SEIF SEIF documentation
SEIF decides, proves and versions what each AI agent receives: which facts, from which source and revision, checked when, and what stays out and why. No model is called to decide it. Start with the first three pages; the rest is organized by what you want to do.
Start here
| Why SEIF | The problem, where SEIF fits next to agent runtimes and RAG, honest limits |
| Quickstart | Two minutes in the browser, five in your coding agent, ten in your application |
| How resolution works | The pipeline, every status and reason, the guarantees and the tests that prove them |
Concepts
Context Projects · File-native projects · Connecting sources · Connectors · Consumers and objectives · Context Profiles and rules · Retrieval and budget · Context Tests · Test troubleshooting · Context Builds · Context Runtime · Context Traces · Runtime Context Providers · Deployment Bindings
Use it
| Goal | Read |
|---|---|
| Work in the browser (Contexto, Agentes, Conexões) | The SEIF Suite · Web and desktop parity |
| Give Claude Code, Codex or Cursor the project's context | SEIF in the agent you already use · CLI and console · Language Server |
| Use the same project in Trinity, OpenBot or any runtime | One project, any runtime · Trinity · OpenBot |
| Edit, inspect and experiment like an IDE | Context Lab · Perspective Mixer · Context World |
| Connect databases and put an agent online (chat link, WhatsApp, widget, API) | Data & go live |
| Run an AI team with approvals | Team · Sales funnel on WhatsApp · One agent system |
| Give an agent tools with exact approval for writes | Context and tool delivery · Agent Factory |
| Gate pull requests on context | Context CI · Context Regression |
| Promote and roll back across environments | Context Registry · Environments and promotion · Deployments · Linux server |
| Investigate an incident and keep it from recurring | Investigating a trace · Reproduction · Context reliability · Reliability Impact |
Reference
Integration API (seif.context-runtime.v1) · Python SDK · TypeScript SDK · Operational events and webhooks · Connector SDK · Authorization model · Service identities · Team and RBAC · Release attestation
Security and privacy
Security boundary · Trace privacy and retention
Architecture and decisions
Architecture · ADR 001 Desktop and boundaries · ADR 002 Build, Runtime, Trace · ADR 003 Commercial control plane · ADR 004 Runtime context and identities · ADR 005 Investigation · ADR 006 Context and tool contracts · Delivery · Advanced: SEIF provenance · Workspace inspector