Files
agent-manager/config.yaml
T
bruno 7fb65c257d
release / release (push) Successful in 13m7s
release / macos (push) Canceled after 0s
feat: am setup — provider openrouter + modeles proposes ; am ai — role am-copilot par defaut ; fix migrate import (config active)
- setup: openrouter ajoute aux CHOICES (base URL, OPENROUTER_API_KEY,
  6 modeles declares dans le registre embarqué, proposes au choix)
- setup: pick_model liste les modeles disponibles (registre) + autre modele
- am ai sans --role: am-copilot par defaut (regeneration auto des roles si
  fichier manquant, degradation douce si echec; role explicite inconnu = erreur)
- fix migrate --import: import_bundle ecrivait la config dans user_config_dir()
  GLOBAL -> ecrasait la config reelle de la machine (et cargo test detruisait
  la config utilisateur a chaque suite via export_import_round_trip).
  L'import ecrit desormais dans app.paths.config_dir (config active).
v1.1.2
2026-08-21 08:36:26 -04:00

1185 lines
46 KiB
YAML

# ============================================================================
# agent-manager — default configuration (embedded in the binary)
# ============================================================================
# This is the base catalog. User configuration files
# (./agent-manager.yaml, ~/.config/agent-manager/config.yaml, --config <file>)
# override or extend it. Run 'am config show' to see the effective config.
#
# Install types: npm | pip | uv | cargo | go | bun | curl | binary | git
# ============================================================================
version: "1.0"
# --- Global settings (null = platform default) -------------------------------
settings:
install_dir: null # default: ~/.local/share/agent-manager/agents
log_dir: null # default: ~/.local/state/agent-manager/logs
default_shell: null # install scripts + shell gateway (default: user's shell)
auto_install_deps: true # offer to install missing dependencies
confirm_before_run: true # confirm before running install scripts
stop_timeout_secs: 5 # SIGTERM grace period before SIGKILL
self_update_repo: Projets/agent-manager # repo used by am self-update (release assets)
self_update_base_url: https://git.dracodev.net/api/v1 # serveur de release (Gitea ici; GitHub: https://api.github.com)
models_prune_days: 30 # un modèle est candidat au prune après N jours sans usage (#73)
catalog_cache_ttl_secs: 3600 # TTL du cache des catalogues distants (#60 #67)
catalog_url: null # URL officielle du catalogue distant (défaut: config.yaml du repo) (#60)
backups_keep: 5 # nombre de backups conservés pour am update --rollback (#64)
sessions_retention_days: 90 # purge des sessions terminées au-delà de N jours (#53)
