One approved decision layer for every teammate and every coding agent.
Engineers contribute decisions while they work. Leads approve and reconcile changes. Every Claude Code, Cursor, Codex, and Gemini CLI session receives the current approved version, so nobody re-explains the same constraint twice.
Same decision layer. Different scope of sharing.
Open Source
Local, individual. Capture, review, version, and replay decisions from plain JSON in ~/.contexer/, with the opt-in local commit guard. No account.
Individual developers who want zero accounts and full local control.
Personal Cloud
Synced, individual. OAuth login, sync across your own machines with an outbound preview before anything leaves, plus web management. Not shared with teammates.
Developers working across multiple machines who still want personal-scoped decisions.
Teams
Shared team governance. Approved team decisions land in every member's agent sessions, leads review candidates, and reconcile with a server-authoritative preview. Contexer Guard checks each pull request against approved decisions and blocks merge when a violation is found.
Engineering teams who want one approved source of truth across every AI coding session.
Organization
Planned multi-team governance: cross-team hierarchy, org-wide policy, SSO, centralized audit, and merge-gate enforcement. Not built yet.
Orgs planning multi-team rollout who want to shape it as design partners.
What's in each tier.
Swipe the table sideways to compare tiers →
Teams inherits everything in Open Source and Personal Cloud, then adds shared governance. Reconciliation previews the server diff before anything changes: the approved team version stays active until a lead approves, retries are idempotent, and pulls never silently overwrite local decisions. Two guards keep decisions enforced. The local commit-time guard runs on your machine: it warns before commit and only blocks when regex or secret checks are explicitly armed locally. Contexer Guard runs on GitHub pull requests: it checks every PR against approved team decisions and blocks the merge when a violation is found. Personal Cloud is individual only, Teams is early access, and everything marked Planned is not shipped yet.
Safe reconciliation, humans keep authority.
When a developer corrects the wording of a shared decision locally, it does not overwrite anything. contexer reconcile shows a server-authoritative diff preview before you confirm, then atomically syncs your personal revision and creates a team candidate for lead review.
- The currently approved team version stays active in every agent session until the candidate is approved.
- Interrupted submissions retry safely, and a changed server head forces a fresh preview instead of a blind retry.
- Changes made in the web app arrive locally as a Suggested Update for review; they never silently overwrite the standing local revision.
- A later in-sync pull clears only the team-created proposal and records the reconciliation outcome.
- 01Local correctionDeveloper edits a shared decision on their machine.
- 02Server-authoritative previewDiff against the current server head, confirm or abort.
- 03Lead-reviewed candidateTeam candidate created; approved version stays active.
- 04Approved team versionReplayed into every member's agent sessions.
One source of truth for every AI coding session.
Engineering decisions flow from the people who make them, through one approved knowledge layer, into every AI assistant your team uses.
Every AI coding session starts with the same approved engineering decisions.
Teams is early access and team-scoped: shared approved decisions across Claude Code, Cursor, Codex, and Gemini CLI, lead review, reconciliation with preview, the opt-in local commit guard (warnings, with blocking only for explicitly armed regex or secret checks), and Contexer Guard, which checks pull requests against approved decisions and blocks merge on violations.
Organization is not built yet. Multi-team hierarchy and cross-team sharing, org-wide policy, SSO (SAML/OIDC), centralized audit and access controls, merge-gate enforcement, and dedicated onboarding are planned, and we are designing them with early partners.
Planning a multi-team rollout? Help us design it.
Organization is on the roadmap, not shipped. We are working with design partners on multi-team hierarchy, org-wide policy, SSO, centralized audit, and merge-gate enforcement. Tell us about your setup and we will come back with scope and timing, typically within one business day.