agent-action-handshake

Resolve FSM action and agent prompt mismatches in Friday workflow runtimes.

97|5|Updated Apr 23, 2026
One-click install
npx skills add https://github.com/friday-platform/friday-studio --skill agent-action-handshake
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: agent-action-handshake
Source: https://github.com/friday-platform/friday-studio/tree/main/packages/system/skills/agent-action-handshake
Command: npx skills add https://github.com/friday-platform/friday-studio --skill agent-action-handshake

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill eliminates the frequent misdiagnosis of platform bugs and empty-output failures caused by gaps between how FSM actions are authored and what agent prompts expect, ensuring your Friday workflows run reliably without unexpected runtime errors.

Core Features & Use Cases

  • Invocation Path Reference: Clear breakdown of the four FSM-to-agent invocation kinds (inline LLM, agent-to-LLM, agent-to-atlas, agent-to-user) and their unique runtime handling rules.
  • Contract Enforcement: Documents auto-injected tools, outputTo/inputFrom semantics, and complete tool requirements to prevent common output failures.
  • Use Case: Use this when authoring multi-step Friday workflows that chain agent outputs between steps, or debugging why an agent step is not returning expected results to the FSM.

Quick Start

Consult the agent-action-handshake skill whenever you are building or troubleshooting FSM actions that invoke agents to validate handshake compliance and resolve output issues.

Frequently Asked Questions about agent-action-handshake

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
Why does my Friday workflow agent step return empty output to the FSM?

Empty output in Friday workflow agent steps usually stems from mismatches between FSM action authoring and agent prompt expectations, specifically missing outputTo mappings or complete tool requirements in the runtime handshake contract.

How do I configure outputTo and inputFrom mappings for chained agent workflows?

You configure outputTo and inputFrom mappings in your FSM action definitions to chain agent outputs across multi-step state machine workflows, ensuring the runtime auto-injection contract passes documents correctly between sequential agent steps.

What is the FSM-agent handshake contract in Friday workflow runtimes?

The FSM-agent handshake contract defines how auto-injected platform tools, the complete tool, and input/output document chaining rules govern communication between FSM actions and agent prompts to ensure consistent workflow execution.

How do the different FSM-to-agent invocation kinds affect runtime handling?

Friday workflows support four invocation kinds—inline LLM, agent-to-LLM, agent-to-atlas, and agent-to-user—each with unique runtime handling rules that dictate how agent prompts receive inputs and return outputs to the state machine.

When should I check the agent-action-handshake contract to debug workflow failures?

You should check the handshake contract when debugging multi-step Friday workflows that chain agent outputs, especially when an agent step is not returning expected results or you suspect a misdiagnosed platform bug in the FSM runtime.

Do I need to manually inject platform tools when authoring FSM agent actions?

No, the Friday runtime handles auto-injection of platform tools based on the invocation kind, but your FSM action configuration and agent prompts must align with the defined output contract to prevent output failures and communication mismatches.