Discovery — live, auth-free
Two public surfaces an agent reads before it touches anything.
Both live on the app host app.entradigm.com, are
CORS-enabled and require no authentication. They contain only platform
documentation and the public schema — never tenant data.
Orientation (llm.txt)
Plain-text orientation for LLM agents: the platform docs flattened to markdown, the list of live endpoints, and copy-paste curl recipes.
Open llm.txt →Introspection (intent schema)
The machine-readable JSON schema of every intent — request, response, slides and provenance shapes — plus the capabilities block: apiVersion and per-intent live flags stating what this deployment can do. Readable from any origin.
Open schema →The handshake
Orient → introspect → act — in three calls.
The first two steps are auth-free; the third is bearer-token gated and rate-limited — the agent passes the tenant token to trigger a cycle.
# 1. orient
curl -s https://app.entradigm.com/.well-known/llm.txt
# 2. introspect the capabilities block
curl -s https://app.entradigm.com/strategist/schema | jq '.capabilities.intents'
# 3. act (bearer-gated)
curl -s -X POST https://app.entradigm.com/strategist/run \
-H "Authorization: Bearer $ENTRADIGM_TOKEN" \
-H "Content-Type: application/json" \
-d '{"intent":"generate-cycle","tenant":"your-tenant","cycleId":"2026-07-23-demo"}'In production, POST /strategist/run additionally expects the tenant's configuration passed inline as tenantConfig (optional in the schema, required by the production runtime) — see the request shape in the schema above.
An external agent — your n8n workflow, your Make scenario, your coding agent —
drives Entradigm through the same schema-typed contract and the same
human-review gate as an operator in the interface: agent-triggered cycles pause
for human approval before anything is published.
Three further public schema endpoints (/strategist/tools/schema,
/memory/schema, /library/schema)
are indexed in llm.txt.
What we do better
A surface an agent can already read today — and a contract designed for it.
Live capabilities carry a Live tag. Properties of the future
public, versioned API carry a Design goal tag — the formal
/v1/ path is on the roadmap below.
Auth-free discovery
The public llm.txt and the intent schema at /strategist/schema — readable from any origin, no token.
Intents through one endpoint
Agent intents run bearer-gated and rate-limited through /strategist/run — the same gates, budgets and human review as the in-app trigger.
Machine-readable capability matrix
The capabilities block in the schema tells an agent what this deployment can do — per-intent live flags, supported platforms and content types — before it tries.
Versioned API
The public API is designed for a versioned path (/v1/), so the contract does not shift under an agent.
Unified error envelope
The public API will return errors as {error:{code,message,context}} — machine-readable and consistent instead of free-form prose.
Cursor pagination & rate-limit headers
The public API is designed to walk lists via stable cursors and carry the remaining budget in every response via X-RateLimit-* headers. Signed webhook payloads for cycle events are also part of the target picture.
Roadmap
What comes next for the agent surface.
Each of these carries its roadmap tag on every surface that mentions it. The live contract above is what you build on today.
MCP server
A single endpoint through which any LLM agent wires in Entradigm as a tool with full schema introspection.
n8n node
A native node for n8n automation workflows.
SKILL.md
An agent-readable skill manifest for coding agents.
Public /v1/ API
The formal public, versioned API with error envelope, cursor pagination and rate-limit headers.