The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires cloning an external, untrusted repository (https://github.com/YuhangChen1/Paper2All.git) to obtain the necessary scripts for processing academic papers.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a series of high-risk operations including installing unverified Python packages from the untrusted repository (pip install -r requirements.txt) and running various processing scripts (python pipeline_all.py, python scripts/generate_schematic.py).
[COMMAND_EXECUTION]: The installation guide explicitly instructs the user to use sudo for installing system dependencies like libreoffice and poppler-utils, which introduces a privilege escalation risk.
[CREDENTIALS_UNSAFE]: The skill requires the user to store highly sensitive API keys (OPENAI_API_KEY, OPENROUTER_API_KEY, GOOGLE_API_KEY) in a .env file within the cloned repository directory.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality:
Ingestion points: Reads and processes untrusted LaTeX and PDF files provided by the user (described in SKILL.md and references/paper2web.md).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands within the ingested files are present.
Capability inventory: The skill has broad capabilities including Bash command execution, Write access to the filesystem, and network operations via API calls.
Sanitization: There is no evidence of sanitization or validation logic to prevent malicious instructions embedded in paper content from influencing the agent's behavior during extraction or generation tasks.

paper-2-web