Claude Code Plugins

Community-maintained marketplace

Feedback

Tracer Issue Tracker

@scottatron/claude-skills
0
0

Manage tasks and dependencies with Tracer CLI. Use for issue tracking, dependency management, finding ready work, and AI agent workflows.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name Tracer Issue Tracker
description Manage tasks and dependencies with Tracer CLI. Use for issue tracking, dependency management, finding ready work, and AI agent workflows.

Overview

Tracer is a blazing-fast, CLI-based issue tracker designed specifically for AI agents. It enables tracking tasks, managing dependencies, and discovering ready work—all from the command line with git-friendly JSONL storage.

Key Features

  • Dependency-aware: Track what blocks what, automatically discover ready work
  • Lightning-fast: ~5ms per operation, built in Rust
  • AI-optimized: JSON output on all commands for programmatic parsing
  • Git-native: JSONL storage syncs via git, clean diffs
  • Distributed: Share work across agents and sessions

Installation

# Quick install
cargo install --git https://github.com/Abil-Shrestha/tracer

# Verify
tracer --version
tr --version  # Both commands work identically

Prerequisites: Rust toolchain

Git Configuration

After initializing Tracer, configure git to track the JSONL file while ignoring the database:

# Add the database to .gitignore (only needs to be done once)
echo ".trace/bd.db" >> .gitignore

# Commit .gitignore if it's a new entry
git add .gitignore
git commit -m "Ignore Tracer database file"

# Always commit the JSONL file to share issues across sessions/agents
git add .trace/issues.jsonl
git commit -m "Update Tracer issues"

Why this matters:

  • .trace/issues.jsonl - Human-readable, git-friendly, should be committed and shared
  • .trace/bd.db - Binary database file, should NOT be committed (regenerated from JSONL)

Quick Start

Initialize a project

tracer init

Learn interactively

tracer learn  # Interactive tutorial

Core Commands

Creating Issues

# Basic issue
tracer create "Fix login bug"

# With priority and type
tracer create "User authentication system" -p 1 -t epic
tracer create "Design database schema" -t task

# Get JSON output (essential for AI agents)
tracer create "New feature" --json

Types: epic, task, bug, feature Priorities: 1 (highest) to 5 (lowest)

Listing Issues

# List all open issues
tracer list

# Filter by status
tracer list --status open
tracer list --status in_progress
tracer list --status closed

# Filter by priority
tracer list --priority 1

# Get JSON output
tracer list --json

Viewing Issue Details

tracer show <id>
tracer show test-42 --json

Updating Issues

# Change status
tracer update <id> --status in_progress
tracer update <id> --status blocked

# Common statuses: open, in_progress, blocked, closed

# Update with JSON output
tracer update test-5 --status in_progress --json

Closing Issues

tracer close <id> --reason "Completed"
tracer close test-10 --reason "Won't fix"

Dependency Management

Adding Dependencies

# "from" depends on "to"
tracer dep add <from> <to> --type blocks

# Example: API depends on schema being done first
tracer dep add test-3 test-2 --type blocks

Dependency Types:

  • blocks: Task A blocks Task B (B can't start until A is done)
  • parent-child: Hierarchical relationship
  • discovered-from: B was discovered while working on A
  • related: General relationship

Viewing Dependencies

# Show dependency tree for an issue
tracer dep tree <id>
tracer dep tree test-5 --json

Finding Work

Ready Work (Unblocked Tasks)

# Show tasks with no blockers
tracer ready

# Limit results
tracer ready --limit 5

# Get JSON output
tracer ready --json

This is crucial for AI agents: Tracer automatically identifies which tasks can be started right now based on dependency chains.

Blocked Tasks

# Show what's currently blocked
tracer blocked
tracer blocked --json

Data Management

Export

# Export to file
tracer export -o backup.jsonl

# Export to stdout
tracer export

Statistics

tracer stats
tracer stats --json

Workflow Examples

1. Building a Feature with Dependencies

# Create epic
tracer create "User authentication system" -t epic

# Create dependent tasks
tracer create "Design database schema" -t task
# → Returns test-1

tracer create "Build login API" -t task
# → Returns test-2

tracer create "Create login UI" -t task
# → Returns test-3

# Link dependencies
tracer dep add test-2 test-1 --type blocks  # API needs schema
tracer dep add test-3 test-2 --type blocks  # UI needs API

# Find what to work on
tracer ready
# → Shows test-1 (unblocked)

# Start work
tracer update test-1 --status in_progress

# Complete and move to next
tracer close test-1 --reason "Schema created"
tracer ready
# → Now shows test-2 (newly unblocked)

