The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill executes shell commands that pipe data from the GitHub API directly into a Python interpreter. The Python code is provided inline within the instructions for the purpose of JSON parsing.
[COMMAND_EXECUTION]: The skill uses system tools including curl, grep, cut, and the gh CLI to interact with remote repositories and manage synchronization state in GitHub issues.
[EXTERNAL_DOWNLOADS]: The skill fetches repository metadata and file content from api.github.com and raw.githubusercontent.com to identify and retrieve upstream changes.
[DATA_EXFILTRATION]: Configuration state, including commit SHAs and sync history, is stored and updated in a designated GitHub issue. No sensitive local files or credentials are targeted for external transmission.
[PROMPT_INJECTION]: The skill processes content from an external, third-party repository to generate or modify agent instructions, creating a surface for indirect prompt injection.
Ingestion points: SKILL.md files and other component files fetched from the obra/superpowers repository.
Boundary markers: No specific delimiters or "ignore" instructions are present to encapsulate the fetched content.
Capability inventory: The agent has the ability to write files, execute shell commands, and access the network.
Sanitization: Relies on the agent's classification logic and manual user confirmation before applying changes.

sync-superpowers