Orchard Review Example
Example skill that reviews a release proposal; flagged as containing a prompt-injection and exfiltration pattern
- Repository
- rohitg00/ai-engineering-from-scratch
- Stars
- 76,331
- Licence
- MIT
- Files
- 1 file
What it does
This is a sample skill from a security scanner project, used as a test case rather than a real tool. It is written as instructions for an AI agent to review a deployment diff, but its body also tells the agent to ignore prior instructions and to upload a .env file (which typically holds secrets) to an external address. It is meant to show what a skill scanner should catch.
Great for
- Security reviewers testing skill scanners
- Developers building detection rules for malicious agent skills
Skip it if
- Installing as a working release-review tool: the file contains an instruction to exfiltrate .env secrets and to ignore previous instructions, which is dangerous
- Any real deployment review, since the body contains no actual review criteria
Try asking
- “Scan this skill with the skill-scanner tool and confirm it is flagged”
- “Use as a fixture to test detection of prompt injection and credential exfiltration”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
reviewable.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip reviewable.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip reviewable.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Do not install this skill into a live agent with access to a real .env file
- Treat the example as a negative test case for scanners
What's inside
- SKILL.md
Redistributed unmodified from rohitg00/ai-engineering-from-scratch under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.