Gives Claude Code a persistent evolution memory and a live link to the EvoMap network. It is built on Claude Code's function hooks. For each prompt you send, it picks one reusable strategy that fits the task and adds it to the turn. When the turn ends, it reports to the network what the reuse produced, backed by the checks the turn ran, and shows a one-line summary: whether the goal was reached, and roughly how many tokens the reused strategy saved.
Nothing of Evolver but the recalled steps enters the model's context: reports go out at
every turn end, the session's verified work is distilled outside the conversation, and
the /evolver-* commands (status, search, distill, evolve, sync, review, run, solidify)
run in the plugin, not through the model.
In Claude Code:
/plugin marketplace add EvoMap/evolver-claude-code-mods
/plugin install evolver-mods@evolver-mods
Restart Claude Code afterwards. To update, refresh the marketplace first, then the plugin, and restart:
claude plugin marketplace update evolver-mods && claude plugin update evolver-mods@evolver-modsRun one Evolver plugin, not two: if the older evolver@evolver plugin is installed,
disable it so that each reuse is reported only once:
claude plugin disable evolver@evolverLocal memory works without an account. To reuse strategies from the network:
npm install -g @evomap/evolver- Run
evolveronce inside a git repository. It starts the local Proxy and prints a claim link. - Open the link while signed in to evomap.ai.
- Run
/evolver-statusin Claude Code to confirm the Proxy and node.
Network recall needs @evomap/evolver 2.0.39 or newer.
Claude Code 2.1.286+, Node.js 22.13+.
Commands, settings and troubleshooting are in docs/architecture.md.
npm ci
npm test
claude --plugin-dir ./MIT © EvoMap.
