The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill downloads executable JavaScript and CommonJS scripts (catalog.cjs, session-check.cjs, and opencode-discovery.js) from an external GitHub repository and configures the agent to execute them.\n- [PERSISTENCE]: The workflow modifies sensitive agent configuration files, specifically ~/.claude/settings.json, ~/.codex/config.toml, and opencode.json. It adds the downloaded scripts to the SessionStart hooks or plugin lists, causing the code to run automatically whenever the agent is initialized.\n- [EXTERNAL_DOWNLOADS]: Content is fetched from https://raw.githubusercontent.com/LilMGenius/paperthin/. While GitHub is a well-known service, the scripts are retrieved from a non-trusted repository, introducing a risk of supply chain compromise where external changes could affect the agent's behavior.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands including npx skills, gh api, and local file manipulation commands to perform updates and modify system settings. It also uses gh auth status to verify credentials for authenticated API calls.\n- [DYNAMIC_EXECUTION]: The skill implements a mechanism for dynamic code execution by fetching remote scripts and integrating them into the host agent's runtime environment via configuration overrides.\n- [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect injection as it processes external data without sufficient controls.\n
Ingestion points: Data returned by npx skills list and scripts fetched from the LilMGenius/paperthin repository as specified in SKILL.md.\n
Boundary markers: Absent; there are no delimiters or instructions provided to the agent to treat ingested content as untrusted or to ignore embedded instructions.\n
Capability inventory: The skill possesses extensive capabilities including subprocess execution (npx, gh), file-system write access for configuration files, and network access for downloading scripts.\n
Sanitization: Absent; no evidence of input validation, escaping, or filtering of the external scripts or command outputs before they are processed or written to the filesystem.

re0-upgrade