Intertool Documentation

Intertool gives each engineer a private memory vault and turns confirmed engineering learnings into sourced context another engineer's fresh coding session can retrieve. A team can begin with one repository, then expand the same reviewed workflow across an organisation without giving up ownership, permissions, or evidence.

The private My memory layer can import durable local Codex memory documents. It is scoped to the user and is never returned to the team. The Team memory layer stores reviewed decisions, conventions, warnings, discoveries, ownership, and runbooks—not raw conversations, prompts, source files, or terminal history. Every published team memory has an author and timestamp, may include a source, and remains inside its authenticated organisation.

Read Personal memory for the import and privacy model.

From one team to an enterprise

• Small teams get one dependable place for the decisions, warnings, conventions, and runbooks that coding agents need.
• Growing organisations keep retrieval precise with organisation, repository, path, status, and relevance scopes.
• Enterprise teams retain tenant isolation, owner/admin/member roles, revocable personal tokens, immutable published-state versions, reports, and audited lifecycle actions.

The workflow

1. Codex, Claude, or an engineer proposes a team draft.
2. A human reviews the exact wording and scope.
3. The human explicitly publishes the draft.
4. An authorised MCP client retrieves it when the repository, task, and paths match.
5. Engineers report stale or conflicting context without silently deleting it.

Start locally

pnpm install
cp .env.example .env.local
docker compose up -d postgres
pnpm db:migrate
pnpm db:seed
pnpm dev

Continue with Getting started, then read Publishing memories and the API overview.