Agent capabilities
V-Sum exposes read-only access to technical briefings, events, companies, and people through multiple machine-readable interfaces. No authentication is required for public data. This page summarizes what agents can do and where to find each surface.
Discovery endpoints
- — agents.json — machine-readable capability manifest
- — auth.md — authentication documentation for agents
- — agent-permissions.json — scoped permission manifest
- — agent-card.json — A2A agent card
- — ai-plugin.json — ChatGPT plugin manifest
- — MCP server card
- — API catalog (RFC 9727)
- — schemamap.xml
MCP server
POST https://v-sum.com/mcp — JSON-RPC 2.0, streamable HTTP transport, no authentication.
Available tools (all read-only, idempotent, closed-world):
- —
search_briefings— full-text search over briefings - —
get_briefing— single briefing by slug with chapters - —
list_briefings— paginated list, filter by event or tag - —
list_events— chronological events with briefing counts - —
list_companies— companies with latest briefing slug
Registry: superduperdot/vsum-mcp
Public Data API
api.v-sum.com — read-only JSON, no auth, hypermedia-driven (_links).
OpenAPI spec.
Authentication
All public content and APIs require no authentication. Operator APIs (/api/admin/*)
require a Google OAuth staff session. See auth.md
for the full auth model.
- — OAuth AS metadata (RFC 8414)
- — Protected Resource metadata (RFC 9728)
- — OpenID Configuration
SDKs
- — npm:
v-sum—npm install v-sum - — npm:
vsum-api-client—npm install vsum-api-client - — PyPI:
vsum-api-client—pip install vsum-api-client - — Go:
vsum-sdk-go—go get github.com/superduperdot/vsum-sdk-go - — CLI:
vsum-cli—npm install -g vsum-cli
Content signal
All V-Sum pages emit Content-Signal: ai-train=yes, search=yes, ai-input=yes. AI crawlers are
welcome to index, train on, and cite V-Sum content. See robots.txt
for crawler-specific directives.
For the full developer hub: v-sum.com/developers