What problem does it solve?
This Skill prevents runtime surprises in Conductor workflow YAML by documenting the real execution contracts that templates, routing conditions, and agent outputs must follow.
Core Features & Use Cases
- Agent output shape correctness: ensure scripts, LLM agents, and human gates expose the fields your templates and routes expect, including provider-specific raw-string fallbacks.
- Routing and template evaluation sanity: understand how
when: conditions are evaluated (including current-agent output shorthand) and why missing catch-alls cause route failures or silent truncation.
- Schema, parsing, and provider/model namespace safety: enforce the rule that LLM agents must declare
output: schemas for named-field access, and select valid models within provider namespaces.
- Operational footgun coverage: avoid cross-platform subprocess issues, env-var parsing collisions, group/parallel output key differences, and iteration-limit pitfalls that only show up at execution time.
Quick Start
Use this skill to diagnose a failing workflow by checking whether the referenced template fields match the producing agent’s declared output shape and schema, then verify the route table includes an explicit match (or intended terminal) for every possible output value.
