How do I install the Workflow Work skill?

Run npx skillmds@latest add andrewdryga/workflow-work in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Workflow Work safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Workflow Work?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Workflow Work free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Workflow Work?

AndrewDryga (@andrewdryga) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Workflow Work by andrewdryga — Agent Skill

Name: Workflow Work
Rating: 4.5 (4 reviews)
Author: AndrewDryga

File contents

Complete the implementation

Follow the root and touched project AGENTS.md. The user's request or accepted plan defines the complete deliverable. Use workflow-spec when design is needed, then continue under the existing authorization.

Work in coherent steps. Give concise updates about findings and the next check. Use Elixir context/testing skills for Portal, go-engineer for runner/MCP, and security-engineer for trust boundaries. Verify unfamiliar APIs against source or documentation before using them.

Run focused checks for the behavior being changed. For Portal, use focused Credo after coherent Elixir edits and relevant ExUnit tests. Keep required happy, denial, cross-account, and abuse cases. Fix a red check before dependent work; gather missing evidence and repair within scope instead of stopping at the first failure. Never weaken an assertion or gate to hide a defect.

Adapt routine implementation details as evidence develops. Preserve the full agreed behavior and security contracts. For a material scope or authority change, complete independent work and ask for that decision. Nearby unrelated issues belong in a follow-up. Use targeted edits unless most of the file is changing.

The lead reviews delegated changes and runs each touched project's local final gate on the final tree. Portal uses ./run gate portal --changed; run the complete ./run gate portal before a push or release. Other projects use ./run gate <project>. Delegates provide focused check evidence; a handoff does not itself require duplicate full gates. Rerun or broaden verification after changes, failures, or unresolved concerns. Manual, skill, and KB-only work needs ./run check agent-setup plus ./run check docs; executable tooling changes also need ./run gate tooling.

Finish the lifecycle required by AGENTS.md: focused commit, task notes, and completion state. Report actual outcomes and validation, distinguishing source completion from push, deployment, and live proof. Preserve pending work and user decisions across compaction. Do not end with a promise to perform an authorized step that remains undone.

---name: workflow-workdescription: Complete an authorized Emisar implementation with focused feedback, canonical final gates, and scope discipline. Use for a plan, checklist, or request to implement or continue work.---# Complete the implementationFollow the root and touched project AGENTS.md. The user's request or accepted plandefines the complete deliverable. Use workflow-spec when design is needed, thencontinue under the existing authorization.Work in coherent steps. Give concise updates about findings and the next check.Use Elixir context/testing skills for Portal, go-engineer for runner/MCP, andsecurity-engineer for trust boundaries. Verify unfamiliar APIs against source ordocumentation before using them.Run focused checks for the behavior being changed. For Portal, use focused Credoafter coherent Elixir edits and relevant ExUnit tests. Keep required happy, denial,cross-account, and abuse cases. Fix a red check before dependent work; gather missingevidence and repair within scope instead of stopping at the first failure.Never weaken an assertion or gate to hide a defect.Adapt routine implementation details as evidence develops. Preserve the full agreedbehavior and security contracts. For a material scope or authority change, completeindependent work and ask for that decision. Nearby unrelated issues belong in afollow-up. Use targeted edits unless most of the file is changing.The lead reviews delegated changes and runs each touched project's local finalgate on the final tree. Portal uses `./run gate portal --changed`; run the complete`./run gate portal` before a push or release. Other projects use`./run gate <project>`. Delegates provide focused check evidence; a handoff doesnot itself require duplicate full gates. Rerun or broaden verification afterchanges, failures, or unresolved concerns. Manual, skill, and KB-only work needs`./run check agent-setup` plus `./run check docs`; executable tooling changes alsoneed `./run gate tooling`.Finish the lifecycle required by AGENTS.md: focused commit, task notes, andcompletion state. Report actual outcomes and validation, distinguishing sourcecompletion from push, deployment, and live proof. Preserve pending work and userdecisions across compaction. Do not end with a promise to perform an authorizedstep that remains undone.

Frequently asked questions

Workflow Work