# Prix des modèles (USD par million de tokens) pour am stats --costs (#49).
# cost_models: # décommenter pour surcharger
# claude-sonnet-4: { input_per_mtok: 3.0, output_per_mtok: 15.0 }
# deepseek-chat: { input_per_mtok: 0.27, output_per_mtok: 1.10 }
# Seuils d'alerte de am monitor (#50) : 0 désactive le seuil.
# monitor_thresholds:
# cpu_pct: 80.0
# mem_mb: 2048
# Plugins d'événements (#75) : scripts exécutés sur on_install/on_start/
# on_stop/on_update, contrat JSON stdin/stdout, timeout, échec non bloquant.
# Le répertoire <state>/plugins/ contient les scripts (+ manifeste .yaml).
# plugins:
# timeout_secs: 10 # temps max d'un plugin (défaut 10 s)
# enabled: [notify] # vide = tous les plugins du répertoire
# Sauvegarde git de l'état (#66) : dépôt où am sync pousse state.json,
# le journal et l'historique (logs/ et backups/ exclus automatiquement).
# sync_repo: https://git.dracodev.net/bruno/am-state.git
# sync_on_exit: true # pousse automatiquement à la fermeture du REPL (opt-in)
# Télémétrie anonyme opt-in (#76) : compteurs agrégés uniquement (jamais de
# chemins, commandes ni identifiants). Désactivée par défaut — RIEN n'est
# collecté ni envoyé tant que enabled n'est pas explicitement true.
# telemetry:
# enabled: false
# endpoint: https://exemple.tld/v1/ping # optionnel : envoi batch périodique
# am ask (#78) : langage naturel → commande am. Les règles locales marchent
# toujours, hors-ligne, sans dépendance. Le raffinement LLM optionnel passe
# par le registre providers (base_url + token du keyring + modèle) — seul
# le texte de la requête est envoyé, jamais les données du journal.
# ask:
# enabled: true # false désactive entièrement am ask
# provider: deepseek # optionnel (défaut: settings.default_provider)
# model: deepseek-chat # optionnel (défaut: modèle par défaut du provider)
# Registre communautaire (#77) : sources de catalogues de confiance +
# auteur par défaut pour am registry publish. L'installation d'une source
# inconnue exige une confirmation (checksum sha256 vérifié avant tout).
# registry:
# sources:
# - https://git.dracodev.net/bruno/am-catalog/raw/branch/main/am-catalog.yaml
# author: bruno
# am ai (#96 #97) : politique de sécurité Shell AI. dry-run par défaut —
# aucune commande modifiante n'est exécutée sans confirmation explicite
# (--yes). `confirm` demande validation pour les commandes risky ;
# `auto` exécute tout (déconseillé). Les motifs supplémentaires s'ajoutent
# aux patterns intégrés (rm -rf, dd if, mkfs, chmod -R 777, ...).
# shell_ai:
# default_safety: dry-run # dry-run | confirm | auto
# risky_patterns:
# - "rm -rf"
# - "> /dev"
# - "dd if"
# Profil sandbox par agent (#79) : commandes autorisées (binaires), répertoires
# de travail et politique réseau. Désactivé par défaut (mode non sandboxé).
# Les tentatives refusées sont journalisées (events sandbox) pour l'audit.
# Exemple dans un agent :
# sandbox:
# enabled: true
# commands: [python, git]
# dirs: ["~/workspace"]
# network: false
# Registre des providers LLM (#88) : nom -> base_url, modèles, modèle par
# défaut. Le provider par défaut est utilisé à l'install/au run quand aucun
# n'est donné (issue #90). Les tokens ne vont JAMAIS ici — ils vivent dans
# le keyring OS sous providers/<nom>/api_key (issue #89, am providers set-token).
# Pour retirer un provider du catalogue par défaut : am providers remove <nom>
# (écrit "<nom>: null" dans votre config — le null supprime l'entrée au merge).
default_provider: anthropic
providers:
anthropic:
base_url: https://api.anthropic.com/v1
default_model: claude-sonnet-4-5
models: [claude-sonnet-4-5, claude-opus-4-1, claude-haiku-4-5]
openai:
base_url: https://api.openai.com/v1
default_model: gpt-5.2
models: [gpt-5.2, gpt-5.1-mini, o4-mini]
deepseek:
base_url: https://api.deepseek.com
default_model: deepseek-chat
models: [deepseek-chat, deepseek-reasoner]
google:
base_url: https://generativelanguage.googleapis.com/v1beta
default_model: gemini-2.5-pro
models: [gemini-2.5-pro, gemini-2.5-flash]
openrouter:
base_url: https://openrouter.ai/api/v1
default_model: deepseek/deepseek-chat-v3-0324
models:
- deepseek/deepseek-chat-v3-0324
- anthropic/claude-sonnet-4.5
- openai/gpt-4o
- google/gemini-2.5-pro
- meta-llama/llama-3.3-70b-instruct
- mistralai/mistral-large-2411
# --- Command aliases ----------------------------------------------------------
aliases:
cc: claude-code
gemini: gemini-cli
gh-copilot: github-copilot
q: amazon-q
droid: factory-droid
gjc: gajae-code
tiny: tiny-agents
continue: continue-cli
ai: aichat
# --- Groups (start/stop/restart several agents together) ----------------------
groups:
dev: [claude-code, aider, codex]
local: [atomic-agent, nanocoder, gptme, goose]
claw: [openclaw, zeroclaw, picoclaw]
mainstream: [claude-code, opencode, gemini-cli, codex, qwen-code]
# ============================================================================
# Agents
# ============================================================================
agents:
# ---------------------------------------------------------------- installés
- name: claude-code
display_name: "Claude Code"
description: "Agent terminal d'Anthropic, repo-aware. Édite, refactore, exécute des commandes git et itère sur les erreurs. Le standard de référence du coding agent."
category: coding-agent
website: https://github.com/anthropics/claude-code
install:
type: npm
package: "@anthropic-ai/claude-code"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: claude
provider: anthropic
config:
env_map:
api_key: ANTHROPIC_API_KEY
model: ANTHROPIC_MODEL
base_url: ANTHROPIC_BASE_URL
tags: [anthropic, assistant, source-available, mainstream]
- name: deepseek-harness
display_name: "DeepSeek Harness"
description: "CLI officiel DeepSeek (dsh) : boot de profils, gestion des plugins Cordis, orchestration multi-agents, boucles de buts (goals) et interface web. Installez-le pour piloter vos agents depuis am."
category: coding-agent
website: https://github.com/deepseek-ai/deepseek-harness
install:
type: npm
package: "@deepseek-ai/dsh"
dependencies:
- { name: node, min_version: "20.0.0", install_hint: "https://nodejs.org" }
run: dsh
provider: deepseek
tags: [deepseek, harness, multi-agent, plugins, web-ui]
- name: hermes-agent
display_name: "Hermes Agent"
description: "Agent Nous Research multi-modèle (300+ providers). Mémoire persistante, création automatique de skills, exécution de code sandboxée, reach multi-plateforme (Telegram/Slack/Discord/WhatsApp)."
category: assistant
website: https://github.com/NousResearch/hermes-agent
install:
type: curl
script: "https://hermes-agent.nousresearch.com/install.sh"
dependencies:
- { name: curl, install_hint: "https://curl.se" }
- { name: sh, install_hint: "un shell POSIX" }
run: hermes
platforms: [linux, macos]
tags: [nous-research, memory, skills, multi-channel]
- name: atomic-agent
display_name: "Atomic Agent"
description: "Agent local-first optimisé pour les modèles locaux (llama.cpp/Ollama). Contrôle navigateur (CDP) + OS, 14 skills intégrés (docker, github, notion, obsidian, pdf, ffmpeg...)."
