Documentation
Everything about running, connecting and administering Contextator, in one place.
Getting started
- ContextatorA self-hosted documentation server that speaks the Model Context Protocol, giving AI agents semantic search over your own docs.
- Quick StartFrom nothing to an agent answering questions about your documentation — Docker Compose, one project, one connected client, about ten minutes end to end.
- InstallationEvery way to run Contextator — Docker Compose, plain docker run, building the image yourself, or from source — plus the first account and a reverse proxy.
- ConfigurationEvery environment variable Contextator reads — server, accounts, database, storage, documents, uploads, embeddings, search, sync, observability and reverse proxies.
- Dashboard TourA guided tour of the Contextator dashboard — signing in, the account menu, top bar, project list and detail, keyboard shortcuts, footer and the Users page.
Sources
- ProjectsThe unit everything else belongs to — how to create a project, control MCP access, read its statuses, and re-index or delete it.
- Document SourcesThe six kinds of source a project can hold, the mount name every path is prefixed with, and how syncing and failures work across all of them.
- Local Directory SourceIndexing a folder already on the server in place, the allowed-roots security boundary, and what gets scanned and skipped.
- Git Repository SourceIndexing documentation straight from a git repository — private repos, provider tokens, how syncing and push-triggered re-indexing work.
- Upload SourceDropping files, folders and archives onto the dashboard for documentation that has no other home, and the limits that apply to it.
- Obsidian VaultsIndexing an Obsidian vault as a local directory or an upload, and exactly which wikilink and callout syntax gets rewritten first.
- NotionIndexing a Notion workspace through the API — creating an integration, sharing pages, what gets imported, and the export-zip alternative.
- ConfluenceIndexing Confluence Cloud or Data Center — deployment differences, token types, the private-address boundary, and the Data Center webhook.
- Documentation Site SourceIndexing a public site's own sitemap, llms.txt or a crawl — detection, the five crawl ceilings, robots.txt, and how freshness is checked.
- Content TypesThe four content-type flavors a source can pick — plain, Obsidian vault, Notion export and OpenAPI/Swagger — and what each one transforms before chunking.
Operating
- IndexingWhat happens during an index run, why re-indexing an unchanged project embeds nothing, how documents are chunked, and what is refused while indexing is in progress.
- Embedding ModelsThe default local model, switching to OpenAI, changing models safely, how search actually scores results, and running fully air-gapped.
- Connecting AI ClientsReady-to-paste setup for Claude Code, Cursor, Claude Desktop and any other MCP client, plus tokens, remote access and verifying an endpoint without an agent.
- MCP ToolsThe three read-only tools every project publishes — search_docs, list_topics and read_document — their arguments, their limits, and what they cannot do.
- Push WebhooksSetting up a per-source push webhook for GitHub, GitLab, Gitea, Bitbucket and Confluence Data Center, what happens on delivery, and rotating a secret.
- Running on KubernetesInstalling the Helm chart — database and SECRET_KEY handling, ingress, persistence, probes, security context, resource sizing and uninstalling.
Administration & security
- Admin APIThe dashboard's own JSON REST API — auth, SSO, projects, search, sources, members, MCP tokens, webhooks, metrics, the audit log and the error shapes every endpoint shares.
- Accounts and PermissionsHow sign-in, roles, project membership, passwords and sessions work in the dashboard — and what to do when nobody can sign in.
- SecurityContextator's threat model — a new project's MCP endpoint requires a token by default, but can be switched open — and the operator checklist for locking it down.
- Backup and DataWhere Contextator stores its database, models and materialised sources, and how the built-in backup and restore command takes and checks an archive.
- TroubleshootingExact error messages for startup, sign-in, source, indexing and search problems, matched to their causes and fixes.
- FAQShort, self-contained answers about data, accounts, sources, the MCP endpoint, indexing and the AGPL licence.
- Single Sign-OnFederated sign-in over OIDC — turning it on, auto-provisioning, linking and unlinking your own account, and the guardrails that keep root a local-only account.