AgentSkillsCN

Worklog

工作日志

SKILL.md

Agent Work Log

Log completed work activities to a centralized database for tracking and analysis.

When to Use

Use the aw task command to log work when you complete:

  • •Feature implementations: New functionality, components, or capabilities
  • •Bug fixes: Resolved issues or defects
  • •Refactoring: Significant code reorganization or architectural changes
  • •Configuration changes: Infrastructure, deployment, or build setup
  • •Performance improvements: Optimizations that impact system behavior
  • •Documentation: Meaningful additions to docs, READMEs, or guides

When NOT to Log

Do not log:

  • •Reading files or exploring code
  • •Failed attempts or incomplete work
  • •Trivial changes (typos, formatting, single-line edits)
  • •Planning or research activities
  • •Tool invocations without meaningful outcomes

Usage

bash
aw task "description of completed work" --category <category>

Session Tracking: Set CLAUDE_SESSION_ID environment variable at the start of your session to correlate all work from that session.

Categories

Always include a category to help organize work. Choose from:

  • •feature - New functionality or capabilities
  • •bugfix - Fixed defects or issues
  • •refactor - Code restructuring without behavior change
  • •docs - Documentation updates
  • •config - Build, deployment, or infrastructure setup
  • •test - Test additions or improvements
  • •perf - Performance optimizations
  • •infra - Infrastructure or tooling changes
  • •security - Security improvements or fixes

Examples

Good Examples

bash
aw task "Implemented JWT authentication with refresh tokens" --category feature
aw task "Fixed memory leak in WebSocket connection handler" --category bugfix
aw task "Refactored database layer to use connection pooling" --category refactor
aw task "Added comprehensive error handling to API endpoints" --category feature
aw task "Optimized search query performance by 10x with indexing" --category perf
aw task "Updated API documentation with new endpoints" --category docs
aw task "Added unit tests for authentication module" --category test
aw task "Configured CI/CD pipeline for automated deployments" --category config

Bad Examples

bash
# Too vague
aw task "Made some changes"

# Not a completion
aw task "Started working on authentication"

# Too trivial
aw task "Fixed typo in variable name"

# Just exploration
aw task "Read the authentication code"

Best Practices

  1. •Always specify category: Use --category to categorize the work
  2. •Be specific: Include what was done and what feature/area it affects
  3. •Focus on completion: Only log when work is done, not when starting
  4. •One task per log: Don't combine multiple unrelated accomplishments
  5. •Use clear language: Avoid jargon or overly technical details
  6. •Include context: Mention the relevant technology, component, or system

Category Selection Guide

When choosing a category:

  • •feature - Added new API endpoint, created new component, implemented new capability
  • •bugfix - Resolved error, fixed incorrect behavior, patched security issue
  • •refactor - Reorganized code structure, improved code quality, extracted functions
  • •docs - Wrote guides, updated README, added code documentation
  • •config - Modified package.json, updated CI config, changed build settings
  • •test - Added unit tests, integration tests, improved test coverage
  • •perf - Optimized queries, reduced memory usage, improved response time
  • •infra - Set up deployment, configured monitoring, added logging
  • •security - Fixed vulnerabilities, added authentication, improved permissions

Integration

This skill automatically invokes the aw CLI tool, which:

  • •Stores entries in ~/.aw/worklog.db (SQLite database)
  • •Adds timestamps automatically (ISO 8601 format)
  • •Works from any directory (global installation)
  • •Provides instant feedback on successful logging

Technical Details

  • •Command: aw task "<description>" --category <category>
  • •Database: ~/.aw/worklog.db
  • •Schema: work_entries table with:
    • •id - Auto-incrementing primary key
    • •timestamp - ISO 8601 timestamp
    • •task_description - Description of the work
    • •session_id - From CLAUDE_SESSION_ID env var (nullable)
    • •category - Work category (nullable)
    • •project_name - Auto-detected from git repo or directory
    • •git_branch - Auto-detected current branch (nullable)
    • •working_directory - Directory where command was run
    • •created_at - Unix timestamp of database insertion
  • •Output: ✓ Logged: <description> [category] on success
  • •Auto-detection: Project name, git branch, and working directory are collected automatically