category: local-first
website: https://github.com/AtomicBot-ai/atomic-agent
install:
type: binary
repo: AtomicBot-ai/atomic-agent
binary: atomic-agent
run: atomic-agent
tags: [local, ollama, browser, skills]
- name: pi
display_name: "Pi"
description: "Harness terminal minimal et adaptable (pi-mono). API LLM unifiée, TUI, skills et MCP. C'est la base sur laquelle Prime Agent est construit."
category: coding-agent
website: https://github.com/badlogic/pi-mono
install:
type: npm
package: "@earendil-works/pi-coding-agent"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: pi
tags: [harness, mcp, tui]
- name: tiny-agents
display_name: "Tiny Agents"
description: "Squad d'agents composables basés sur l'Inference Client HuggingFace + MCP. Chaque agent est défini par un agent.json."
category: coding-agent
website: https://huggingface.co/datasets/tiny-agents/tiny-agents
install:
type: pip
package: tiny-agents
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: tiny-agents
note: "usage : tiny-agents run <chemin-vers-agent.json>"
tags: [huggingface, mcp, experimental]
- name: nemoclaw
display_name: "NemoClaw"
description: "CLI NVIDIA pour provisionner et gérer des sandboxes sécurisées (OpenShell runtime) exécutant des agents. Politiques réseau, filesystem et process."
category: coding-agent
website: https://github.com/NVIDIA/NemoClaw
install:
type: npm
package: nemoclaw
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: nemoclaw
tags: [nvidia, sandbox, docker]
# --------------------------------------------------------------- Prime Agent
- name: prime-agent
display_name: "Prime Agent"
description: "Agent open-source (MIT) de Prime Intellect. RLM (Recursive Language Model) — contexte traité comme variables, outils comme appels de fonction dans un REPL IPython persistant. Construit sur pi."
category: coding-agent
website: https://github.com/PrimeIntellect-ai/prime-agent
install:
type: curl
script: "https://app.primeintellect.ai/prime-agent/install.sh"
dependencies:
- { name: curl, install_hint: "https://curl.se" }
- { name: sh, install_hint: "un shell POSIX" }
run: prime-agent
platforms: [linux, macos]
tags: [rlm, daemon, self-improving]
# --------------------------------------------------- Poids lourds open source
- name: opencode
display_name: "OpenCode"
description: "Agent terminal 75+ providers, intégration LSP, design privacy-first."
category: coding-agent
website: https://github.com/anomalyco/opencode
install:
type: npm
package: opencode-ai
methods:
- type: curl
script: "https://opencode.ai/install"
note: "installateur officiel (bash)"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: opencode
tags: [multi-provider, lsp, privacy]
- name: claw-code
display_name: "Claw Code"
description: "Réécriture clean-room Rust/Python de l'architecture Claude Code. Le repo le plus rapide de l'histoire de GitHub à atteindre 100K stars."
category: coding-agent
website: https://github.com/ultraworkers/claw-code
install:
type: binary
repo: ultraworkers/claw-code
binary: claw-code
run: claw-code
note: "projet en mode musée — maintenance 100 % automatisée"
tags: [rust, clean-room, fork]
- name: gemini-cli
display_name: "Gemini CLI"
description: "Agent terminal officiel Google (Apache-2.0), outils repo + recherche."
category: coding-agent
website: https://github.com/google-gemini/gemini-cli
install:
type: npm
package: "@google/gemini-cli"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: gemini
tags: [google, mainstream]
- name: codex
display_name: "Codex CLI"
description: "Agent local officiel OpenAI, lecture/édition/exécution de code, TUI interactif (Apache-2.0)."
category: coding-agent
website: https://github.com/openai/codex
install:
type: npm
package: "@openai/codex"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: codex
config:
env_map:
api_key: OPENAI_API_KEY
model: OPENAI_MODEL
base_url: OPENAI_BASE_URL
tags: [openai, mainstream]
- name: openhands
display_name: "OpenHands"
description: "Environnement de développement agentique open-source (ex-OpenDevin), avec entrées CLI et web."
category: coding-agent
website: https://github.com/All-Hands-AI/OpenHands
install:
type: pip
package: openhands
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: openhands
note: "une image Docker officielle existe aussi (allhandsai/openhands)"
tags: [agentic-dev, web-ui]
- name: cline
display_name: "Cline CLI"
description: "Agent autonome multi-modèle pour planning, édition de fichiers, exécution de commandes et browser use."
category: coding-agent
website: https://github.com/cline/cline
install:
type: npm
package: "@cline/cline"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: cline
tags: [multi-model, planning, browser]
- name: goose
display_name: "Goose"
description: "Agent local et extensible, conçu pour tourner on-device, intégration MCP."
category: local-first
website: https://github.com/aaif-goose/goose
install:
type: npm
package: "@aaif-goose/goose"
methods:
- type: binary
repo: aaif-goose/goose
binary: goose
note: "binaire depuis les releases GitHub"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: goose
tags: [local, mcp, on-device]
- name: aider
display_name: "Aider"
description: "Pair-programming par diffs/patches, excellent support git et multi-fichiers. Le vétéran du genre."
category: coding-agent
website: https://github.com/Aider-AI/aider
install:
type: uv
package: aider-install
methods:
- type: pip
package: aider-install
note: "via pip (plus lent)"
post_install: ["aider-install"]
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: aider
note: "aider-install installe aider dans ~/.local (style pipx)"
tags: [diffs, git, pair-programming]
- name: continue-cli
display_name: "Continue CLI"
description: "Extension terminal multi-modèle, focus local/privacy."
