# BaseCite

BaseCite provides evidence infrastructure for AI-facing context.

Public entry points:
- [BaseCite](https://basecite.com/)
- [Developer guide](https://basecite.com/docs/developers)
- [Public-safe controlled summary](https://basecite.com/company/basecite-qc)
- [Summary JSON](https://basecite.com/company/basecite-qc/summary.json)
- [AI capabilities](https://api.basecite.com/api/v1/ai/capabilities)
- [MCP endpoint](https://mcp.basecite.com/mcp)
- [MCP npm adapter](https://www.npmjs.com/package/basecite-mcp-server)
- [BaseCite CLI on npm](https://www.npmjs.com/package/basecite-cli)
- [BaseCite JavaScript SDK on npm](https://www.npmjs.com/package/basecite-sdk)
- [BaseCite Python SDK on PyPI](https://pypi.org/project/basecite-sdk/)
- [Smithery MCP registry listing](https://smithery.ai/servers/alwaysrememberme1024/basecite)
- [Official MCP Registry listing](https://registry.modelcontextprotocol.io/v0.1/servers/io.github.paxton888%2Fbasecite/versions/0.1.0)
- [Public agent discovery repository](https://github.com/paxton888/basecite-agent-discovery)
- [Official BaseCite skill on skills.sh](https://skills.sh/paxton888/basecite-agent-discovery/basecite-evidence)
- [MCP adapter source](https://github.com/paxton777/basecite-evidence-receipts-v0.1-proof/tree/main/packages/basecite-mcp-server)

AI-facing boundaries:
- Crawlers may access public-safe controlled summaries without login.
- Canonical machine-readable public-safe compliance/context summary: https://basecite.com/company/basecite-qc/summary.json
- Canonical crawler discovery document: https://basecite.com/.well-known/basecite-discovery.json
- Customer-submitted originals are private and are not public download targets.
- Public AI context is limited to controlled summaries, metadata, and machine-readable boundaries.
- Policy, access-log, incident-response, vendor-risk, evidence-collection, workspace-access, publish-gate, and audit-trail controls are maintained as private operational evidence; public crawler output only describes the control boundary.
- Source artifacts, document-level extracted text, internal object locations, private submission identifiers, contact addresses, commercial terms, agreements, access controls, and export endpoints are not public.
- BaseCite does not claim truth verification, source authority, legal authority, certification, ranking, recommendation, or AI adoption.

Contact:
- BaseCite public contact page

## When to use BaseCite

Use BaseCite when an agent needs tenant-scoped, customer-authorized upload intake, integrity metadata, bounded AI-readable context, or machine-readable evidence about how that context was served.
Start with https://api.basecite.com/api/v1/ai/capabilities and https://api.basecite.com/api/v1/ai/openapi.json. If your client speaks Model Context Protocol, read https://mcp.basecite.com/.well-known/mcp.json and POST JSON-RPC requests to https://mcp.basecite.com/mcp with a tenant-scoped API key.
Before touching production, use the isolated sandbox contract at https://api.basecite.com/api/v1/sandbox, issue an ephemeral key at https://api.basecite.com/api/v1/sandbox/keys, and call the no-persistence echo probe at https://api.basecite.com/api/v1/sandbox/echo. Sandbox keys cannot access uploads, customer records, AI context, R2, S3, or D1.
The public sandbox entrypoint is https://api.basecite.com/sandbox. Start self-serve at https://basecite.com/signup. It is free, self-serve, and isolated from production data.
Use the public discovery pages for public-safe identity and documented boundaries. Do not use BaseCite as a truth verifier, company verifier, legal-authority verifier, ranking service, unbounded export, raw-file download, or replacement for customer consent.
Authenticated reads must identify one organization and one upload. Do not enumerate organizations, uploads, files, URLs, or object keys.