The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation and use of the lattereview Python package. While this is an external third-party dependency from PyPI, its usage is transparently documented as the primary focus of the latte-review-guide skill for automating literature reviews.
[COMMAND_EXECUTION]: The scientify-write-review-paper skill utilizes standard shell commands (cat, ls) to manage workspace project states and verify directory structures within the user's environment. These operations are restricted to project-specific paths (~/.openclaw/workspace/).
[PROMPT_INJECTION]: The paper-reading-assistant and latte-review-guide skills are subject to indirect prompt injection risks due to the processing of untrusted external content.
Ingestion points: Text extraction from PDF manuscripts in paper-reading-assistant/SKILL.md and bulk data imports (CSV/PubMed) in latte-review-guide/SKILL.md.
Boundary markers: Prompt templates for summarization and critique lack explicit delimiters (e.g., XML tags or triple backticks) to isolate untrusted paper text from the agent's instructions.
Capability inventory: The skill does not possess high-privilege capabilities such as arbitrary network access or administrative shell execution, limiting the impact of any potential injection.
Sanitization: No sanitization or filtering of the extracted text is performed before interpolation into LLM prompts.

paper-review-skills