Shared memory for AI coding agents
Your team debugs it once.
Every agent remembers.
OpenWolfPack is the workspace where your team's Claude Code, Codex and Gemini agents pool what they learn. One shared memory, searchable by meaning, in every session.
Free to start · No credit card · Bring your own LLM key
Your agents remember, together
Every learning, decision and fix an agent makes lands in one shared workspace instead of dying in someone's local context window.
Search by meaning, not keywords
Semantic search over the team's memory. Ask “when does the API quota reset” and find the note that says “rate limit resets at UTC midnight”.
One key, one bill, your models
Bring your own LLM key — Anthropic, OpenAI, Groq, Mistral or a local endpoint. Encrypted at rest, never shown again, never ours.
Three steps, then it just runs
- 1
Connect your agent
Drop in the owpack connector. Claude Code, Codex and Gemini push what they learn as they work.
- 2
Invite the team
One workspace, roles for owners, admins, members and read-only viewers. Everyone reads the same memory.
- 3
Stop re-solving things
Before an agent guesses, it recalls. The fix your colleague found last Tuesday is already there.
Simple pricing
You pay for the shared workspace. The model bill stays yours — connect your own key and use whichever provider you already trust.
Free
Solo devs trying it out
€0forever
- 1 project
- Up to 3 members
- Full-text search
- Connector, CLI and MCP server
Team
Teams shipping with agents daily
€99per month
- Unlimited members and projects
- Semantic search (bring your own embeddings)
- AI consolidation of duplicate memory
- Roles, invites and an exportable audit log
- Priority support