category: coding-agent
website: https://github.com/continuedev/continue
install:
type: npm
package: "@continuedev/continue"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: continue
tags: [multi-model, privacy]
- name: deep-agents-code
display_name: "Deep Agents Code"
description: "Agent coding terminal officiel LangChain (Deep Agents SDK). TUI, opérations fichiers, shell, subagents, mode headless, human-in-the-loop."
category: coding-agent
website: https://github.com/langchain-ai/deepagents
install:
type: pip
package: deepagents-code
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: deepagents-code
tags: [langchain, subagents]
- name: crush
display_name: "Crush"
description: "TUI agentic Go signé Charmbracelet, multi-provider, LSP-aware, riche UI terminal."
category: coding-agent
website: https://github.com/charmbracelet/crush
install:
type: go
package: github.com/charmbracelet/crush@latest
dependencies:
- { name: go, min_version: "1.22.0", install_hint: "https://go.dev/dl" }
run: crush
tags: [go, tui, lsp]
- name: qwen-code
display_name: "Qwen Code"
description: "Agent CLI officiel Alibaba pour les modèles Qwen coder (Apache-2.0)."
category: coding-agent
website: https://github.com/QwenLM/qwen-code
install:
type: npm
package: "@qwen-code/qwen-code"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: qwen-code
tags: [alibaba, qwen]
- name: grok-build
display_name: "Grok Build"
description: "Harness et TUI de coding officiel xAI, fullscreen, interactif à la souris (Apache-2.0)."
category: coding-agent
website: https://github.com/xai-org/grok-build
install:
type: npm
package: "@xai-org/grok-build"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: grok-build
platforms: [linux, macos]
tags: [xai, tui]
- name: kimi-cli
display_name: "Kimi CLI"
description: "Agent CLI officiel Moonshot AI, skills, MCP, intégration ACP IDE (Apache-2.0). Installé via uv (paquet PyPI kimi-cli)."
category: coding-agent
website: https://github.com/MoonshotAI/kimi-cli
install:
type: uv
package: kimi-cli
dependencies:
- { name: uv, min_version: "0.4.0", install_hint: "https://docs.astral.sh/uv" }
run: kimi
tags: [moonshot, skills, mcp]
- name: plandex
display_name: "Plandex"
description: "Agent CLI plan-first pour construire des features multi-fichiers avec étapes structurées et 2M tokens de contexte."
category: coding-agent
website: https://github.com/plandex-ai/plandex
install:
type: curl
script: "https://plandex.ai/install.sh"
dependencies:
- { name: curl, install_hint: "https://curl.se" }
- { name: sh, install_hint: "un shell POSIX" }
run: plandex
platforms: [linux, macos]
note: "développement au ralenti (dernier commit oct. 2025)"
tags: [plan-first, large-context]
- name: swe-agent
display_name: "SWE-agent"
description: "Agent pour résoudre de vrais issues de repo et tâches PR (workflows SWE-bench)."
category: coding-agent
website: https://github.com/SWE-agent/SWE-agent
install:
type: pip
package: swe-agent
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: sweagent
tags: [swe-bench, issues]
- name: open-interpreter
display_name: "Open Interpreter"
description: "Outil terminal qui exécute du code et des actions — faire des choses sur ma machine."
category: coding-agent
website: https://github.com/OpenInterpreter/open-interpreter
install:
type: pip
package: open-interpreter
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: interpreter
tags: [code-execution]
- name: mimo-code
display_name: "MiMo Code"
description: "Agent terminal officiel Xiaomi. TUI + mode non-interactif, MCP, skills, hooks, git worktrees, sessions reprenables."
category: coding-agent
website: https://github.com/XiaomiMiMo/MiMo-Code
install:
type: npm
package: "@xiaomi-mimo/cli"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: mimo
tags: [xiaomi, worktrees]
- name: smol-developer
display_name: "Smol Developer"
description: "Agent développeur embarquable qui génère des codebases entières depuis un prompt."
category: coding-agent
website: https://github.com/smol-ai/developer
install:
type: pip
package: smol-developer
dependencies:
- { name: python, min_version: "3.8.0", install_hint: "https://python.org" }
run: smol-developer
note: "projet inactif (dernier commit sept. 2023)"
tags: [codegen, pioneer]
- name: groq-code-cli
display_name: "Groq Code CLI"
description: "CLI léger propulsé par l'inférence ultra-rapide de Groq, outils/commandes extensibles."
category: coding-agent
website: https://github.com/build-with-groq/groq-code-cli
install:
type: npm
package: groq-code-cli
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: groq-code
tags: [groq, fast]
# ------------------------------------------------------- Trésors open source
- name: jcode
display_name: "jcode"
description: "Agent TUI Rust optimisé RAM et latence de démarrage (~28 MB/session). Mémoire agent, swarm mode, browser automation, MCP, 40+ providers avec OAuth intégré."
category: coding-agent
website: https://github.com/1jehuang/jcode
install:
type: binary
repo: 1jehuang/jcode
binary: jcode
run: jcode
tags: [rust, swarm, tui]
- name: agentty
display_name: "agentty"
description: "Agent terminal C++26, binaire statique 13.6 MB, cold start sub-milliseconde, zéro dépendance runtime, sandbox par défaut."
