The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources, such as job postings and scraped website text, and interpolates it into LLM prompts for analysis.
Ingestion points: Untrusted data enters the context through scripts/lib/common.ts and via scraping tools documented in provider-playbooks/firecrawl.md.
Boundary markers: The skill relies on simple variable interpolation without robust delimiters or sanitization logic in its scripts to isolate external content from instructions.
Capability inventory: The skill can execute local binaries (execFileSync), perform network requests (fetch), and write to the local file system (writeFileSync).
Sanitization: There is no explicit sanitization of input data before it is passed to LLM providers.
[EXTERNAL_DOWNLOADS]: The skill requires the @cargo-ai/cli package, which is downloaded from the official npm registry. This dependency is consistent with the author's identity and is a standard requirement for the tool's operation.
[COMMAND_EXECUTION]: The script scripts/lib/common.ts executes the local npm command to identify the global package root. The command is invoked with fixed arguments and does not pose a command injection risk.

cargo-gtm