How do I install the Agent Framework skill?

Run npx skillmds@latest add hummingbot/agent-framework in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Agent Framework safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Agent Framework?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Agent Framework free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Agent Framework?

hummingbot (@hummingbot) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Agent Framework by hummingbot — Agent Skill

Name: Agent Framework
Author: hummingbot

File contents

Agent Framework — Interaction Playbook

Tool map

Intent	Tool
Read an agent's journal	`trading_agent_journal_read(agent="<slug>", limit=N)`
Write to an agent's journal	`trading_agent_journal_write(agent="<slug>", entry="...")`
List all agents	`manage_agents(action="list")`
Get one agent's config/identity	`manage_agents(action="get", agent="<slug>")`
List strategies an agent owns	`manage_strategies(action="list", agent="<slug>")`
Get a strategy detail	`manage_strategies(action="get", agent="<slug>", strategy="<name>")`
List running instances	`control_agent(action="list")`
Start an agent instance	`control_agent(action="start", agent="<slug>")`
Stop a running instance	`control_agent(action="stop", instance_id="<id>")`
Pause / resume	`control_agent(action="pause"/"resume", instance_id="<id>")`
Query available LLM models	`get_available_models()`
Read an agent's memory	`manage_memory(action="list")` then `manage_memory(action="read", name="...")` with `agent="<slug>"`
Read an agent's skills	`manage_skill(action="list", agent="<slug>")`
Run a routine belonging to an agent	`manage_routines(action="run", name="<routine>")`

Steps

Identify the agent slug from the [AGENTS] index or from manage_agents(action="list").
Pick the right tool from the map above. One tool call usually suffices — don't chain five tools when one answers the question.
Journal reads — pass limit to cap output (default can be large). Filter by date if the agent has been running a while.
Controlling instances — control_agent(action="list") first to get instance IDs before stop/pause/resume.
Strategies vs instances — strategies are the recipe (what to do); instances are the running process. Listing strategies answers "what is this agent designed to do?"; listing instances answers "is it running right now?".
Memory/skills are per-agent — always pass agent="<slug>" when reading another agent's memory or skills, otherwise you'll read Condor's own library.

Do NOT use this skill for

Creating or deleting agents → agent_builder skill
Authoring or editing routines → routine_cookbook skill + background delegate
Building or editing strategies (the playbook code) → strategy_builder skill

---name: agent-frameworkdescription: Interact with the live agent runtime — journals, strategies, instances, models, memory, skills, and routines.---## Agent Framework — Interaction Playbook### Tool map| Intent | Tool ||---|---|| Read an agent's journal | `trading_agent_journal_read(agent="<slug>", limit=N)` || Write to an agent's journal | `trading_agent_journal_write(agent="<slug>", entry="...")` || List all agents | `manage_agents(action="list")` || Get one agent's config/identity | `manage_agents(action="get", agent="<slug>")` || List strategies an agent owns | `manage_strategies(action="list", agent="<slug>")` || Get a strategy detail | `manage_strategies(action="get", agent="<slug>", strategy="<name>")` || List running instances | `control_agent(action="list")` || Start an agent instance | `control_agent(action="start", agent="<slug>")` || Stop a running instance | `control_agent(action="stop", instance_id="<id>")` || Pause / resume | `control_agent(action="pause"/"resume", instance_id="<id>")` || Query available LLM models | `get_available_models()` || Read an agent's memory | `manage_memory(action="list")` then `manage_memory(action="read", name="...")` with `agent="<slug>"` || Read an agent's skills | `manage_skill(action="list", agent="<slug>")` || Run a routine belonging to an agent | `manage_routines(action="run", name="<routine>")` |### Steps1. **Identify the agent slug** from the [AGENTS] index or from `manage_agents(action="list")`.2. **Pick the right tool** from the map above. One tool call usually suffices — don't chain five tools when one answers the question.3. **Journal reads** — pass `limit` to cap output (default can be large). Filter by date if the agent has been running a while.4. **Controlling instances** — `control_agent(action="list")` first to get instance IDs before stop/pause/resume.5. **Strategies vs instances** — strategies are the *recipe* (what to do); instances are the *running process*. Listing strategies answers "what is this agent designed to do?"; listing instances answers "is it running right now?".6. **Memory/skills are per-agent** — always pass `agent="<slug>"` when reading another agent's memory or skills, otherwise you'll read Condor's own library.### Do NOT use this skill for- Creating or deleting agents → `agent_builder` skill- Authoring or editing routines → `routine_cookbook` skill + background delegate- Building or editing strategies (the playbook code) → `strategy_builder` skill

Frequently asked questions

Agent Framework