designing-workflow-skills

Featured

Guides the design and structuring of workflow-based Claude Code skills with multi-step phases, decision trees, subagent delegation, and progressive disclosure. Use when creating skills that involve sequential pipelines, routing patterns, safety gates, task tracking, phased execution, or any multi-step workflow. Also applies when reviewing or refactoring existing workflow skills for quality.

AI & Automation 8,273 stars 715 forks Updated 6 days ago CC-BY-SA-4.0

Install

View on GitHub

Quality Score: 93/100

Stars 20%
100
Recency 20%
100
Frontmatter 20%
70
Documentation 15%
100
Issue Health 10%
50
License 10%
100
Description 5%
100

Skill Content

# Designing Workflow Skills Build workflow-based skills that execute reliably by following structural patterns, not prose. ## Essential Principles <essential_principles> <principle name="description-is-the-trigger"> **The `description` field is the only thing that controls when a skill activates.** Claude decides whether to load a skill based solely on its frontmatter `description`. The body of SKILL.md — including "When to Use" and "When NOT to Use" sections — is only read AFTER the skill is already active. Put your trigger keywords, use cases, and exclusions in the description. A bad description means wrong activations or missed activations regardless of what the body says. "When to Use" and "When NOT to Use" sections still serve a purpose: they scope the LLM's behavior once active. "When NOT to Use" should name specific alternatives: "use Semgrep for simple pattern matching" not "not for simple tasks." </principle> <principle name="numbered-phases"> **Phases must be numbered with entry and exit criteria.** Unnumbered prose instructions produce unreliable execution order. Every phase needs: - A number (Phase 1, Phase 2, ...) - Entry criteria (what must be true before starting) - Numbered actions (what to do) - Exit criteria (how to know it's done) </principle> <principle name="tools-match-executor"> **Tools must match the executor.** Skills use `allowed-tools:` in frontmatter. Agents use `tools:` in frontmatter. Subagents get tools from their `subagent_type`. Neve...

Details

Author
trailofbits
Repository
trailofbits/skills
Created
10 months ago
Last Updated
6 days ago
Language
Python
License
CC-BY-SA-4.0

Similar Skills

Semantically similar based on skill content — not just same category

AI & Automation Listed

designing-workflow-skills

Design, refactor, package, merge, or restructure runtime-neutral workflow skills whose multi-step behavior spans executable coordination, validated skill settings, event-triggered guards, MCP or external-tool services, delegated workers, and explicit command entrypoints. Use when workflows need routing, phases, concurrency, dependencies, resume, bounded loops, configurable behavior, safety gates, progressive disclosure, validated adapters, or durable failure recovery, when bundling skills and components for host discovery and distribution, when absorbing or consolidating one or many skills into one target without losing capabilities or copying host-specific syntax into core guidance, and when auditing or safely moving the files of a skill, bundle, or the repository holding them without breaking discovery, imports, builds, tests, or packaging.

0 Updated 3 weeks ago
clearlane
AI & Automation Listed

dynamic-workflow-patterns

Pattern taxonomy, agent role combinations, model routing, unit-of-work sizing, and resilience discipline for Claude Code dynamic workflows. ALWAYS load this skill before authoring or running any Workflow tool script, and ALWAYS load it when the user mentions "workflow" or "ultracode" in any form -- or when the task calls for multi-agent orchestration such as fan-out, tournaments, adversarial verification, triage at scale, ranking large lists, deep verification of claims, or root-cause hunting; do not hand-roll a workflow from memory when this skill applies.

7 Updated 5 days ago
alex-feel
AI & Automation Listed

claude-mastery-expert

This skill should be activated when the user wants to build an agent, design tools for an agent, write skills, structure a CLAUDE.md, optimize prompt caching, set up worktrees, use subagents, design verification loops, write hooks, structure a project for Claude Code, or improve their Claude Code workflow. Relevant when the user says "how should I structure this", "write a skill for", "design this agent", "optimize for caching", "set up verification", "use subagents", "parallel development", "write a CLAUDE.md", "improve my workflow", "design tools", "build an agent", "context engineering", or "autonomous research".

11 Updated 1 weeks ago
johnkozaris