category: coding-agent
website: https://github.com/1ay1/agentty
install:
type: binary
repo: 1ay1/agentty
binary: agentty
run: agentty
config:
provider_default: openai
env_map:
api_key: AGENTTY_API_KEY
model: AGENTTY_MODEL
base_url: AGENTTY_BASE_URL
files:
- path: ~/.config/agentty/config.toml
keys: [model, base_url]
tags: [c++, static, sandbox]
- name: nullclaw
display_name: "NullClaw"
description: "Plus petit agent compatible OpenClaw, en Zig. 678 KB, ~1MB RAM, <2ms startup, 23+ providers, 18 channels."
category: assistant
website: https://github.com/nullclaw/nullclaw
install:
type: binary
repo: nullclaw/nullclaw
binary: nullclaw
run: nullclaw
tags: [zig, tiny, openclaw]
- name: nanocoder
display_name: "Nanocoder"
description: "Agent CLI local-first communautaire. BYO model (Ollama, OpenRouter, API OpenAI-compatible), tool calling natif + fallback XML, MCP, tools personnalisés."
category: local-first
website: https://github.com/Nano-Collective/nanocoder
install:
type: uv
package: nanocoder
methods:
- type: pip
package: nanocoder
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: nanocoder
tags: [local, ollama, byo-model]
- name: gptme
display_name: "gptme"
description: "Agent dans ton terminal avec agents persistants auto-modifiants, mémoire git-backed. Exécute du code, édite des fichiers, browse le web."
category: coding-agent
website: https://github.com/gptme/gptme
install:
type: uv
package: gptme
methods:
- type: pip
package: gptme
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: gptme
tags: [persistent, memory]
- name: devon
display_name: "Devon"
description: "Pair programmer open-source avec TUI, planification, exécution et debugging autonomes dans les workflows git."
category: coding-agent
website: https://github.com/entropy-research/Devon
install:
type: pip
package: devon-agent
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: devon
note: "projet inactif (dernier commit juil. 2024)"
tags: [tui, autonomous]
- name: ra-aid
display_name: "RA.Aid"
description: "Agent de coding autonome sur LangGraph avec pipeline research -> plan -> implement, intégration aider optionnelle."
category: coding-agent
website: https://github.com/ai-christianson/RA.Aid
install:
type: pip
package: ra-aid
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: ra-aid
note: "développement au ralenti (juin 2025)"
tags: [langgraph, pipeline]
- name: letta-code
display_name: "Letta Code"
description: "Agent CLI memory-first (ex-MemGPT). Mémoire persistante entre sessions, model-agnostique, skill learning, context repositories."
category: coding-agent
website: https://github.com/letta-ai/letta-code
install:
type: pip
package: letta-code
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: letta-code
tags: [memory, memgpt]
- name: dvalincode
display_name: "DvalinCode"
description: "Agent local-first provider-neutre orienté gouvernance : policy engine, egress réseau forcé, audit trail hash-chaîné, approval diff inline, Web GUI embarqué."
category: local-first
website: https://github.com/arthurpanhku/dvalincode
install:
type: binary
repo: arthurpanhku/dvalincode
binary: dvalincode
run: dvalincode
tags: [governance, audit]
- name: san
display_name: "San"
description: "Runtime Go provider-agnostique pour agents spécialisés. Exécute les skills/plugins/MCP Claude Code sans modification, 9+ providers, mémoire auto-évolutive."
category: coding-agent
website: https://github.com/genai-io/san
install:
type: binary
repo: genai-io/san
binary: san
run: san
tags: [go, skills, runtime]
- name: zap
display_name: "Zap"
description: "Agent TUI Rust skill-first, injecte seulement le contexte nécessaire. Code-indexé SQLite, MCP, modèles locaux via LM Studio."
category: coding-agent
website: https://github.com/zap-coding-agent/zap-coding-agent
install:
type: binary
repo: zap-coding-agent/zap-coding-agent
binary: zap
run: zap
tags: [rust, tui, sqlite]
- name: smelt
display_name: "Smelt"
description: "Agent TUI Rust multi-provider, 4 modes (Normal/Plan/Apply/Yolo), permissions granulaires, subagents parallèles, keybinds vim, mode headless."
category: coding-agent
website: https://github.com/leonardcser/smelt
install:
type: cargo
package: smelt
dependencies:
- { name: cargo, min_version: "1.75.0", install_hint: "https://rustup.rs" }
run: smelt
tags: [rust, tui, vim]
- name: grinta
display_name: "Grinta"
description: "Agent terminal local-first pour exécution autonome long-horizon : état durable, récupération, gestion de contexte, orchestration d'outils, LSP/DAP."
category: local-first
website: https://github.com/josephsenior/Grinta-Coding-Agent
install:
type: pip
package: grinta
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: grinta
tags: [long-horizon, lsp]
- name: waveloom
display_name: "Waveloom"
description: "Agent Go avec TUI Bubble Tea, DeepSeek V4 prompt caching, 4 niveaux de compaction de contexte, 3 modes subagent (Fork/Cold/Explore)."
