issue-tracking
npx machina-cli add skill a5c-ai/babysitter/issue-tracking --openclawIssue Tracking
Overview
Track work through Gas Town's bead system: git-backed atomic work units that carry persistent attribution. Each bead is an issue/task with full lifecycle tracking from creation through completion.
When to Use
- Creating new beads for a convoy
- Tracking bead progress and status
- Managing bead dependencies
- Collecting attribution data for agent evaluation
Bead Lifecycle
- Created: Bead defined with scope and acceptance criteria
- Assigned: Bead placed on agent's hook
- In Progress: Agent actively working (GUPP enforced)
- Review: Work complete, awaiting merge review
- Done: Merged and verified (
gt done) - Destroyed: Wisps only - cleaned up after landing
Attribution
All work carries persistent attribution:
- Which agent completed the bead
- Time taken and quality score
- Used for A/B testing agent configurations
- Feeds into agent evaluation and scoring
Tool Use
Used within convoy management and orchestrator processes.
Source
git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/methodologies/gastown/skills/issue-tracking/SKILL.mdView on GitHub Overview
Issue-tracking for Gas Town beads turns tasks into git-backed, atomic work units with persistent attribution. Each bead represents an issue from creation through completion, enabling lifecycle visibility and reliable provenance. It supports convoy management and orchestrator processes by tracking progress, dependencies, and attribution data.
How This Skill Works
Beads are created with defined scope and acceptance criteria and can be assigned to agents via their hook. The system tracks lifecycle stages (Created, Assigned, In Progress with GUPP, Review, Done, Destroyed) and records attribution data such as the completing agent, duration, and quality score. These beads are git-backed and feed into convoy orchestration and evaluation workflows.
When to Use It
- Creating new beads for a convoy
- Tracking bead progress and status
- Managing bead dependencies
- Collecting attribution data for agent evaluation
- Auditing bead lifecycle and reporting
Quick Start
- Step 1: Create a bead with scope and acceptance criteria
- Step 2: Assign the bead to an agent and start tracking progress
- Step 3: Complete, review, merge, and mark as Done (or Destroyed when cleanup is needed)
Best Practices
- Define clear scope and acceptance criteria at bead creation
- Use explicit bead dependencies to model task order
- Enforce the In Progress state with GUPP when work starts
- Move beads through Review to Done with merge verification (`gt done`)
- Capture attribution data (agent, time, quality) for evaluation and A/B testing
Example Use Cases
- Bead created for a convoy with scope and criteria
- Bead assigned to an agent's hook
- Bead in Progress under GUPP enforcement
- Bead reviewed and merged, then marked Done
- Post-landing cleanup and Destroyed (Wisps) state