The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to manage its environment, specifically invoking "npm install" for dependency management and "npx tsx" to run the generated parsing scripts.
[DYNAMIC_EXECUTION]: A core feature of the skill is the generation and execution of TypeScript code at runtime. The agent creates these scripts based on templates found in "scripts/example.ts" to interact with the LlamaParse API.
[INDIRECT_PROMPT_INJECTION]: The skill processes unstructured files provided by users, which are considered untrusted data. The content extracted from these files is returned to the agent's context, presenting a vulnerability surface where malicious content within a document could attempt to influence the agent's subsequent actions.
Ingestion points: Document files supplied by the user and processed via the "LlamaCloud" client as described in "SKILL.md" and "scripts/example.ts".
Boundary markers: No explicit delimiters or warnings are specified to separate the parsed document content from the agent's internal instructions.
Capability inventory: The generated scripts have the capability to read and write files locally, perform network requests to LlamaParse APIs, and download content via presigned URLs.
Sanitization: The skill does not describe any specific validation or sanitization steps for the data extracted from the parsed documents.
[EXTERNAL_DOWNLOADS]: The skill fetches the "@llamaindex/llama-cloud" library from the public npm registry and references documentation from the "developers.llamaindex.ai" domain to guide its code generation process.

llamaparse