category: coding-agent
website: https://github.com/Menfre01/waveloom
install:
type: binary
repo: Menfre01/waveloom
binary: waveloom
run: waveloom
tags: [go, tui, deepseek]
- name: every-code
display_name: "Every Code"
description: "Fork de Codex CLI avec validation, automation, intégration browser, multi-agents, theming et orchestration multi-provider (OpenAI/Claude/Gemini)."
category: coding-agent
website: https://github.com/just-every/code
install:
type: git
repo: "https://github.com/just-every/code"
build: ["npm", "install"]
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: "every code"
tags: [fork, codex, multi-provider]
- name: free-code
display_name: "Free Code"
description: "Fork de Claude Code sans télémétrie, guardrails retirés, toutes les features expérimentales activées."
category: coding-agent
website: https://github.com/paoloanzn/free-code
install:
type: git
repo: "https://github.com/paoloanzn/free-code"
build: ["npm", "install"]
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: free-code
tags: [fork, no-telemetry]
- name: claurst
display_name: "Claurst"
description: "Claude Code réécrit en Rust idiomatique, avec découvertes du leak source (KAIROS, buddy system)."
category: coding-agent
website: https://github.com/Kuberwastaken/claurst
install:
type: git
repo: "https://github.com/Kuberwastaken/claurst"
build: ["cargo", "build", "--release"]
binary_path: target/release/claurst
dependencies:
- { name: cargo, min_version: "1.75.0", install_hint: "https://rustup.rs" }
run: claurst
tags: [rust, rewrite]
- name: mistral-vibe
display_name: "Mistral Vibe"
description: "Assistant CLI Mistral pour interaction conversationnelle et éditions de repo."
category: coding-agent
website: https://github.com/mistralai/mistral-vibe
install:
type: npm
package: "@mistralai/mistral-vibe"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: mistral-vibe
tags: [mistral]
- name: amazon-q
display_name: "Amazon Q Dev CLI"
description: "Chat terminal agentic AWS pour build, debugging et DevOps en langage naturel."
category: coding-agent
website: https://github.com/aws/amazon-q-developer-cli
install:
type: npm
package: "@aws/amazon-q-developer-cli"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: q
tags: [aws, devops]
- name: lazycodex
display_name: "LazyCodex"
description: "Harness d'agent pour codebases complexes — package OmO comme distribution Codex. Mémoire projet, planning, exécution et complétion vérifiée à l'intérieur de Codex. L'équivalent de LazyVim pour Codex."
category: coding-agent
website: https://github.com/code-yeongyu/lazycodex
install:
type: npm
package: lazycodex-ai
post_install: ["lazycodex-ai", "install"]
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: "npx lazycodex-ai"
# Sonde PATH explicite : le premier mot de `run` est npx (interpréteur) —
# sans `detect`, npx sur le PATH suffirait à déclarer l'agent « external ».
detect: lazycodex-ai
note: "s'intègre dans Codex (nécessite @openai/codex)"
tags: [harness, codex, plan]
- name: gajae-code
display_name: "Gajae-Code"
description: "Harness de coding externe (gjc) — se connecte au plan que tu paies déjà (Claude, Codex, Cursor, Copilot...). Workflow plan-gated, réponses aux questions depuis le téléphone."
category: coding-agent
website: https://github.com/Yeachan-Heo/gajae-code
install:
type: npm
package: gajae-code
methods:
- type: bun
package: gajae-code
note: "via bun (plus rapide)"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: gjc
tags: [harness, subscription, telegram]
# ------------------------------------------------------------ Local-first
- name: openyak
display_name: "openyak"
description: "Agent desktop local-first, zéro compte, zéro télémétrie. Modèles locaux via Ollama/Rapid-MLX ou BYO."
category: desktop
website: https://github.com/openyak/openyak
installable: false
note: "application desktop — téléchargez-la depuis https://github.com/openyak/openyak/releases"
tags: [desktop, ollama]
- name: kun
display_name: "Kun"
description: "Workspace local-first desktop GUI + TUI pour coding, écriture, design, recherche et automation."
category: desktop
website: https://github.com/KunAgent/Kun
installable: false
note: "application desktop — https://github.com/KunAgent/Kun/releases"
tags: [desktop, workspace]
- name: ipollowork
display_name: "iPolloWork"
description: "Workspace AI local-first source-available, alternative à Codex et Claude Code pour code, fichiers, docs, sites web."
category: local-first
website: https://github.com/Devin-AXIS/iPolloWork
install:
type: git
repo: "https://github.com/Devin-AXIS/iPolloWork"
run: ipollowork
note: "étapes de build spécifiques au repo — consultez son README"
tags: [workspace, source-available]
- name: cc-haha
display_name: "cc-haha"
description: "Workspace desktop local-first cross-platform pour Claude Code/agents : multi-agent, git worktrees, diffs, skills marketplace."
category: desktop
website: https://github.com/NanmiCoder/cc-haha
installable: false
note: "application desktop — https://github.com/NanmiCoder/cc-haha/releases"
tags: [desktop, multi-agent]
- name: claw-empire
display_name: "claw-empire"
description: "Simulateur de bureau d'agents AI local-first — orchestre CLI, OAuth et API depuis un CEO Desk."
