install -g ruflo@latest
claude mcp add ruflo -- npx -y ruflo@latest mcp start
cd /path/to/project && ruflo init && ruflo memory init && ruflo daemon start

# Check current state
python3 -c "import json; d=json.load(open('.mcp.json')); print('autoStart:', d['mcpServers']['claude-flow'].get('autoStart'))"

# Enable (REQUIRED for full swarm functionality)
python3 -c "
import json
with open('.mcp.json') as f: d = json.load(f)
d['mcpServers']['claude-flow']['autoStart'] = True
with open('.mcp.json', 'w') as f: json.dump(d, f, indent=2)
print('MCP autoStart enabled')
"

# List untested modules by size (biggest = best target)
for f in lib/**/*.ex; do
  base=$(basename "$f" .ex)
  count=$(find test/ -name "${base}_test.exs" 2>/dev/null | wc -l | tr -d ' ')
  [ "$count" = "0" ] && echo "$(wc -l < "$f")L $f"
done | sort -rn | head -10

# Pre-flight: scan public API
grep -n "^  def " lib/your_module.ex
# Pre-flight: check existing test patterns
head -30 test/some_existing_test.exs

# Launch (ALWAYS background, NEVER nohup)
exec(
  command='cd /project && claude --permission-mode bypassPermissions --mcp-config .mcp.json --print "SWARM MODE: ... TASK: ..."',
  background=True,
  timeout=300
)

# Poll for completion
process(action="poll", sessionId="xxx", timeout=120000)

# Verify the output actually compiles/passes
mix test test/path/to/new_test.exs

SWARM MODE: Initialize hierarchical swarm with MCP tools.

COORDINATION:
1. Call swarm_init with topology "hierarchical", maxAgents 4, strategy "specialized"
2. Call agent_spawn for: architect (analyze module), coder (write tests), reviewer (verify)
3. Call task_create for the test generation task
4. After completion: call memory_store with key findings and task_complete

TASK: Write comprehensive ExUnit tests for [MODULE_PATH] ([LINE_COUNT] lines, [DESCRIPTION]).

Public API:
[PASTE grep -n "^  def " output here]

Key data to validate:
- [List specific assertions: required struct keys, value ranges, URL formats, etc.]
- [List known edge cases: unknown inputs, nil, empty string, integer where string expected]

Requirements:
- File: test/[matching_path]_test.exs
- Use async: true (pure functions, no state)
- Group tests by function (describe blocks)
- Test ALL variants, not just a sample (e.g., all 8 hotels, not just 2)
- Include edge cases: nil input, empty string, unknown keys
- Do NOT modify any source files

When done: call memory_store with test count + key findings, then output results.

SWARM MODE: Initialize hierarchical swarm with MCP tools.

COORDINATION:
1. Call swarm_init with topology "hierarchical", maxAgents 4, strategy "specialized"
2. Call agent_spawn for: architect (analyze GenServer behavior), coder (write tests), reviewer (verify)
3. Call task_create for the test generation task
4. After each iteration: call memory_store with what failed and why
5. After completion: call task_complete with final results

TASK: Write comprehensive ExUnit tests for [MODULE_PATH] ([LINE_COUNT] lines, GenServer).

Public API:
[PASTE grep output]

CRITICAL — Test Isolation for GenServers:
- The module registers as a named process (__MODULE__). It's already supervised globally.
- Do NOT use start_supervised! — it conflicts with the app-supervised instance.
- Pattern: stop the global instance, restart with test config, re-stop at end.
- OR: if start_link accepts a name: option, use unique names per test.
- async: false for GenServer tests that touch global state.

Requirements:
- Test GenServer lifecycle (init, handle_call, handle_cast, handle_info)
- Test crash recovery: missing catch-all handlers are REAL BUGS worth flagging
- Test state transitions and side effects
- Do NOT modify any source files

SWARM MODE: Initialize hierarchical swarm (maxAgents 4, strategy specialized).
Spawn: architect (analyze deps + plan mocks), coder (write tests), reviewer (verify coverage).

TASK: Write comprehensive ExUnit tests for [MODULE_PATH].

This module depends on: [LIST DEPENDENCIES]
Mock strategy: [Mox / manual mock / test config override]
Reference existing mocks in test/support/ if any.

Requirements:
- Mock all external calls (HTTP, DB, external services)
- Test happy path AND error paths (timeouts, 4xx, 5xx, malformed responses)
- async: true if using Mox with allowances
- Do NOT modify source files

SWARM MODE: Initialize hierarchical swarm (maxAgents 6, strategy specialized).
Spawn: architect (plan structure), coder (implement), tester (tests), reviewer (quality).
Architect plans FIRST. Coder implements. Tester validates. Reviewer catches issues.

TASK: [Feature description with clear acceptance criteria]

Architecture constraints:
- [List patterns to follow from existing codebase]
- [List modules/files to reference for conventions]

HARD LIMIT: Maximum 5 iterations if quality loop needed.

SWARM MODE: Initialize anti-drift hierarchical swarm (maxAgents 4).
Spawn: architect (plan + checkpoint), coder (execute), reviewer (validate each step).

ANTI-DRIFT RULES:
- Architect creates numbered plan before ANY code is written
- Coder implements ONE step at a time
- Reviewer validates EACH step before proceeding
- If reviewer rejects twice: STOP and report
- Checkpoint state after each successful step

TASK: [Refactor description]
HARD LIMIT: Maximum 8 iterations.

SWARM MODE: Security-focused hierarchical swarm (maxAgents 5, strategy specialized).
Spawn: security-architect (threat model), auditor (scan), coder (fix), tester (verify).

TASK: Security audit of [scope].

Checklist:
- [ ] Dependency vulnerabilities (mix audit / npm audit)
- [ ] Hardcoded secrets in source
- [ ] Injection vectors (SQL, XSS, command)
- [ ] Auth/authz bypass paths
- [ ] GenServer catch-all handlers (handle_info, handle_cast) — these are REAL BUGS
- [ ] Error messages leaking internal state
- [ ] Rate limiting gaps
- [ ] CORS/CSP headers

Output: findings table with severity, file, line, fix.
HARD LIMIT: Maximum 5 iterations.

SWARM MODE: Initialize ring swarm (maxAgents 4, strategy adaptive).
Spawn: coder, tester, reviewer, coordinator.
HARD LIMIT: Maximum 10 iterations.

TASK: Iterate on [target] until [score threshold].

Per iteration:
1. Coder fixes based on reviewer feedback
2. Tester runs full suite, reports pass/fail count
3. Reviewer scores against rubric
4. Score >= threshold → STOP, report final score
5. Iteration == 10 → STOP regardless, report score and remaining gaps