lynxai-team/agent-smith/create-agent

Claude Code and agent skill by lynxai-team/agent-smith

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

use when asked to create, define, or modify an agent in YAML for Agent Smith, optionally exposes it as a subagent tool

Install

npx vskill@latest install lynxai-team/agent-smith/create-agent

SKILL.md

Create an Agent in YAML

Define, register, and verify an AI agent from a YAML file in the Agent Smith framework.

Workflow

  1. Ensure a registered features directory (first time only) — edit the conf file to add your features/ root path, then run lm conf. The conf file lists the directories the framework scans for feature files; it is the source of truth for feature paths.
  2. Write the agent YAML — create <name>.yml in features/agents/. The filename becomes the agent's default name (e.g., my-agent.yml → agent my-agent). Fill in the required fields (description, prompt, model) plus any optional ones from the field reference below.
  3. Register — run lm update. This triggers feature discovery: it scans registered directories, parses agent metadata, and upserts into the agent SQLite table (and creates ToolSpec entries in the tool table).
  4. (Optional) Expose as a subagent — add a tool: section to the YAML so other agents can call it, then run lm update again.
  5. Verify:
bash
lm agents # agent appears in the list
lm my-agent "Hello" # runs the agent interactively

Field reference

Required: name (from filename), prompt (system prompt template; {prompt} = user input), description, model (e.g., qwen4b).

Optional: category (UI navigation), template (system prompt, stop sequences), inferParams (sampling: temperature, top_p, top_k, min_p, repetition_penalty), variables (input variable definitions with required:/optional:), toolsList (references to registered tools by name), tools (inline tool definitions), models (alternative models for UI selection), shots (few-shot examples), skills, workflow (pre/post steps), mcp.

Examples

Minimal agent:

yaml
description: "My first agent"
category: myagents/general
prompt: |-
User query: {prompt}
model: qwen4b

Full annotated example:

yaml
# features/agents/my-coder.yml
description: "A coding agent with file and shell access"
category: development/coding
prompt: |-
{prompt}
template:
system: |-
You are a coding agent. Follow instructions precisely.
Report back with a short summary when done.
model: qwen35b
inferParams:
min_p: 0
top_k: 20
top_p: 0.85
temperature: 0.6
repetition_penalty: 1
variables:
required:
workspace:
description: The local directory path where to operate
toolsList:
- read-file
- write-file
- edit-search-replace
- shell

Subagent (callable by other agents) — add a tool: section defining the interface:

yaml
# features/agents/my-search.yml
tool:
name: my-search
description: Perform web search and return results
arguments:
prompt:
description: The search query or question
required: true
description: "A web search agent"
prompt: |-
Research the following: {prompt}
model: qwen4b
toolsList:
- ddsearch
- openpage

Other agents reference it by its tool name in their toolsList (e.g., - my-search), then run lm update.

Gotchas

  • Simply running lm conf without first editing the conf file will NOT add a new features directory — the conf file is the source of truth for feature paths.
  • Every change to an agent's tool list requires re-registration: after adding, removing, or modifying entries in toolsList (or inline tools), run lm update again — the agent will not see new tools until then.
  • A subagent is NOT usable until registered: the tool: section only defines the interface; lm update creates the ToolSpec entry in the tool table that makes it callable. Without this step, parent agents referencing it in their toolsList get a "tool not found" error.
  • Renaming the YAML file changes the agent's default name — keep filenames stable or update references in other agents' toolsList.

References (load when needed)

  • Full schema and field-by-field guide: agent-smith/docsite/public/doc/reference/1.howto-agents.md
  • Minimal example agent: agent-smith/examples/features/agents/general-agent.yml
  • Complex real-world example with template, variables, tools: lynx-coder/dist/agents/lx-coder.yml
  • Feature discovery pipeline internals: agent-smith/docsite/public/doc/libraries/core/3.feature-discovery.md

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/lynxai-team/agent-smith/create-agent/badge)](https://verified-skill.com/skills/lynxai-team/agent-smith/create-agent)

Something wrong with this skill or its page? Tell us.