Who it is forA team lead whose members use different coding agents on the same repository.
The problemEach person and each agent keeps separate context, so the team gives conflicting instructions and loses decisions.
The outcomeEvery agent in the repo reads the same AGENT_PROTOCOL.md and uses the same committed runtime, so memory stays consistent.
What you need first
- Codex in the ChatGPT desktop app or Codex CLI
- Python 3.9 or newer available as python3
- Git, when installing from GitHub
Steps
- Initialize EchoesVault in the project by sending: Initialize EchoesVault for this project.
- Commit the generated protocol, portable runtime, and adapters with Git.
- Push the branch so teammates receive the EchoesVault folder and .echoes-vault/echoes_vault.py.
- Teammates clone the repo and do not need a global EchoesVault install to use the committed runtime.
- Ask Claude or OpenCode in the project to restore project memory and confirm it reads AGENT_PROTOCOL.md.
Prompt for your agent
In the project at <project-path>, use the committed EchoesVault files from psinetron/echoes-vault-codex to restore project memory with Claude or OpenCode. Expected output: a short summary of the latest session entries and the protocol version. Verify that the agent reads EchoesVault/AGENT_PROTOCOL.md before writing.
Caveat. Teammates must use the same protocol version, and the README says a protocol mismatch should stop writes until the adapter or plugin is updated.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for psinetron/echoes-vault-codex and "Share one project memory across teammates and agents"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer