coddy-agent

Name: coddy-agent
Author: coddy-project

Verified

General-purpose agent in one static Go binary. ReAct loop, ACP server for IDEs, OpenAI-compatible REST API with embedded web UI, Telegram gateway, cron scheduler, long-term memory, context compaction, rules, skills, MCP. Distroless-ready, works with any OpenAI-compatible provider.

128stars

13forks

Installation

# Add to your Claude Code skills
git clone https://github.com/coddy-project/coddy-agent

Getting Started

Guides for using ai agents skills like coddy-agent.

Caveman: Cut Claude Token Use by 65%
How agent-side prompt compression works, when to use it, and when not to.
What is an AI Skills Marketplace?
Definitions, how marketplaces work, and how to choose between them in 2026.
Getting Started with AI Skills

Security ReportVerified

Last scanned: 7/22/2026

{
  "issues": [
    {
      "file": "README.md",
      "line": 99,
      "type": "remote-install",
      "message": "Install command (remote install script piped to a shell — review the source before running): \"curl -fsSL https://coddy.dev/install.sh | bash\"",
      "severity": "low"
    }
  ],
  "status": "PASSED",
  "scannedAt": "2026-07-22T06:27:22.427Z",
  "npmAuditRan": true,
  "pipAuditRan": true,
  "promptInjectionRan": true
}

README.md

Frequently Asked Questions

What is coddy-agent?

coddy-agent is an open-source ai agents skill for AI coding assistants such as Claude Code, Codex CLI, and ChatGPT, built by coddy-project. General-purpose agent in one static Go binary. ReAct loop, ACP server for IDEs, OpenAI-compatible REST API with embedded web UI, Telegram gateway, cron scheduler, long-term memory, context compaction, rules, skills, MCP. Distroless-ready, works with any OpenAI-compatible provider. It has 128 GitHub stars.

Is coddy-agent safe to use?

Yes. coddy-agent passed SkillsLLM's automated security scan — a dependency vulnerability audit plus prompt-injection heuristics — with no high-severity issues. You can read the full report in the Security Report section on this page.

How do I install coddy-agent?

Clone the repository with "git clone https://github.com/coddy-project/coddy-agent" and add it to your Claude Code skills directory (see the Installation section above).

What programming language is coddy-agent written in?

coddy-agent is primarily written in Go. It is open-source under coddy-project on GitHub, so you can review or fork the full source.

Are there alternatives to coddy-agent?

Yes. SkillsLLM lists many other AI Agents skills you can browse and compare side by side. Open the AI Agents category from the badge at the top of this page, or use the Related Skills and comparison links further down to weigh coddy-agent against similar tools.

Agentic AI for Beginners

Build your first AI agent from scratch - tool use, ReAct pattern, memory, deployment

41 minBeginner

Comments (0)

to leave a comment.

No comments yet. Be the first to share your thoughts!

Related Skills

skills

by mattpocock

Skills for Real Engineers. Straight from my .agents directory.

250,518

claude-source-leaked easyeda-copilot

⚠️ Third-Party Software Notice

This skill is third-party open-source software developed and hosted independently on GitHub. SkillsLLM is an informational directory and does not control or maintain the underlying repository.

Any security checks, ratings, or warnings displayed by SkillsLLM are automated and limited in scope. They do not constitute a security certification or guarantee that the software is safe, error-free, or free from malicious code, vulnerabilities, compromised dependencies, or prompt-injection risks.

Review the source code, permissions, dependencies, and configuration before installing or running any third-party skill. Use is at your own risk. To the maximum extent permitted by applicable law, SkillsLLM is not liable for losses arising from third-party software.

Read the Terms of Service

Desktop (1920×1080)	Mobile (390×844)

Chat	Mobile chat

History	Scheduler

Scheduler — job editor	Settings

Settings — Skills	Settings — Appearance

Settings - MCP servers (project trust gate)	Background tasks

Screenshots: desktop at 1920×1080, mobile at 390×844 from the embedded UI (coddy http). Spec and dev workflow: docs/ui.md, layout tokens: DESIGN.md.

Console TUI (-tags cli) - bare coddy in a terminal, here in Konsole:

Console startup

Console turn with a tool call

More console states and the capture sets: docs/cli.md.

Coddy is a distroless-friendly harness: drop it into minimal images (scratch, distroless, read-only workspaces) without a full OS shell. The same agent core is reachable from a terminal through the console TUI, an IDE over ACP, a browser or any OpenAI client over HTTP, Telegram through the messenger gateway, and cron through the scheduler. The grep and glob filesystem tools use system ripgrep when available and fall back to built-in Go implementations when it is not. The harness layer (ACP RPC, sessions, prompts, providers) stays the same if you tighten the toolset or drive it from automation instead of an IDE. The design also targets container fleets - many Coddy instances in Docker (orchestrator-defined limits, read-only rootfs, mounted workspace) with full control of each container, similar in spirit to agent OS / swarm-style agents, not a single shared chat pool.

