Pipeline Status

Display which agents have run in the current session and verify pipeline completeness.

What This Command Does

Shows:

• Which agents have executed (researcher, planner, test-master, etc.)
• Agent status (started, completed, failed)
• Execution timestamps and duration
• Tools used by each agent
• Pipeline completeness (are all expected agents present?)

This helps you:

• ✅ Verify workflow - Confirm full SDLC pipeline executed
• ✅ Debug issues - See which agents ran and when
• ✅ Track progress - Monitor autonomous execution in real-time
• ✅ Prove compliance - Show that all required steps were followed

How It Works

Reads the structured pipeline log file for today:

• Location: docs/sessions/{date}-{time}-pipeline.json
• Format: JSON with agent invocations, timestamps, status
• Created by: SubagentStop hook (automatic)
• Updated by: Orchestrator and agents during execution

Usage

Just run the command:

/pipeline-status

Example Output

📊 Agent Pipeline Status (20251103-143022)

Session started: 2025-11-03T14:30:22
Session file: 20251103-143022-pipeline.json

✅ researcher           COMPLETE  14:35:10 (285s) - Found 3 JWT patterns (tools: WebSearch, Grep, Read)
✅ planner              COMPLETE  14:40:25 (315s) - Created implementation plan (tools: Read)
✅ test-master          COMPLETE  14:45:40 (195s) - Wrote 12 tests (all failing) (tools: Write, Bash)
✅ implementer          COMPLETE  14:52:15 (395s) - Implemented JWT auth (tools: Edit, Write, Bash)
✅ reviewer             COMPLETE  14:54:30 (135s) - Quality approved (tools: Read, Grep)
✅ security-auditor     COMPLETE  14:56:10 (100s) - No security issues found (tools: Grep, Bash)
✅ doc-master           COMPLETE  14:57:30 (80s) - Updated README and API docs (tools: Edit)

======================================================================
Pipeline: COMPLETE ✅
Total duration: 25m 25s

When Pipeline is Incomplete

📊 Agent Pipeline Status (20251103-103022)

Session started: 2025-11-03T10:30:22
Session file: 20251103-103022-pipeline.json

✅ researcher           COMPLETE  10:32:15 (115s) - Research findings saved
⏳ planner              RUNNING   10:34:20 - Planning architecture...

======================================================================
Pipeline: INCOMPLETE ⚠️  Missing: test-master, implementer, reviewer, security-auditor, doc-master
Total duration: 4m 58s

No Pipeline File

If no pipeline file exists for today:

📊 Agent Pipeline Status

No pipeline file found for today.

This usually means:
- Agents haven't been invoked yet today
- Feature work was done manually (without /auto-implement)
- Session log directory doesn't exist

To create pipeline logs:
1. Use /auto-implement for feature work
2. Agents will automatically log to docs/sessions/
3. Run /pipeline-status to verify execution

Implementation

This command simply wraps the agent_tracker.py script:

# Show pipeline status
python plugins/autonomous-dev/scripts/agent_tracker.py status

The script:

1. Finds today's most recent pipeline JSON file
2. Parses agent execution data
3. Displays formatted status with timestamps
4. Calculates pipeline completeness
5. Shows total duration

Integration with Autonomous Workflow

The pipeline status is automatically tracked during /auto-implement:

1. Orchestrator invokes agent → Logs "started" to pipeline file
2. Agent executes → Work happens
3. Agent completes → SubagentStop hook logs "completed" with details
4. User checks status → /pipeline-status shows current state
5. Commit happens → PreCommit hook can verify pipeline (optional)

File Format (JSON)

{
  "session_id": "20251103-143022",
  "started": "2025-11-03T14:30:22",
  "agents": [
    {
      "agent": "researcher",
      "status": "completed",
      "started_at": "2025-11-03T14:30:25",
      "completed_at": "2025-11-03T14:35:10",
      "duration_seconds": 285,
      "message": "Found 3 JWT patterns",
      "tools_used": ["WebSearch", "Grep", "Read"]
    },
    {
      "agent": "planner",
      "status": "completed",
      "started_at": "2025-11-03T14:35:15",
      "completed_at": "2025-11-03T14:40:25",
      "duration_seconds": 310,
      "message": "Created implementation plan",
      "tools_used": ["Read"]
    }
  ]
}

Troubleshooting

"No pipeline file found"

Cause: Agents haven't been invoked, or manual implementation was used

Solution:

• Use /auto-implement for feature work
• Agents will automatically create pipeline logs
• Manual code changes don't create logs (by design)

"Agent shows 'started' but never 'completed'"

Cause: Agent invocation failed or was interrupted

Solution:

• Check agent output for errors
• Re-run /auto-implement to complete workflow
• Agent will start fresh (idempotent)

"Tools not showing in output"

Cause: Tool extraction is best-effort from agent output

Solution:

• This is expected - Claude Code doesn't provide tool usage directly
• The hook parses output for tool mentions
• Missing tools don't affect pipeline verification

Related Commands

• /auto-implement - Invoke orchestrator (creates pipeline logs)
• /status - Show PROJECT.md goal progress
• /health-check - Validate all plugin components

Benefits

1. Transparency - See exactly which agents ran and when
2. Debugging - Identify pipeline gaps or failures
3. Compliance - Prove that SDLC steps were followed
4. Metrics - Track agent usage patterns and timing
5. Confidence - Verify autonomous workflow executed correctly

---

This command provides visibility into the autonomous development pipeline, ensuring all SDLC steps were completed.