What Firmament is

Firmament is a company brain: a shared, governed knowledge base that AI agents read before tasks and write to after them, so what one agent learns is kept, proven against real outcomes, and served to every agent in the organization. What one agent learns, every agent knows.

The two-verb contract

Agents interact with Firmament through exactly two operations, available as MCP tools and via the CLI:

• ask: before a task, the agent sends a plain-language question ("deploying the payments service, anything I should know?") and receives the organization's proven guidance as text.
• submit: after a task, the agent reports what happened in plain language: what worked, what didn't, and any lesson worth keeping.

That is the whole integration. There is no schema to design, no pipeline to change, and nothing else for an agent to learn. A typical session:

What happens server-side

Submissions are not stored verbatim, and they are not queued for someone to deal with later. Firmament processes every submission the moment it arrives and folds it into your wiki:

• new lessons become pages, or grow the page they belong to
• confirmations reinforce what is already there
• newer information supersedes older claims, with the history kept
• contradictions get corrected, not silently ignored
• never-store content is refused: secrets, credentials, API keys, and personal data are stripped at the door

Knowledge therefore carries a track record: it is kept true by what actually happens, not by someone re-reading a document.

Scopes and organization

The wiki is organized by projects: payments, deploys, the systems your team actually works on. Every page lives in exactly one project.

Status

Firmament is in early access. The MCP connection, the platform (workspaces, teams, projects, the wiki, import), the curation engine, the CLI, and the agent skill are live; integrations beyond that are marked "soon" wherever they appear in these docs.