The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external, untrusted content such as scientific papers, research claims, and study protocols. This creates a potential attack surface for indirect prompt injection, where an adversary could embed malicious instructions within a document to hijack the agent's logic.
Ingestion points: User-supplied research papers and scientific text (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are provided in the skill body.
Capability inventory: The skill utilizes Read, Write, and Edit tools and provides instructions for executing shell commands via the uv run utility.
Sanitization: There are no provisions for escaping or validating the content extracted from external documents.
[COMMAND_EXECUTION]: The documentation includes explicit shell command examples for the agent to execute, specifically using the uv run command to invoke external Python scripts for schematic generation (SKILL.md).
[EXTERNAL_DOWNLOADS]: The skill references the uv tool from Astral (astral.sh) and integrates with the OpenRouter API (openrouter.ai) for optional visual aid generation. These references to well-known technology services are documented for their intended use in the skill's workflow.
[METADATA_POISONING]: A discrepancy exists between the provided author identity 'kinandigital' and the metadata in SKILL.md, which lists the author as 'K-Dense Inc.'
[NO_CODE]: The skill package is composed entirely of Markdown documentation and reference files; it does not contain any standalone executable scripts or binaries.

scientific-critical-thinking