Skip to content
Back to skills

Improve

ASecurity

Audit a codebase or prepare an implementation plan for a specified improvement.

  • 2 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added November 7, 2026
business

Security analysis

A100/100

Pro scans all 10 files and shows the line behind each finding

Scanned November 7, 2026

npx -y skills add BumpyClock/agent-templates --skill improve --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Improve?

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

Security grade badge for Improve
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/bumpyclock-improve/badge)](https://www.skillsdirectory.com/skills/bumpyclock-improve)

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: improve
description: Audit a codebase or prepare an implementation plan for a specified improvement.
license: MIT
metadata:
  author: shadcn
  version: "1.2.0"
---

# Improve

Return evidence-backed recommendations or a self-contained plan for the requested scope.
User instructions define the deliverable and override these defaults.

## Select the route

- Audit or roadmap: use [Audit](references/audit.md).
- Create or revise a plan: use [Plans](references/plans.md).
- Apply findings or execute a plan: use [Execution](references/closing-the-loop.md) and [Programming](../programming/SKILL.md).
- Reconcile existing plans or publish requested issues: use [Plan maintenance](references/closing-the-loop.md).

Read only the route needed for the current task.
An audit does not require a plan, repository scoreboard, or source changes.
An explicit request to apply findings authorizes the selected implementation scope without another approval gate.

## Scope and evidence

Inspect the source and contracts needed to answer the question.
Distinguish observed defects from inferred risks and optional product directions.
Support each recommendation with a source location, concrete cost, and confidence.
State material tradeoffs and limits.

Keep audit operations read-only.
Write plan files only when requested.
Report credential locations without secret values.
Do not commit, push, or publish without authorization.

Use repository checks that address the affected contract.
Record unavailable checks and unresolved assumptions instead of inventing commands or results.

Files in this skill

  • SKILL.md1.5 KB
  • agents/openai.yaml259 B
  • attribution.txt616 B
  • references/audit-playbook.md18.5 KB
  • references/audit.md2.8 KB
  • references/closing-the-loop.md1.9 KB
  • references/plan-template.md2.3 KB
  • references/plans.md1.1 KB
  • scripts/validate-evidence.mjs7.5 KB
  • scripts/validate-evidence.test.mjs4.8 KB

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.