One backend. Every AI environment.
We standardize on MCP and add thin adapters for the runtimes, IDEs, and CI systems your team already uses. No vendor-specific plugin rewrites.
Standard
Remote MCP server
BetaPrimary integration surface. Tools for search, bundle retrieval, changed-items, acknowledgement. Resources for project, repo, app, and org-wide bundles.
REST API + SDK
BetaFor environments that do not yet support MCP. Same canonical model, same auth, same audit logs.
Event webhooks
BetaOn every change: context.item.updated, context.bundle.published, policy.breaking_change. Downstream caches invalidate and re-fetch.
Chat & agents
ChatGPT custom connector
BetaSearch/fetch over your shared context inside ChatGPT Business knowledge connections.
OpenAI Responses API
BetaDrop-in remote MCP server reference for OpenAI tool calls.
Anthropic Claude MCP
GANative MCP client support. Works with Claude Code, Claude Desktop, and Claude API tools.
IDEs
Cursor / Cline / Claude Code
BetaMCP-aware coding agents query the same context tools at runtime.
VS Code extension
PlannedLocal context cache, real-time invalidation, policy linter.
JetBrains plugin
PlannedIntelliJ / WebStorm / PyCharm context cache + linter.
CI
GitHub Action
BetaBlock merges that violate current rules or fail to acknowledge mandatory context versions. Already shipping in shared-context/actions/sctxt-validate.
GitLab CI step
PlannedSame enforcement primitives for GitLab pipelines.
Sync
CLI: `sctxt sync`
BetaPull current bundles into .ai/context for teams that want files in-repo. Already shipping in shared-context/cli.
Repo materialization bot
PlannedAuto-opens PRs into .ai/context/ and .ai/rules/ when bundles change.
Notify
Slack / Teams change notifier
PlannedPush human-readable change summaries to the channels that own each bundle.
Need a connector that’s not here?
The MCP layer is generic. New runtimes usually need a thin adapter, not a rewrite. Tell us your stack and we’ll scope it.
Talk to us