The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of shell commands for Git workflow management, environment state verification, and build/test execution.
Evidence includes calls to git, date, npm run build, npm test, and launchctl for service management.
[EXTERNAL_DOWNLOADS]: The skill connects to remote repositories to fetch code updates.
It defaults to https://github.com/qwibitai/pepper.git but allows the user to provide an arbitrary upstream URL.
[REMOTE_CODE_EXECUTION]: By design, the skill merges remote code into the local environment and subsequently executes it via npm build and test scripts.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface in Step 6 and Step 7.
Ingestion points: Reads the content of CHANGELOG.md and lists remote branches via git branch -r from the fetched upstream source.
Boundary markers: No explicit delimiters are used when parsing the file content.
Capability inventory: The skill can execute shell commands (git, npm, launchctl) and invoke other skills via the Skill tool.
Sanitization: The skill identifies potential migration tools by parsing /[skill-name] patterns in the changelog and offers to execute them using the Skill tool. While this is mitigated by AskUserQuestion requiring user consent, a compromised upstream repository could attempt to influence the agent to run malicious skills.

update-pepper