Agents Machine

Skills Editor

Create and manage reusable AI skills.

The Skills Editor provides a visual interface for creating, testing, and managing skills. Skills are reusable AI tools that can be invoked by name and can inject secrets from the vault.

Creating a Skill

Define the Skill Fill in the basics: name (e.g., code-reviewer),

description, instructions (Markdown system prompt), and tags.

Configure Options Optionally set a model override, attach an

agent prompt, and define env var requirements with vault mappings.

Test Use the built-in sandbox to test the skill with a sample message.

Testing does NOT save — it runs in an isolated context.

Secret Injection

Skills can access vault secrets through environment mapping:

Required env vars: TELEGRAM_BOT_TOKEN, CHAT_ID
Env mapping:
  TELEGRAM_BOT_TOKEN → TELEGRAM_BOT_TOKEN (vault key)
  CHAT_ID → TELEGRAM_CHAT_ID (vault key)

At runtime, secrets are injected as process.env.TELEGRAM_BOT_TOKEN — no manual retrieval needed. The vault handles encryption/decryption transparently.

Examples

Skill Instructions
You are a senior code reviewer. Analyze the provided code for:
- Code quality and readability
- Security vulnerabilities
- Performance issues
- TypeScript best practices

Output format:
1. Summary (1-2 sentences)
2. Issues found (severity: critical/warning/info)
3. Suggestions for improvement
Skill Instructions
You send notifications via Telegram.
Use process.env.TELEGRAM_BOT_TOKEN and process.env.CHAT_ID.
Format messages with Markdown for readability.

Env mapping: TELEGRAM_BOT_TOKEN → vault key, CHAT_ID → vault key TELEGRAM_CHAT_ID

Invoking Skills

💬 Chat

"Invoke the code-reviewer skill on this function"

🔌 MCP Tools

invoke_skill tool from your IDE.

🤖 Agents

Agents can call skills during execution.

🔄 Pipelines

Skills can be nodes in automation pipelines.

On this page