Which API should you use with NanoClaw?

NanoClaw runs Claude Code through the Claude Agent SDK, so the endpoint you give it must speak Anthropic's Messages API. For an always-on NanoClaw, the best cost-optimized option is Standard Compute's Messages endpoint: set NANOCLAW_ANTHROPIC_BASE_URL to https://api.stdcmpt.com and your key as NANOCLAW_ANTHROPIC_AUTH_TOKEN before setup, give the agent the model anthropic/claude-standardcompute, and the router sends each step to the cheapest model that can do it (Claude Fable 5.1 or GPT-6 Astra only when it counts), on a plan from $19 a month with a $20 compute budget. NanoClaw's setup also accepts a Claude Pro or Max subscription or an Anthropic API key. NanoClaw itself is free and MIT-licensed.

Light or occasional use

If you already pay for Claude Pro or Max and your agents stay inside its usage limits, NanoClaw's subscription sign-in adds no new bill. For a few messages a day, an Anthropic API key billed per token on a small Claude model may cost the least.

Daily use, cost-optimized

When NanoClaw answers chats and runs scheduled tasks around the clock, Standard Compute's router sends each step to the cheapest model that can do it, and for every model we vet providers for price, uptime and speed, with automatic fallback. Opt-in Europe or US Compliance rules apply per key, team or company, and we don't train on your prompts or store them. The bill stays at the monthly plan price: requests stop with HTTP 402 when the compute budget is used, so a runaway agent cannot bill you on top.

API and subscription options for NanoClaw

Prices and allowances can change. NanoClaw's custom endpoint setup, vault behavior and model settings come from its credentials, environment-variable and agent-provider docs and its README, read on October 15, 2026. NanoClaw: credentials · NanoClaw: agent providers · NanoClaw: environment variables · NanoClaw repo. Spot an error? Tell us.

Point NanoClaw at Standard Compute

export NANOCLAW_ANTHROPIC_BASE_URL=https://api.stdcmpt.com
export NANOCLAW_ANTHROPIC_AUTH_TOKEN=sc_live_…   # your Standard Compute key
bash nanoclaw.sh

# then, per agent group: let the router pick the model per step
ncl groups config update --id <agent-id> --model anthropic/claude-standardcompute
ncl groups restart --id <agent-id>

Setup writes ANTHROPIC_BASE_URL to .env and stores the key in the credential vault, keyed to api.stdcmpt.com; the agent container only sees the base URL and a placeholder token. Use the base URL without /v1: the Claude Agent SDK adds /v1/messages itself. For every group at once, set NANOCLAW_DEFAULT_MODEL to anthropic/claude-standardcompute instead and restart the host.

NanoClaw questions

Can NanoClaw use an OpenAI-compatible API?

Not with its default Claude provider, which speaks only Anthropic's Messages API; use an Anthropic-compatible endpoint such as https://api.stdcmpt.com. NanoClaw's /add-opencode skill swaps the agent brain for OpenCode, which reaches OpenRouter, OpenAI, DeepSeek and other providers through OpenCode's own config, and /add-codex adds Codex.

Why does NanoClaw refuse ANTHROPIC_AUTH_TOKEN in the container environment?

NanoClaw's session admission rejects credential values in container environments: real keys live in the credential gateway (OneCLI's Agent Vault by default) and are injected on the wire. For a custom Anthropic endpoint, set NANOCLAW_ANTHROPIC_BASE_URL and NANOCLAW_ANTHROPIC_AUTH_TOKEN before setup; the container gets ANTHROPIC_BASE_URL and a gateway-managed placeholder token.

Which model does NanoClaw use on a custom endpoint?

The group's model if set, then NANOCLAW_DEFAULT_MODEL, then the Claude Agent SDK's default. Left unset, NanoClaw asks for Claude models by name and Standard Compute serves those Claude models from your budget. Set anthropic/claude-standardcompute to let the router pick the cheapest capable model per step, or an exact ID from /v1/models such as anthropic/claude-sonnet-5.5 to pin one.