Monitors Claude Code context usage and, when a session nears its limit, writes a handoff summary so work can resume in a fresh session. Adds a statusline indicator and commands to configure the trigger threshold and summary length.
Claude Code
$ npx -y skills add povvo/claudikins-automatic-context-manager --agent claude-code
Codex
$ npx skills add povvo/claudikins-automatic-context-manager -a codex
Cursor
$ npx skills add povvo/claudikins-automatic-context-manager -a cursor
Gemini CLI
$ npx skills add povvo/claudikins-automatic-context-manager -a gemini-cli
GitHub Copilot
$ npx skills add povvo/claudikins-automatic-context-manager -a github-copilot
OpenCode
$ npx skills add povvo/claudikins-automatic-context-manager -a opencode
Windsurf
$ npx skills add povvo/claudikins-automatic-context-manager -a windsurf
v1.1.1-beta - Native hooks implementation. This is beta software - expect rough edges. Please report bugs here.
Automatic context management for Claude Code. Monitors token usage and triggers at 60% context saturation, prompting for session handoff with an auto-generated summary that opens in a fresh session.
Claude Code CLI and Python 3.
LLM attention weights distribute across all context tokens. As context fills, per-token attention decreases and earlier information receives less weight during generation.
UserPromptSubmit hookAskUserQuestion for handoff confirmation.claude/claudikins-acm/handoff-state.json and human-readable handoff.mdclaude commandSessionStart hook detects and loads structured state into fresh context/plugin marketplace add elb-pr/claudikins-marketplace/plugin → navigate to Claudikins Marketplace → installManual
git clone https://github.com/elb-pr/claudikins-automatic-context-manager.git
./install.sh
flowchart TD
A[Statusline monitors context] -->|every 300ms| B{Context >= 60%?}
B -->|No| A
B -->|Yes| C[Create threshold flag file]
C --> D[UserPromptSubmit hook fires]
D --> E[AskUserQuestion prompt]
E --> F{User response}
F -->|YES| G[Capture structured state]
G --> H[Save handoff-state.json + handoff.md]
H --> I[Open new terminal session]
I --> J[SessionStart hook loads state]
F -->|SNOOZE| K[Wait 5 minutes]
K --> A
F -->|DISMISS| L[Disable for session]
Run /acm:config for interactive setup, or edit ~/.claude/claudikins-acm.conf directly.
| Variable | Default | Range | Description |
|---|---|---|---|
THRESHOLD | 60 | 50-90 | Context percentage trigger |
SNOOZE_DURATION | 300 | 60-3600 | Snooze period in seconds |
SUMMARY_TOKENS | 500 | 200-2000 | Max tokens for handoff summary |
Supported terminals and methods
| Platform | Terminal | New Tab Method |
|---|---|---|
| Windows native | Windows Terminal | wt.exe new-tab (Git Bash or PowerShell) |
| Windows native | Git Bash | New window via start git-bash |
| Windows native | PowerShell | New window via Start-Process |
| WSL | Windows Terminal | wt.exe new-tab wsl |
| WSL | Warp | SendKeys* |
| macOS | iTerm2 | osascript |
| macOS | Terminal.app | osascript (Cmd+T) |
| macOS | Warp | osascript |
| Linux | GNOME Terminal | --tab flag |
| Linux | Konsole | --new-tab flag |
| Linux | XFCE Terminal | --tab flag |
| Linux | Kitty | @ launch --type=tab |
| Fallback | Any | Clipboard + instructions |
*Warp on WSL: Uses PowerShell SendKeys automation (focus window, Ctrl+Shift+T, paste, enter). Fragile but functional until Warp adds CLI tab support.
Handoff not triggering Check context has reached threshold. Snooze may be active, or handoff already triggered this session.
New terminal not opening
Platform detection may have failed. Check platforms/ for your terminal. Falls back to clipboard with instructions.
Handoff not loading in new session
Must be in the same project directory. Verify .claude/claudikins-acm/handoff.md exists and SessionStart hook is installed.
/plugin → Installed → select plugin → Uninstall
Stars on GitHub
View @povvo’s profileKeeps Claude Code from charging ahead on its own: work moves through planning, execution, verification and shipping, with a review gate between each step. Changes reach you already checked and approved, not as a rushed refactor.
@povvoA development methodology for coding agents that makes the agent brainstorm before building, follow plans, run parallel work and finish branches in a disciplined way. You get fewer skipped steps and more consistent results from your coding assistant.
@obraHelps you take a change from a rough idea or bug report through to a tested, reviewed result, with each step handled by a focused skill. Its skills cover pinning down domain terms, designing module interfaces, diagnosing hard bugs, and reviewing changes against your repo's coding standards.
@mattpocockGives your coding agent a planning, testing and review routine to follow on every task. It plans before building, checks changes with tests, has a fresh reviewer look for regressions, and saves repeated wins as reusable skills.
@affaan-mDeveloper tools for working on the React codebase itself. Provides skills for extracting error messages, managing and debugging feature flags across release channels, running Flow type checks, and fixing lint and formatting issues before committing.
@facebookMonitors Claude Code context usage and, when a session nears its limit, writes a handoff summary so work can resume in a fresh session. Adds a statusline indicator and commands to configure the trigger threshold and summary length. Install it with the command on its page. It includes acm-config, acm-handoff. Its skills do not fire on their own yet. Free and open source.
Is this plugin yours?
Claim it with GitHubSubmit a pluginPromote it