Назад към всички

auto-context-manager

// AI-powered automatic project context management. Detects which project the user is referring to and maintains context awareness.

$ git log --oneline --stat
stars:1,933
forks:367
updated:March 4, 2026
SKILL.mdreadonly

Auto Context Manager

AI-powered automatic project context management. Detects which project the user is referring to and maintains context awareness.

Activation

Auto-detect on: Use at session start or when context is unclear. Manual call: When user asks about projects or context switching.

Quick Commands

Run from the skill directory:

# Detect project from message
python acm.py detect "your message here"

# List all projects
python acm.py list

# Get current active project
python acm.py current

# Switch to different project
python acm.py switch <project_id>

Usage in Sessions

When user message contains project keywords:

  1. Run python acm.py detect "<message>" to identify context
  2. Use context to prioritize relevant skills/memory
  3. Respond with project-aware knowledge

Example:

User: "Check my portfolio"
-> Detect: financial/trading project
-> Use relevant financial skills
-> Check trading-related memory files

Configuration

Projects are stored in ~/.auto-context/projects.json and can be customized:

{
  "projects": {
    "my-project": {
      "name": "My Project",
      "description": "Project description",
      "keywords": ["keyword1", "keyword2", "keyword3"]
    }
  },
  "current_project": "default"
}

Files

  • auto_context_manager.py - Core module
  • acm.py - CLI wrapper
  • ~/.auto-context/projects.json - Project config (auto-created)

Adding Projects

from auto_context_manager import AutoContextManager
acm = AutoContextManager()
acm.create_project('project_id', 'Project Name', ['keyword1', 'keyword2'], 'Description')

Integration Notes

  • Fully local, no external APIs
  • Data stored in ~/.auto-context/
  • Confidence score indicates match strength
  • Always returns a result (defaults to "default" project)