Skip to content
Back to skills

brief-first-execution

ASecurity

Create a single working brief before substantial implementation, audit, or migration work. Use when tasks are non-trivial, risks are real, or the session is likely to drift without one source of truth for scope, non-goals, verification, and done criteria.

  • 16 stars
  • 0 votes
  • 1 copy
  • 4 views
  • Added June 15, 2026
ai-agentsgo

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned June 15, 2026

npx -y skills add markoblogo/abvx-agent-skills --skill brief-first-execution --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of brief-first-execution?

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

Security grade badge for brief-first-execution
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/markoblogo-brief-first-execution/badge)](https://www.skillsdirectory.com/skills/markoblogo-brief-first-execution)

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: brief-first-execution
description: Create a single working brief before substantial implementation, audit, or migration work. Use when tasks are non-trivial, risks are real, or the session is likely to drift without one source of truth for scope, non-goals, verification, and done criteria.
license: MIT
metadata:
  abvx_status: experimental
  abvx_origin: original
---

# Brief First Execution

Start non-trivial work with one live task brief.

## Goal

Reduce drift by creating a compact source of truth for:

- scope;
- non-goals;
- affected areas;
- risks;
- verification;
- done criteria.

## When To Use

Use for:

- multi-file implementation;
- behavior changes;
- migrations, audits, or refactors;
- long runs where context could drift;
- work with real operational or delivery risk.

Skip for:

- tiny one-file edits with obvious blast radius;
- trivial questions that do not create lasting change.

## Workflow

1. **Name the task**
   - write a short title that matches the real delivery target.
2. **Define scope**
   - describe what is in;
   - keep it concrete and delivery-shaped.
3. **Define non-goals**
   - state what this task will not do;
   - use this to stop opportunistic scope creep.
4. **List affected areas**
   - identify modules, docs, data flows, or systems likely to change.
5. **List risks**
   - focus on failure modes that matter for this task.
6. **Define verification early**
   - list the narrowest useful checks first;
   - add broader checks only where the blast radius warrants them.
7. **Define done criteria**
   - describe what must be true before closing the task.
8. **Keep the brief live**
   - update the brief when scope changes materially;
   - do not leave the brief frozen while the work mutates underneath it.

## Rules

- One task, one brief.
- Scope and non-goals should both be explicit.
- Verification belongs in the brief before heavy implementation starts.
- If the task changes enough, revise the brief rather than pretending it did not.

## Final Report

Return the brief itself and call out any unresolved scope or verification questions.

Files in this skill

  • SKILL.md2.1 KB
  • SKILL_CARD.md1.3 KB
  • agents/openai.yaml218 B

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.