Agent AFK
What is Agent AFK?
Agent AFK is an open-source, local-first runtime for running coding agents unattended. It allows developers to start coding tasks, walk away, and later review verified results through traces, terminal states, and receipts of agent activity. The runtime supports permission boundaries, reversibility gates, isolated worktrees, parallel subagents, adversarial verification, persistent session memory, and Telegram supervision. It works with Claude, Cursor, Claude Code, Codex, custom agents, and other model providers while keeping the runtime and data on the user's machine.
How to use Agent AFK?
Install Agent AFK globally with Node.js 20 or newer using npm install -g agent-afk. Configure it for your repository and connect your preferred model provider. Use the CLI REPL, background daemon, or Telegram bridge to start tasks. Run slash skills such as /diagnose, /review, /shadow-verify, /mint, or /ship to orchestrate coding, verification, and release workflows. Agent AFK records traces and session state under ~/.afk/, notifies you when work finishes or stalls, and lets you inspect, resume, redirect, interrupt, or roll back runs.
Agent AFK's Core Features
Unattended coding-agent execution with daemon and background workflows
Permission boundaries, reversibility gates, isolated worktrees, and risk-aware actions
Parallel subagent orchestration with coordinated interruption
Adversarial verification and self-checking before changes are accepted
Complete local trace history for sessions, subagents, and tool calls
Explicit terminal states including Done, Blocked, Asking, and Interrupted
Telegram supervision for notifications, responses, redirection, and interruption
Composable slash skills such as /mint, /diagnose, /review, /shadow-verify, and /ship
Persistent plans, preferences, conventions, decisions, and session memory
Support for external and local model providers through a BYO-model approach
Agent AFK's Use Cases
Run a feature implementation overnight and review a verified receipt the next day
Diagnose bugs using parallel hypotheses in isolated Git worktrees
Review a pull request with separate correctness, security, API compatibility, testing, and performance reviewers
Run an end-to-end feature pipeline from specification and research through implementation, verification, and shipping
Supervise long-running coding tasks remotely through Telegram
Inspect agent tool calls and traces when a task stalls or produces unexpected results
Safely prepare commits, pushes, releases, or pull requests with approval and verification gates
FAQ from Agent AFK
What is Agent AFK?
Is Agent AFK free?
Does Agent AFK require an account?
How do I install Agent AFK?
Which operating systems are supported?
Which AI models can Agent AFK use?
Can I monitor coding runs remotely?
Where are traces and session data stored?
What happens when the agent is unsure?
What paid services are available?
Agent AFK Reviews (0)
Agent AFK Pricing
Open Source
$0
Free forever. Includes the full CLI, REPL, daemon, Telegram bridge, slash skills, subagents, hooks, MCP, plugins, local traces, memory, and BYO-model support.
Supported
$5,000 per month
Agent AFK wired into your stack and kept running, including repository, daemon, and Telegram setup, custom skills and orchestration, direct maintainer access, and monthly working sessions.
Dedicated
$10,000 per month
Hands-on partnership with everything in Supported, priority response, weekly working sessions, CI and internal-tool integration, custom orchestration, plugins, skill development, and team onboarding.
Analytic of Agent AFK
Agent AFK Website Traffic Analysis
Agent AFK Launch embeds
Use website badges to drive support from your community for your Toolify Launch. They're easy to embed on your homepage or footer.
Alternative of Agent AFK
Social Listening