What problem does it solve?
When a Friday agent session fails with a runtime error, manually parsing error messages and tracing configuration contracts to find a fix is time-consuming and error-prone. This skill eliminates that friction by mapping every common runtime error pattern directly to its precise one-line fix.
Core Features & Use Cases
- Comprehensive error coverage: Handles all standard Friday runtime failure patterns including missing complete tool calls, empty agent outputs, schema validation failures, invalid job/signal config rejections, and signal payload mismatches.
- Actionable fixes: Each error entry includes a clear, specific fix aligned with Friday's official configuration and agent handshake contracts, plus cross-references to deeper documentation for context.
- Guardrails: Explicitly calls out common missteps to avoid, such as retrying contract violations without fixing the root cause or incorrectly migrating workspace patterns to work around prompt errors.
- Use case: For example, if your agent session fails with the error "LLM action with outputTo '<doc>' did not call complete", this skill immediately tells you to update the agent's prompt to explicitly invoke the complete tool with the required response argument.
Quick Start
Use the debugging-runtime-errors skill to fix the runtime error from your most recent failed Friday agent session.
