Skip to main content
The signa skill teaches coding agents (Claude Code, Cursor, Codex, Copilot, and any other agent that supports skills) how the Signa API actually works: authentication, the response envelope, search and filter conventions, the watch query DSL, pagination, and error handling. With the skill installed, an agent writes correct integration code on the first try instead of guessing field names and inventing endpoints.

​
Install

Run this from the project where you’re building against Signa: 
npx skills add signa-so/skills
That installs the skill into your project (for Claude Code, it lands in .claude/skills/signa/), where your agent picks it up automatically. The source lives at github.com/signa-so/skills.

​
What it covers

  • Conventions agents otherwise get wrong: Bearer auth, the list envelope, comma-separated array params, _gte/_lt date operators, POST filters nested under filters, idempotency keys, cursor pagination, and 429 handling.
  • A task-to-endpoint map: which endpoint to reach for by job to be done (clearance, portfolio pulls, monitoring, deadlines).
  • The monitoring surface: the watch query DSL and webhook signature verification, the area agents hallucinate most without it.
  • Live-docs awareness: the skill tells the agent to fetch current details from these docs (every page serves raw Markdown, and llms.txt is the index), so it stays accurate as the API evolves.

​
Skill or MCP server?

They complement each other. The MCP server gives an agent hosted tools to act on Signa conversationally (search, create watches) without writing code. The skill gives a coding agent the knowledge to write code against the API (scripts, backends, scheduled jobs). If your agent is producing an integration, install the skill; if it’s answering questions or operating your account, connect MCP. Many teams use both.