agentafk

Settings Generator

Configure ~/.afk/ without reading the docs.

Pick what you need. Copy the output. Paste it once. Generates ~/.afk/config/afk.config.json and ~/.afk/config/afk.env.

Your input never leaves your browser. No analytics, no telemetry, and zero third-party scripts. The generator runs entirely client-side and makes no network requests. Verify it yourself: open DevTools → Network and reload.

Quick setup: covers 80% of installs

I want to run AFK with using .My ANTHROPIC_API_KEY:

Add what you need

Show all 241 optionsEvery env var + JSON field, by category — skip anything you don't need
Model & Sampling33

Read the docs↗

→ afk.envdocs↗

Default model for agent turns. Tier names (small/medium/large/local) follow whatever that slot is bound to; sonnet/opus/haiku/fable pin one fixed model. Defaults to the medium tier.

→ afk.env

Extended thinking: adaptive lets afk decide per turn, enabled:max always thinks, disabled turns it off. Custom accepts a token budget.

→ afk.env

Reasoning-effort level for providers that expose one.

Advanced options
→ afk.env

TTL for Anthropic prompt-cache blocks.

→ afk.env

Deprecated and ignored by the generation path. Use AFK_MAX_OUTPUT_TOKENS to cap per-response output tokens.

→ afk.env

Cap on output tokens per turn. Falls back to the provider default when unset.

→ afk.config.json

Token cap written to afk.config.json.

→ afk.config.json

Sampling temperature. Leave unset to use the provider default.

→ afk.env

Hard spend ceiling for a single run, in USD. The run stops when it is reached.

→ afk.env

Comma-separated override for image (vision) capability detection on the openai-compatible provider.

→ afk.env

Legacy alias for AFK_MODEL, kept for pre-AFK_* deployments. Prefer AFK_MODEL.

→ afk.env

Experimental streaming connection retry wall budget in milliseconds.

→ afk.env

Disable the temporary within-turn context pressure guard (1/true/yes/on).

→ afk.env

Temporary within-turn context guard operational threshold as percent of the effective provider-route limit (catalog metadata for subscription routes; contextLimitFor() for API-k…

→ afk.env

Disable Anthropic prompt caching when set to 1/true/yes/on.

→ afk.env

Maximum number of simultaneously-running agents across the entire session tree.

→ afk.env

Maximum number of child agents a single agent may have running concurrently (per-agent concurrent child cap).

→ afk.env

Maximum sub-agent/skill nesting depth; 0 disables nested delegation entirely (the agent, skill, AND compose tools all refuse).

→ afk.env

Maximum total agents spawned across the entire session tree (lifetime, not concurrent).

→ afk.env

Byte threshold for delegation tool results (agent/compose/skill).

→ afk.env

Bypass the process-wide rate-limit admission gate (issue #941).

→ afk.env

Jitter ceiling (ms) for the rate-limit admission bucket: each waiter at a window boundary wakes at reset + random(0..ceiling) to avoid re-storming the API.

→ afk.env

When set to 1/true, forked sub-agents that hit an OAuth usage-limit 429 will park and wait (up to 2 hours) for a keychain account hot-swap before surfacing the error — the same…

→ afk.env

Foreground subagent result size cap in bytes.

→ afk.env

Enable LLM-generated empty-prompt suggestions — a proposed next action shown as ghost text when the prompt is blank, accepted with Tab or Right-arrow.

→ afk.env

Numeric temperature override for model sampling.

→ afk.env

Set to 1 to stop publishing/reading the cross-process usage ledger (namespace `usage` in state/kv/kv.db).

→ afk.env

Disable the shared agent workspace (WorkspaceStore + workspace_publish/workspace_query tools + preamble injection).

→ afk.envhuman-gated

Base URL for xAI API-key mode.

→ afk.env

Override the semver sent as x-grok-client-version to the xAI OAuth CLI chat proxy.

→ afk.envhuman-gated

Base URL for xAI SuperGrok / SuperGrok Heavy / X Premium+ OAuth inference.

→ afk.config.json

Default model id / alias.

→ afk.envhuman-gated

Standard OpenAI SDK base URL override.

Model Slots & Local Models16
Provider & Endpoints5
API Keys7
Safety & Permissions24
System Prompt2
Telegram15
Daemon15
Auto-routing5
Interactive & UI56
Worktrees12
Browser Control7
Session & Context18
MCP Servers1
Paths5
Updates1
Debug & Diagnostics19
↓ See your config

Beyond this form

This page writes two files: afk.config.json and afk.env. These parts of setup live in the CLI, where they belong.

  • Check your setupafk doctor

    Health check across auth, config, provider reachability and state directories.

  • Sign inafk login

    Interactive auth. Browser OAuth cannot happen in a config file. Use this instead of pasting a token.

  • Connect Telegramafk telegram setup

    Walks token entry and the chat allowlist without the token passing through anything else. Then afk telegram start.

  • Keep it runningafk service install

    Installs a launchd agent (macOS) or systemd --user unit (Linux) so the bot and daemon survive reboot.

  • Schedule recurring workafk schedule add

    Cron-driven tasks in schedules.json. afk schedule list / logs to inspect runs.

  • Add MCP servers

    Edit ~/.afk/config/mcp.json by hand. A repo-local .mcp.json also works, but is fail-closed: set AFK_ALLOW_PROJECT_MCP above to opt in, because it spawns that repo’s commands.

  • Give afk project context

    Run /init inside the REPL to generate an AFK.md for the repo. It is the highest-leverage file you can add.

  • Change one key laterafk config set <key> <value>

    You do not need this page again. afk config get/set edits both artifacts in place.

Full reference: configuration docs↗