category: local-first
website: https://github.com/GreenSheep01201/claw-empire
install:
type: git
repo: "https://github.com/GreenSheep01201/claw-empire"
run: claw-empire
note: "étapes de build spécifiques au repo — consultez son README"
tags: [simulator, orchestration]
# ---------------------------------------------------------------- Shell AI
- name: aichat
display_name: "AIChat"
description: "All-in-one LLM CLI avec Shell Assistant, Chat-REPL, RAG, AI Tools & Agents et support fichiers/répertoires (-f). Idéal pour exécuter des micro-tâches shell via langage naturel (issue #94)."
category: shell-ai
website: https://github.com/sigoden/aichat
install:
type: binary
repo: sigoden/aichat
binary: aichat
run: aichat
tags: [shell, rust, multi-provider, rag, local, fast]
- name: ai-cli
display_name: "AI CLI"
description: "Traduit le langage naturel en commandes shell (OpenAI, OpenRouter ou serveur local) avec politique de sécurité (risk / certainty) avant exécution (issue #95)."
category: shell-ai
website: https://github.com/kriserickson/ai-cli
install:
type: binary
repo: kriserickson/ai-cli
binary: ai
run: ai
tags: [shell, go, safety, multi-provider]
- name: fabric
display_name: "Fabric"
description: "Framework open source pour augmenter les humains avec l'IA : patterns réutilisables (summarize, extract wisdom, analyse de contenu), CLI Go 43k+ stars (issue #95)."
category: shell-ai
website: https://github.com/danielmiessler/fabric
install:
type: go
package: github.com/danielmiessler/fabric/cmd/fabric@latest
dependencies:
- { name: go, min_version: "1.22.0", install_hint: "https://go.dev/dl" }
run: fabric
tags: [patterns, go, content]
- name: shell-ai
display_name: "Shell AI"
description: "Agent ReAct local-first propulsé par Ollama : zéro réseau, outils filesystem, exécution de commandes shell et intégration MCP (issue #95)."
category: shell-ai
website: https://github.com/nishant9083/shell-ai
install:
type: npm
package: "@shell-ai/cli"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: shell-ai
note: "nécessite Ollama et un modèle (ollama pull gpt-oss) — agent jeune (15 stars)"
tags: [shell, ollama, local, mcp, typescript]
- name: shellgpt
display_name: "ShellGPT"
description: "Outil de productivité CLI propulsé par LLM (GPT-5…) : génère et exécute des commandes shell, écrit du code et des docs, 12k+ stars (issue #95)."
category: shell-ai
website: https://github.com/TheR1D/shell_gpt
install:
type: pip
package: shell-gpt
dependencies:
- { name: python, min_version: "3.10.0", install_hint: "https://python.org" }
run: sgpt
tags: [shell, python, multi-provider]
# ----------------------------------------------------- Écosystème OpenClaw
- name: openclaw
display_name: "OpenClaw"
description: "L'assistant personnel AI local d'origine. CLI avec onboarding, skills, outils, multi-channel (WhatsApp/Slack/Discord)."
category: assistant
website: https://github.com/openclaw/openclaw
install:
type: npm
package: "@openclaw/openclaw"
dependencies:
- { name: node, min_version: "22.0.0", install_hint: "https://nodejs.org" }
run: openclaw
tags: [assistant, multi-channel, skills]
- name: nanobot
display_name: "nanobot"
description: "Réécriture ultra-légère (~4000 lignes Python) d'OpenClaw. Tool use, mémoire persistante, tâches planifiées, Telegram/Discord/WhatsApp."
category: assistant
website: https://github.com/HKUDS/nanobot
install:
type: git
repo: "https://github.com/HKUDS/nanobot"
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: "python -m nanobot"
# Sonde PATH explicite : le premier mot de `run` est python (interpréteur) —
# sans `detect`, /usr/bin/python suffirait à déclarer l'agent « external ».
detect: nanobot
tags: [python, lightweight]
- name: zeroclaw
display_name: "ZeroClaw"
description: "Runtime d'agent AI Rust autonome, architecture pluggable trait-driven, tourne sur matériel minimal (<5MB RAM), exécution sandboxée, recherche hybride."
category: assistant
website: https://github.com/zeroclaw-labs/zeroclaw
install:
type: binary
repo: zeroclaw-labs/zeroclaw
binary: zeroclaw
run: zeroclaw
tags: [rust, minimal, sandbox]
- name: nanoclaw
display_name: "NanoClaw"
description: "Alternative légère security-first à OpenClaw, agents en containers Apple/Docker, sandbox, basé sur Anthropic Agents SDK."
category: assistant
website: https://github.com/gavrielc/nanoclaw
install:
type: git
repo: "https://github.com/gavrielc/nanoclaw"
dependencies:
- { name: python, min_version: "3.9.0", install_hint: "https://python.org" }
run: nanoclaw
tags: [security, sandbox]
- name: picoclaw
display_name: "PicoClaw"
description: "Assistant personnel AI ultra-léger en Go, <10MB RAM, tourne sur du matériel à 10 dollars."
category: assistant
website: https://github.com/sipeed/picoclaw
install:
type: binary
repo: sipeed/picoclaw
binary: picoclaw
run: picoclaw
tags: [go, tiny]
- name: ironclaw
display_name: "IronClaw"
description: "OpenClaw réécrit en Rust par NEAR AI. Sandbox WASM, permissions capability-based, défense anti prompt-injection."