2. AI Agent Session Management

# Start of session - see what's ready
tracer ready --json

# Work on task
tracer update test-5 --status in_progress --json

# Discover new work while coding
tracer create "Refactor error handling" -t task --json
tracer dep add new-id test-5 --type discovered-from

# End of session
tracer update test-5 --status in_progress --json
tracer export -o session-backup.jsonl

3. Multi-Agent Collaboration

# First time setup (any agent, only once)
echo ".trace/bd.db" >> .gitignore
git add .gitignore
git commit -m "Ignore Tracer database"

# Agent 1: Creates and claims work
tracer create "Implement feature X" -t task
tracer update test-10 --status in_progress
git add .trace/issues.jsonl
git commit -m "Agent 1: Start feature X"
git push

# Agent 2: Pulls and finds available work
git pull  # Tracer auto-imports changes from JSONL
tracer ready  # test-10 won't show (in progress)
tracer create "Implement feature Y" -t task
tracer update test-11 --status in_progress
git add .trace/issues.jsonl
git commit -m "Agent 2: Start feature Y"
git push

Best Practices for AI Agents

1. Always Use JSON Output

Add --json to commands when you need to parse results programmatically:

tracer ready --json
tracer list --status open --json
tracer show test-5 --json

2. Check Ready Work First

At the start of each session:

tracer ready --json

This ensures you work on unblocked tasks.

3. Track Work Discovered During Coding

When you discover new tasks while working:

tracer create "New subtask" -t task
tracer dep add new-id current-id --type discovered-from

4. Update Status Frequently

Keep the tracker current:

tracer update <id> --status in_progress  # When starting
tracer update <id> --status blocked --json  # When stuck
tracer close <id> --reason "Completed"  # When done

5. Commit Often

Tracer stores everything in .trace/issues.jsonl (commit this, NOT the .trace/bd.db file):

# Make sure .trace/bd.db is in .gitignore first (see Git Configuration section)
git add .trace/issues.jsonl
git commit -m "Update task status"

6. Use Priorities

Help the AI (or yourself) focus:

tracer list --priority 1 --json  # High-priority work

Pro Tips

Shorthand

Use tr instead of tracer for faster typing:

tr create "New task"
tr ready
tr list --json

Auto-Discovery

Tracer auto-discovers the database from any subdirectory (like git):

cd src/components/
tr list  # Works from anywhere in project

Performance

With 10,000 issues:

  • Create: ~5ms
  • List: ~15ms
  • Ready query: ~10ms

Cycle Detection

Tracer prevents dependency cycles:

tracer dep add test-2 test-1 --type blocks
tracer dep add test-1 test-2 --type blocks  # Will fail

Common Patterns

Daily Standup

# What did I work on?
tracer list --status in_progress

# What's ready for today?
tracer ready --limit 5

# What's blocked?
tracer blocked

Project Planning

# Create epic
tracer create "Q4 Feature Release" -t epic -p 1

# Break down into tasks
tracer create "Backend API" -t task
tracer create "Frontend UI" -t task
tracer create "Testing" -t task

# Link dependencies
tracer dep add test-10 test-9 --type blocks
tracer dep add test-11 test-10 --type blocks

Bug Triage

# Create bugs with priority
tracer create "Critical: Auth failing" -t bug -p 1
tracer create "UI alignment off" -t bug -p 3

# List by priority
tracer list --priority 1 --status open

Integration with Claude

When helping users with Tracer:

  1. Suggest creating issues for complex multi-step tasks
  2. Use dependency tracking for tasks with clear ordering
  3. Check tracer ready before suggesting what to work on
  4. Parse JSON output when programmatic access is needed
  5. Remind to commit .trace/issues.jsonl for version control

Example conversation flow:

User: "I need to build a login system"

Claude: Let's break this down with Tracer:

1. Create an epic for the feature:
   tracer create "User login system" -t epic -p 1

2. Create dependent tasks:
   tracer create "Database schema for users" -t task
   tracer create "Authentication API endpoints" -t task
   tracer create "Login UI component" -t task

3. Link dependencies:
   tracer dep add test-2 test-1 --type blocks
   tracer dep add test-3 test-2 --type blocks

4. See what's ready:
   tracer ready
   
This will show you should start with the database schema first.

Resources

Summary

Tracer is purpose-built for AI agents and developers who want a fast, git-friendly way to track work with dependency awareness. Key advantages:

  • Speed: Operations in milliseconds
  • Simplicity: Pure CLI, no server needed
  • Smart: Automatic ready work detection
  • Portable: JSONL files sync via git
  • AI-native: JSON output on every command

Perfect for managing complex projects with multiple dependent tasks, especially when working with AI coding agents across sessions.