anoman
API Reference

Complete API Reference

Base URL: https://api.anoman.io — All authenticated endpoints require Authorization: Bearer anm-sk-...

OpenAI-compatible

/v1/ — OpenAI-compatible

  • POST
    /v1/chat/completionsMain endpoint. Supports stream:true (SSE) and batch-eligible requests (returns 202 + poll_url).
  • POST
    /v1/completionsLegacy text completions passthrough.
  • GET
    /v1/modelsModel catalog. Each model includes: tier, provider, batch_support, provider_multiplier, tier_multiplier.
  • POST
    /v1/embeddingsEmbeddings passthrough to provider.

Anthropic-compatible

/anthropic/ — Anthropic Messages API

  • POST
    /anthropic/v1/messagesFull Anthropic Messages API proxy. Streaming supported. Set ANTHROPIC_BASE_URL=https://api.anoman.io/anthropic.

Public (no auth)

/public/v1/

  • GET
    /public/v1/pricingModel pricing at 0% markup. No auth. ISR 1h.
  • GET
    /public/v1/statusSystem status, provider health, active incidents. No auth. ISR 60s.
  • GET
    /public/v1/modelsPublic model catalog (is_public=true). No auth.

Anoman-native

API Keys

  • POST
    /anoman/v1/keysCreate virtual API key.
  • GET
    /anoman/v1/keysList keys for authenticated customer.
  • DELETE
    /anoman/v1/keys/{id}Revoke key.
  • PATCH
    /anoman/v1/keys/{id}Update key name, budget, or guardrail_overrides.

Traces & Agent Sessions

  • GET
    /anoman/v1/tracesList traces with filters (model, date, routing_mode, cache_hit).
  • GET
    /anoman/v1/traces/{id}Trace detail with spans.
  • GET
    /anoman/v1/sessionsList agent sessions (filterable by agent_id, status).
  • GET
    /anoman/v1/sessions/{id}/actionsAction timeline for a session.

Usage & Billing

  • GET
    /anoman/v1/usageUsage stats (tokens, cost, requests) for current period.
  • GET
    /anoman/v1/usage/by-modelUsage breakdown by model.
  • GET
    /anoman/v1/usage/savingsCache savings and batch savings breakdown.
  • GET
    /anoman/v1/usage/batch-economicsBatch rate, savings, and SLA compliance.
  • GET
    /anoman/v1/usage/rate-limitsPer-customer token-cap usage by model tier class plus RPM/TPM burst-guard status.

Batch Jobs

  • GET
    /anoman/v1/batch/{job_id}Poll batch job. Returns 202 (processing) or 200 (complete) with full chat completion result.
  • GET
    /anoman/v1/batchList all batch jobs for API key.
  • DELETE
    /anoman/v1/batch/{job_id}Cancel a queued job.

Burst Credits

  • POST
    /anoman/v1/burst/activateBuy temporary rate limit boost (2x/5x/10x for 1h/4h/24h). Deducted from prepaid balance.
  • GET
    /anoman/v1/burst/statusCheck active burst for current key.
  • DELETE
    /anoman/v1/burst/deactivateCancel active burst (no refund).
  • GET
    /anoman/v1/burst/pricingAvailable burst options for current tier.

Policies

  • POST
    /anoman/v1/policiesCreate agent policy.
  • GET
    /anoman/v1/policiesList policies.
  • PUT
    /anoman/v1/policies/{id}Update policy (tool rules, budget limits).
  • DELETE
    /anoman/v1/policies/{id}Delete policy.

MCP Governance

  • POST
    /anoman/v1/mcp/serversRegister MCP server — discovers tools via tools/list.
  • GET
    /anoman/v1/mcp/servers/{id}/toolsList tools exposed by an MCP server.
  • PATCH
    /anoman/v1/mcp/tools/{tool_id}Update tool policy_mode (allow/deny/require_approval) and rate_limit_per_min.
  • GET
    /anoman/v1/mcp/invocationsMCP tool invocation audit log.

Anomalies & Audit

  • GET
    /anoman/v1/anomaliesList detected anomalies with severity and detection_method.
  • POST
    /anoman/v1/anomalies/{id}/ackAcknowledge anomaly.
  • GET
    /anoman/v1/audit/eventsSOC2 audit log — cursor-paginated, NDJSON export available.

Live Feed (SSE)

  • GET
    /anoman/v1/events/streamSSE stream. Events: request, blocked, batch_enqueued, batch_completed, batch_escalated, anomaly, provider_issue.

SSO (Enterprise)

  • POST
    /anoman/v1/sso/connectionsRegister SAML IdP (Enterprise). Returns metadata_url for IdP setup.
  • GET
    /anoman/v1/sso/connectionsList SSO connections.
  • DELETE
    /anoman/v1/sso/connections/{id}Deactivate SSO connection.

Error codes

Standard error responses

HTTPCodeMeaning
401auth_invalidInvalid or missing API key.
402budget_exceededMonthly budget exhausted.
402token_quota_exceededWeekly or monthly token cap hit for a model tier class.
403policy_violationTool call blocked by policy.
403guardrail_triggeredContent blocked by guardrail.
429rate_limit_exceededRPM or TPM burst guard tripped.
429concurrent_limit_exceededPer-key inflight cap hit.
503provider_unavailableAll providers failed after retries.

Response headers

X-Anoman-* transparency headers

HeaderExample value
X-Anoman-Guardrail-Injectionpass score=0.12
X-Anoman-Guardrail-Piimasked 3 entities
X-Anoman-Guardrail-Contentpass
X-Anoman-Cachenone | provider | semantic
X-Anoman-RegionID | SG
X-Anoman-Tierstarter | pro | payg | enterprise
X-Anoman-Burst-Activetrue | false
X-Anoman-Guardrail-Masterdisabled (when all guardrails bypassed)
X-Anoman-Inflight-Cap30 (on 429)