category: assistant
website: https://github.com/nearai/ironclaw
install:
type: git
repo: "https://github.com/nearai/ironclaw"
dependencies:
- { name: cargo, min_version: "1.75.0", install_hint: "https://rustup.rs" }
run: ironclaw
tags: [rust, wasm, security]
- name: moltis
display_name: "Moltis"
description: "Alternative Rust-native à OpenClaw, zéro code unsafe, 2300+ tests, voice I/O intégré, MCP, multi-channel."
category: assistant
website: https://github.com/moltis-org/moltis
install:
type: git
repo: "https://github.com/moltis-org/moltis"
dependencies:
- { name: cargo, min_version: "1.75.0", install_hint: "https://rustup.rs" }
run: moltis
tags: [rust, voice, mcp]
# ---------------------------------------------------------- Propriétaires
- name: cursor-cli
display_name: "Cursor CLI"
description: "Agent CLI officiel Cursor, shell mode, headless/CI, agents parallèles, multi-modèles."
category: proprietary
website: https://cursor.com/cli
install:
type: npm
package: cursor-agent
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: cursor
note: "paquet npm communautaire — l'officiel est livré avec l'app Cursor"
tags: [cursor, proprietary]
- name: warp
display_name: "Warp"
description: "Terminal moderne avec mode agent AI intégré — comprend les tâches, exécute les commandes, édite les fichiers."
category: desktop
website: https://github.com/warpdotdev/Warp
installable: false
note: "application desktop — téléchargez-la depuis https://warp.dev"
tags: [terminal, desktop]
- name: github-copilot
display_name: "GitHub Copilot CLI"
description: "CLI agentic GitHub pour workflows repo/PR/issue, suggestions de commandes, automation headless."
category: proprietary
website: https://github.com/github/copilot-cli
install:
type: npm
package: "@github/copilot"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: copilot
tags: [github, copilot]
- name: devin
display_name: "Devin"
description: "Ingénieur logiciel AI autonome de Cognition, accès shell/browser complet, code auto-réparant, collaboration PR."
category: saas
website: https://devin.ai
installable: false
note: "SaaS — pas d'installation locale; utilisez https://devin.ai (web ou API)"
tags: [saas, autonomous]
- name: factory-droid
display_name: "Factory Droid"
description: "Agent CLI multi-modèle de Factory, #1 sur Terminal-Bench, droids spécialisés, mode headless CI."
category: proprietary
website: https://github.com/Factory-AI/factory
install:
type: npm
package: "@factory-ai/factory"
dependencies:
- { name: node, min_version: "18.0.0", install_hint: "https://nodejs.org" }
run: droid
tags: [factory, benchmark, ci]
- name: amp
display_name: "Amp"
description: "Agent AI de Sourcegraph pour implémenter des tâches sur de vraies codebases."
category: saas
website: https://sourcegraph.com/amp
installable: false
note: "SaaS — via https://sourcegraph.com/amp"
tags: [saas, sourcegraph]
- name: junie-cli
display_name: "Junie CLI"
description: "Agent CLI LLM-agnostique JetBrains (EAP), GPT-5/Claude/Gemini/Grok, plan mode, headless CI/CD."
category: saas
website: https://junie.jetbrains.com
installable: false
note: "SaaS — via https://junie.jetbrains.com (installation gérée par JetBrains)"
tags: [saas, jetbrains]
- name: cortex-code-cli
display_name: "Cortex Code CLI"
description: "Agent CLI data-native de Snowflake pour pipelines, analytics et apps AI avec gouvernance enterprise."
category: saas
website: https://www.snowflake.com/en/product/cortex-code/
installable: false
note: "SaaS — via Snowflake (https://www.snowflake.com)"
tags: [saas, snowflake, data]
- name: tabnine-cli
display_name: "Tabnine CLI"
description: "Assistant coding AI agentic de Tabnine, distribué en container Docker, licence enterprise."
category: saas
website: https://docs.tabnine.com/main/getting-started/tabnine-cli
installable: false
note: "distribué en container Docker (licence enterprise) — voir https://docs.tabnine.com"
tags: [saas, docker]
- name: mentat-cli
display_name: "Mentat CLI"
description: "Agent CLI cloud-native pour gérer des agents Mentat distants depuis le terminal, détection auto repo/branch."
category: saas
website: https://mentat.ai/docs/cli
installable: false
note: "SaaS — via https://mentat.ai/docs/cli"
tags: [saas, remote-agents]
- name: pool
display_name: "Pool"
description: "Agent terminal de Poolside propulsé par les modèles Laguna, TUI + headless (pool exec), ACP client/server."
category: proprietary
website: https://github.com/poolsideai/pool
install:
type: binary
repo: poolsideai/pool
binary: pool
run: pool
note: "EULA propriétaire"
tags: [poolside, laguna]
- name: ante
display_name: "Ante"
description: "Binaire Rust 15 MB (research preview), architecture client-daemon, TUI + headless, inférence GGUF offline embarquée (llama.cpp), 12+ providers."
category: proprietary
website: https://github.com/AntigmaLabs/ante-preview
install:
type: binary
repo: AntigmaLabs/ante-preview
binary: ante
run: ante
tags: [offline, gguf, daemon]