I'mBoardDocs
Recipes

Recipes

Pre-built, machine-readable YAML recipes that orchestrate common I'mBoard agent workflows — financial updates, pipeline syncs, and investor updates.

Recipes are versioned, machine-readable specifications that describe how an AI agent should orchestrate a multi-step workflow against the I'mBoard API. Each recipe is the contract between an agent and the product: point your agent at a recipe and it knows which endpoints to call, in what order, with what inputs — without you hand-writing API calls.

How recipes are published

Each recipe ships in two forms:

  • Human-readable page (these docs) — the use case, the YAML, and an example agent prompt.
  • Machine-readable JSON — served at /api/recipe/{name}/ and indexed at /api/recipe-index/index.json so agent crawlers can discover the full library.

The YAML source lives in the docs repo under content/recipes/{name}.yaml. When a recipe changes, agents pick up the new behavior on their next run.

Available recipes

RecipePersonaCadence
Monthly Investor UpdateFounder's agentMonthly
Board Financial Update (/api/recipe/financial-update/)CFO's agentWeekly
Sales Pipeline Update (/api/recipe/sales-pipeline-update/)VP Sales's agentWeekly

Recipe schema

Every recipe shares the same shape:

  • name, title, version, description — identity and summary.
  • triggerschedule and a human description of when to run.
  • inputs — typed parameters the agent must collect (name, type, description, required).
  • sources — where the agent can pull each input from.
  • steps — ordered actions, each with an action, description, and optional api_call.
  • output — the resulting resource schema and its fields.

When a step depends on an endpoint that does not exist yet, the recipe says so explicitly in a note and instructs the agent to prepare the payload for a human rather than call a fabricated endpoint.

On this page