Skip to content
Back to skills

Workflow Spec

ASecurity

Plan an Emisar change when the approach needs design, several layers interact, or the user requests a plan. Define the complete outcome, decisions, and verification before implementation.

  • 391 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
ai-agentsbash

Works with

  • mcp

Security analysis

A100/100

Scanned November 7, 2026

npx -y skills add AndrewDryga/emisar --skill workflow-spec --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Workflow Spec?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Workflow Spec
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/andrewdryga-workflow-spec/badge)](https://www.skillsdirectory.com/skills/andrewdryga-workflow-spec)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: workflow-spec
description: Plan an Emisar change when the approach needs design, several layers interact, or the user requests a plan. Define the complete outcome, decisions, and verification before implementation.
effort: medium
argument-hint: "<what you want to build>"
allowed-tools: Read, Grep, Glob, Bash
---

# Plan the requested change

Read the root and touched project AGENTS.md, then the relevant implementation and
references. Preserve the user's requirements, accepted decisions, and authority.
A multi-file change alone does not require a separate planning conversation.

Define the complete outcome, the smallest implementation that delivers it, the
affected layers, and meaningful verification. Use existing contexts, components,
and tools. State assumptions and explain consequential tradeoffs briefly.
Recommend scope changes when needed; do not silently defer requested behavior.

For Portal context changes, name the happy, denial, and cross-account paths.
For runner/MCP/packs, identify affected execution contracts and abuse cases.
For infrastructure, distinguish source validation, plan, apply, and live proof.
For agent instructions, check discovery, conflicting rules, and actual behavior.
Name the touched projects' canonical final gates and other required checks.

Consult specialist skills when the change needs their judgment. Keep the plan
proportional: outcome, implementation steps, verification, and unresolved decisions.

If the user requested only a plan or assessment, deliver that and stop. If they
authorized implementation, continue with workflow-work after planning; no second
approval is needed for routine steps. Ask only when a missing decision or authority
materially changes the result, after completing independent authorized work.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.