Features
Quick start
How to update
Operating modes
Editor and IDE integration
Rules
Skills
MCP server integration
Messenger gateway
Configuration (reference)
Architecture
Documentation
Examples (ACP over stdio)
Persistent sessions
Development
License

Features

Harness-first - ACP server, session lifecycle, prompts, LLM backends, MCP merge, distroless-ready binary
Four surfaces, one agent core - coddy acp for editors, coddy http for the REST gateway and embedded web UI, bare coddy for the interactive console (-tags cli), and coddy gateway for messengers; all four share $CODDY_HOME sessions - see Console, HTTP API
Remote control - point the console or ACP at a running server with --remote <name|host:port|url> (plus --remote-token / CODDY_REMOTE_TOKEN), and switch the web UI between local and remote from the composer environment chip - see Remote control
Self-configuration - the agent edits its own YAML through staged uci-like commands (config_get / config_set / config_changes / config_commit / config_revert / config_rollback): nothing touches the file until a commit you approve, which then validates, snapshots, and hot-reloads skills, rules, tools, and MCP servers - see Configuration reference
ReAct loop - LLM alternates between reasoning, acting (tool calls), and observing results (coding-agent persona out of the box)
Three operating modes - agent (full tool access), plan (planning + text files only), and ask (read-only research: the model can only read the repository and search the web)
Rules - auto-discovers .cursor/rules/, .coddy/rules/, .claude/rules/, .codex/rules/, and nested **/AGENTS.md (agents.md) under the session cwd - see Rules
Skills - slash commands and SKILL.md packs from skills.dirs (defaults: ~/.agents/skills, ~/.coddy/skills, ${CWD}/.coddy/skills; later dirs override earlier) - see Skills
Background tasks - run_command can run detached (background: true plus the model's own expected_seconds estimate); background_list / background_output / background_wait / background_stop collect the result later, a Tasks drawer in the UI shows what is still running with a status ticker, and the permission dialog can widen a grant to a whole program (curl, git status) so a batch of similar calls asks once - see Background tasks
Subagents - the model delegates a bounded, self-contained task to a child agent with its own context window and session (spawn_agent, foreground or detached); definitions are markdown files with YAML frontmatter under ~/.coddy/agents and .coddy/agents (Claude Code's .claude/agents load too), two built-ins (general, explore) ship embedded, project files need a one-time approval (coddy agents trust <name>), tools and permission mode can only narrow, and every run is a background task with a read-only child transcript reachable from the Tasks drawer - see Subagents
MCP server integration - connect any MCP server for additional tools
Multi-provider LLM - OpenAI, Anthropic, Ollama, any OpenAI-compatible API
Context compaction - built-in /compact [instructions] command and automatic summarization when the context reaches compaction.threshold_percent (default 80%) of the model's max_context_tokens; the last compaction.keep_recent_turns (default 2) user turns stay verbatim and the full transcript is preserved on disk - see Configuration
Multimodal / file attachments - attach images and files via the composer (📎) when multimodal: true in the model config; assets saved to ~/.coddy/sessions/<id>/assets/ and injected into the agent context; file chips displayed in the user bubble
Reasoning level - for reasoning models (gpt-5, o-series, Claude thinking models) a composer dropdown picks the effort level (minimal/low/medium/high), mapped to OpenAI reasoning_effort or Anthropic extended-thinking budget_tokens; levels auto-detect from the model id and are configurable per model — see Configuration
ACP protocol - Coddy is an ACP server (coddy acp); pair it with editors or scripts that implement an ACP client (see Editor and IDE integration)
SSH remote execution - built-in ssh_run_command tool runs commands on remote hosts over pure-Go SSH (no external binary); authenticates via SSH agent (SSH_AUTH_SOCK) or ~/.ssh key files — see Configuration
Messenger gateway - optional Telegram bot adapter (-tags gateway.telegram); per-user sessions, group isolation modes, admin ACL; extensible to Discord, Slack, etc. — see Messenger Gateway

Editor and IDE integration

Coddy is an ACP server (coddy acp). Obsidian, VS Code, Zed, scripts, and the bundled coddy http UI are clients that share the same CODDY_HOME sessions when configured with the same home directory.

Configure clients with the absolute path to the binary rather than relying on PATH — some harnesses spaw