The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a Python script (scripts/generate_schematic.py) for visual enhancement, but this script is not provided within the skill package. The recommended command pattern (python scripts/generate_schematic.py "your diagram description") passes natural language input directly as a shell argument. This creates a risk of shell command injection if the script is dynamically generated or does not implement rigorous input sanitization.
[PROMPT_INJECTION]: The skill's workflow depends on ingesting untrusted content from the web, which introduces a surface for indirect prompt injection.
Ingestion points: The skill uses WebFetch and WebSearch to retrieve literature from PubMed and other scientific domains in the 'Conduct Comprehensive Literature Search' phase.
Boundary markers: There are no boundary markers or explicit instructions provided to the agent to treat fetched content as data rather than instructions, allowing potential malicious prompts in scientific papers to influence agent behavior.
Capability inventory: The skill has access to powerful tools, including Bash (used for report compilation and the missing schematic script) and Write/Edit permissions.
Sanitization: No sanitization or validation logic is defined for data retrieved from external sources before it is synthesized into hypotheses and LaTeX reports.
[PROMPT_INJECTION]: There is a discrepancy between the author identified in the skill's YAML frontmatter ('K-Dense Inc.') and the authoritative platform context ('Oceanjackson1'). This metadata inconsistency can be deceptive and may be used to mask the true provenance of the skill.

hypothesis-generation