formula-authoring
npx machina-cli add skill a5c-ai/babysitter/formula-authoring --openclawFormula Authoring
Overview
Create and manage TOML-based Formula templates that define repeatable multi-step workflows. Formulas are cooked into Protomolecules (frozen, ready to instantiate) and then activated as Molecules (durable, checkpoint-able workflows).
When to Use
- Defining repeatable workflow templates
- Creating multi-step processes with variable binding
- Building workflows that must survive restarts
- When NDI (Nondeterministic Idempotence) is needed
Process
- Define formula steps and variables in TOML format
- Validate formula structure and dependencies
- Cook into protomolecule (resolve variables, freeze)
- Test by instantiating a trial molecule
- Register in the formula library for reuse
Formula Lifecycle
Formula (TOML template) -> Protomolecule (frozen) -> Molecule (active, durable)
Key Concepts
- Formula: TOML-based workflow template with variables
- Protomolecule: Frozen template ready to instantiate
- Molecule: Active durable workflow surviving restarts
- NDI: Nondeterministic Idempotence - useful outcomes from unreliable processes
Tool Use
Invoke via babysitter process: methodologies/gastown/gastown-molecule
Source
git clone https://github.com/a5c-ai/babysitter/blob/main/plugins/babysitter/skills/babysit/process/methodologies/gastown/skills/formula-authoring/SKILL.mdView on GitHub Overview
Formula Authoring lets you create TOML-based workflow templates that become Protomolecules (frozen templates) and, when instantiated, active Molecules (durable workflows). It enables repeatable, restart-safe multi-step processes and supports NDI for unreliable environments.
How This Skill Works
Define steps and variables in TOML, then validate the structure and dependencies. Cook the approved template into a Protomolecule, test by instantiating a trial Molecule, and finally register the formula in the library for reuse.
When to Use It
- Defining repeatable workflow templates for multiple runs
- Creating multi-step processes with variable bindings
- Building durable workflows that survive restarts
- When nondeterministic idempotence (NDI) is required for reliability
- Centralizing workflow templates by registering them in a formula library
Quick Start
- Step 1: Define the TOML-based steps and variables for your workflow
- Step 2: Validate the TOML and cook it into a Protomolecule
- Step 3: Test with a trial Molecule and register in the library
Best Practices
- Keep TOML schemas stable and well-documented
- Explicitly validate formula structure and dependencies before cooking
- Test with a trial Molecule to verify runtime behavior
- Version and tag protomolecules to track changes over time
- Document variable bindings, defaults, and error handling
Example Use Cases
- Data ingestion pipeline template that resolves credentials and transforms data
- Order processing workflow with multi-step checks and checkpointing
- ETL job with configurable source, destination, and retry logic
- Sensor data aggregation with deterministic outcomes across restarts
- QA automation pipeline with reproducible steps and versioned templates
