</>MCP Agents Market
Plugin

openwolf

by cytostack2.3kTypeScriptUpdated 2026-08-29

Portable project memory across Claude Code, Codex and OpenCode, plus token accounting measured from harness transcripts. Local file I/O, no API calls, no telemetry.

Claude CodeCodexOpenCodeCursorGemini CLI

OpenWolf is a Claude Code plugin that maintains unified project memory across multiple AI coding assistants (Claude Code, Codex, OpenCode) while tracking token consumption directly from session transcripts. It intercepts file reads and command outputs to prevent duplicate context loading, condenses oversized bash output before it enters the agent's context window, and preserves learned conventions and corrections across sessions. The plugin operates entirely through local file I/O without making API calls or sending telemetry, installing lifecycle hooks that govern context usage and measure actual costs.

Key Features

Shared .wolf/ memory directory that persists project context, conventions, and bug fixes across Claude Code, Codex, and OpenCode sessions
Bash output governor that condenses oversized command results (grep, git show, file dumps) before they consume context tokens, preserving full text on disk
Token accounting measured from actual harness transcripts with attribution for cache rebuilds, model switches, and context compaction events
Durable project index enabling sub-1k-token symbol lookup and targeted file reads of specific functions instead of entire files
Automatic context restoration after compaction, re-injecting in-flight state, user rules, and path-scoped instructions that platforms drop
Lifecycle hooks for session start, file reads, bash execution, compaction, and stop events with health monitoring and verification
Local web dashboard showing verified token savings, per-agent breakdown, hook health status, and cache rebuild attribution
Built-in skills for handoff generation, security audits, and framework migration available across all wired agents

Use Cases

  • 01Prevent token waste when switching between Claude Code, Codex, and OpenCode by maintaining a single project knowledge base
  • 02Track and attribute actual API costs per session and per agent using measurements from platform transcripts
  • 03Reduce context bloat from repeated file reads and oversized command outputs (grep floods, git logs) in long coding sessions
  • 04Preserve coding conventions, bug fixes, and project learnings across agent sessions and share them through version control
  • 05Diagnose expensive cache rebuild events and identify whether model switches, expirations, or compaction triggered them
  • 06Run A/B benchmarks comparing agent performance and token usage with and without context optimization

Related Plugins

View more

andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

Plugin216.1k

claude-mem

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

Plugin95k

Understand-Anything

Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.

Plugin84.9k

rtk

CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies

Plugin82.1k

openwolf — FAQ

What is OpenWolf?+

OpenWolf is a Claude Code plugin that creates portable project memory across multiple AI coding assistants (Claude Code, Codex, OpenCode) and measures token usage from actual session transcripts. It operates entirely through local file hooks without making API calls or sending telemetry.

How do I install OpenWolf?+

Install OpenWolf globally via npm with 'npm install -g openwolf', navigate to your project directory, then run 'openwolf init' to detect installed agents and wire the lifecycle hooks. The plugin then works automatically as you use your normal agents.

Which AI coding assistants does OpenWolf work with?+

OpenWolf provides full integration with Claude Code (12 hooks, output governor, skills), core hooks for Codex CLI and OpenCode (session, read, write, compaction), and context-only support for Cursor, Gemini CLI, and Antigravity through shared rules files.

Does OpenWolf require API keys or send data externally?+

No, OpenWolf uses only local file I/O with no API calls, no telemetry, and no external network requests. All hooks are plain Node.js scripts that run locally, and the dashboard binds to 127.0.0.1 with per-project token authentication.

What are the prerequisites for using OpenWolf?+

You need Node.js 20 or higher and at least one supported AI coding agent (Claude Code, Codex, OpenCode, Cursor, Gemini CLI, or Antigravity) installed. It works on macOS, Linux, and Windows; full-text bug search requires Node.js 22.5+ for built-in SQLite.

Is OpenWolf free to use?+

OpenWolf is open source software licensed under AGPL-3.0, available free of charge. The plugin itself has no costs, though it tracks and optimizes token usage for the underlying AI agent APIs you're already paying for.

How do I install openwolf?+

Open the source repository on GitHub and follow its README. openwolf is a plugin — MCP Agents Market links you directly to the official repo.

Is openwolf free?+

openwolf is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches

Developer Tools PluginsAnalytics & Monitoring PluginsKnowledge & Memory PluginsBest MCP ServersBest SkillsBest AgentsTop rated listingsClaude Code toolsBrowse all categories