Axiom

Ingest API: POST /v1/datasets/{dataset}/ingest with JSON array of events. Events are arbitrary JSON objects — no schema definition required. Timestamps are auto-added if not present. The Ingest API accepts NDJSON, JSON array, and CSV formats. Query API: POST /v1/datasets/{dataset}/query with APL query string. APL supports filtering (where), projection (project), aggregation (summarize), sorting (sort), time buckets (bin), and joins across datasets. Virtual fields compute derived values at query time. Annotations mark events on the timeline. The API is clean: ingest is one endpoint, querying is one endpoint with a powerful query language. Dashboard API provides saved query and visualization management. Webhook monitors trigger alerts on query results.

API errors return JSON with status, message, and error fields. Ingest errors include per-event error details for malformed data. Standard HTTP status codes. Rate limits: ingest is designed for high throughput — millions of events per minute. Query rate limits are per-organization. APL query execution has timeout limits for complex queries. The serverless-native architecture means query performance scales with data volume — no pre-provisioning required. Dataset retention is the primary constraint (30 days free, configurable on paid plans). The main consideration for agents: APL queries on very large datasets benefit from time-range scoping and field-specific filtering. The query optimizer handles most patterns well but extremely complex aggregations may timeout.

Axiom is a cloud-native log management platform with a unique pricing model: the free tier has no data volume limits — you can ingest as much as you want, with retention as the constraint (30 days free, longer on paid plans). For agents, the Ingest API accepts structured JSON events for storage and querying. The APL (Axiom Processing Language) provides SQL-like querying with aggregation, filtering, and time-series analysis. OpenTelemetry integration enables log, trace, and metric ingestion from standard observability pipelines. For agents processing logs — analyzing application behavior, detecting anomalies, debugging issues — Axiom's unlimited ingestion and queryable storage provide a low-friction observability backend. Integrations with Vercel, Next.js, and serverless platforms enable zero-config log forwarding.

API tokens with configurable permissions: ingest-only, query-only, or full access. Tokens can be scoped to specific datasets. This enables least-privilege access: agents ingesting logs use ingest-only tokens, agents querying use query-only tokens. Personal tokens provide user-level access. Organization tokens provide team-level access. Tokens are passed via Authorization: Bearer header. No OAuth. Tokens don't expire by default. For agents, dataset-scoped ingest tokens provide clean isolation — a token for one dataset can't access another. The permission model is more granular than most logging platforms. For agents in multi-tenant environments, per-dataset tokens enable tenant isolation.

Documentation at axiom.co/docs is well-organized with API reference, APL language reference, integration guides, and SDK documentation. The APL reference is comprehensive with operators, functions, and query examples. Integration guides cover Vercel, Next.js, OpenTelemetry, Pino, Winston, and other logging libraries. SDKs for TypeScript/JavaScript, Go, Python, and Rust handle structured logging with automatic batching. The documentation's strength: it focuses on practical integration patterns rather than abstract concepts. The free tier makes experimentation frictionless. For agents, the Ingest API quick start and APL query reference are the essential starting points.