Skip to content
Back to skills

Guides

ASecurity

Turn a repeated repository review into a reusable procedure

  • 147 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 3, 2026
ai-agentspython

Security analysis

A100/100

Pro scans all 21 files and shows the line behind each finding

Scanned September 20, 2026

npx -y skills add prapaa-ai/agav --skill guides --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Guides?

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

Security grade badge for Guides
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/prapaa-ai-guides/badge)](https://www.skillsdirectory.com/skills/prapaa-ai-guides)

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
---
title: Author and Evolve a Skill
description: Turn a repeated repository review into a reusable procedure
guideLevel: advanced
order: 3
---

# Author and Evolve a Skill

Use a skill when you repeat the same procedure across several tasks.

## Scenario: review modules consistently

Create `.agav/skills/module-review/SKILL.md` in the hello-agav repository:

```markdown
---
name: module-review
description: Review a Python module against project conventions
version: 1.0.0
invocation: user
allowed-tools:
  - read_file
  - grep_search
  - find_files
---

# Module Review

1. Read the supplied module and identify its public functions.
2. Check for type annotations, docstrings, and error handling.
3. Find the corresponding test file in `tests/` and list untested paths.
4. Return a table with risk, evidence path, and recommendation.
5. Do not edit files.
```

Restart Agav, then run the skill manually:

```text
/module-review @counter.py
```

`invocation: user` keeps the workflow predictable: it runs only when you call it. The tool allowlist also keeps the review read-only.

## Improve it from real use

Inspect its configuration and usage:

```text
/skills info module-review
```

After several runs, change one instruction when you see a repeated problem — for example, add an edge-case check for malformed JSON if reviews keep missing it. Keep project skills in source control and review instruction changes like code.

## Expected result

Every module review follows the same checks and produces the same compact report format.

Next: [Compose Agents, Skills, and Tools](/guides/agent-skill-tool-combinations).

Files in this skill

  • add-files-as-context.md312 B
  • agent-skill-tool-combinations.md1.7 KB
  • agentic-memory.md281 B
  • build-agent-from-template.md6.3 KB
  • build-native-agent.md6.7 KB
  • byok-ollama-openai.md302 B
  • create-search-skill.md1.6 KB
  • existing-project.md409 B
  • explore-code-with-lsp.md332 B
  • human-in-the-loop.md403 B
  • index.md1.9 KB
  • install-and-configure-agent.md4 KB
  • leverage-memory.md2.7 KB
  • long-running-agent.md5.9 KB
  • loops.md287 B
  • mcp.md4.8 KB
  • modify-and-save-plan.md312 B
  • parallel-agents.md1.6 KB
  • parallel-tools.md1.7 KB
  • plugins.md1.8 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…