REST API (/v1)

Availability

Cloud & Enterprise · per-endpoint plan tiers noted below

This reference covers the /v1 REST endpoints registered by the ee/* modules. Except for the open-core eval/guard endpoints, these paths are not present in Community — the Community collector imports none of the commercial modules, so they return 404. All requests authenticate with an ingest key (Authorization: Bearer <key>); see API overview.

Plan tiers below refer to Splyntra Cloud; the same feature set is available self-hosted under an Enterprise license. See Plans & pricing.

Governance

Append-only ledger, policy engine, delegation, and the central authorization decision. See Governance overview.

Method	Path	Purpose	Plan
GET	/v1/ledger	List activity-ledger entries for the org	Team+
POST	/v1/ledger	Append an entry to the hash-chained ledger	Team+
GET	/v1/policies	List policy rules	Enterprise
POST	/v1/policies	Create a policy rule	Enterprise
DELETE	/v1/policies/{policyID}	Delete a policy rule	Enterprise
GET	/v1/delegation	List agent permissions and delegation state	Enterprise
POST	/v1/delegation/permissions/{id}	Grant an agent permission	Enterprise
DELETE	/v1/delegation/permissions/{id}	Revoke an agent permission	Enterprise
GET	/v1/spend-limits/{id}	Read an agent's spend limit	Team+
POST	/v1/spend-limits/{id}	Set a daily/monthly USD spend cap	Team+
DELETE	/v1/spend-limits/{id}	Remove a spend limit	Team+
GET	/v1/approval-rules/{id}	Read an approval rule	Enterprise
POST	/v1/approval-rules/{id}	Create/update an approval rule	Enterprise
DELETE	/v1/approval-rules/{id}	Delete an approval rule	Enterprise
GET	/v1/approvals/{id}	Read a pending approval request	Enterprise
POST	/v1/approvals/{requestID}/decide	Approve or reject a request	Enterprise
POST	/v1/authorize	Decide whether an agent may take an action	Team+

POST /v1/authorize is the decision endpoint agents call before a consequential action. It evaluates, in order: explicit agent deny → spend limit (fails closed if analytics are unavailable) → policy evaluation (deny-wins) → approval-rule match. It returns {"decision": "allow" | "deny" | "needs_approval"} and writes every outcome to the ledger. Spend-limit checks apply from Team+; policy and approval-rule evaluation require Enterprise. See Delegation & approvals.

Example — POST /v1/authorize

curl -X POST "$SPLYNTRA_ENDPOINT/v1/authorize" \
  -H "Authorization: Bearer $SPLYNTRA_API_KEY" \
  -H "content-type: application/json" \
  -d '{
    "action": "payments.refund",
    "agent_id": "support_agent",
    "context": { "amount": 80 }
  }'

{ "decision": "allow" }

Identity

Agent registry, scoped credentials, cross-agent trust, and OIDC federation. All identity endpoints are Enterprise. See Agent identity.

Method	Path	Purpose
GET	/v1/identity/agents	List registered agents
POST	/v1/identity/agents	Register an agent
DELETE	/v1/identity/agents/{agentID}	Remove a registered agent
GET	/v1/identity/agents/{agentID}/credentials	List an agent's credentials
POST	/v1/identity/agents/{agentID}/credentials	Issue a scoped credential (secret shown once)
DELETE	/v1/identity/credentials/{credID}	Revoke a credential
GET	/v1/identity/trust	List trust policies
POST	/v1/identity/trust	Create a cross-agent trust policy
DELETE	/v1/identity/trust/{policyID}	Delete a trust policy
POST	/v1/identity/authorize	Evaluate cross-agent trust (deny-wins, default deny)
GET	/v1/identity/idp	List trusted OIDC issuers
POST	/v1/identity/idp	Trust an OIDC issuer for federation
DELETE	/v1/identity/idp/{idpID}	Remove a trusted issuer
POST	/v1/identity/token	Exchange a federated token for a scoped credential

See Trust policies and Federation.

Compliance

Framework reports and ledger verification. All compliance endpoints are Team+. See Compliance.

Method	Path	Purpose
POST	/v1/compliance/reports	Generate a SOC 2 / EU AI Act / NIST AI RMF report
GET	/v1/compliance/reports	List generated reports
GET	/v1/compliance/reports/{id}?format=csv	Fetch a report (JSON or CSV)
POST	/v1/compliance/verify-ledger	Verify the ledger hash chain
POST	/v1/compliance/policies/dry-run	Preview what the live policy set would decide

Eval & guard

The scoring and guard endpoints are open core — present in every edition.

Method	Path	Purpose	Plan
POST	/v1/scorers	Run a scorer against results	Open core
POST	/v1/guard	Pre-flight injection check for an inline guard	Open core

The built-in scorers ship in the core; the llm_as_judge scorer is a commercial add-on registered by ee/scorers-pro (Enterprise). See Scorers and Guardrails.