The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it is designed to ingest and process unstructured career data from external sources.
Ingestion points: Processes multiple files from the source-data/ directory, including Obsidian notes, CSV exports, and resume text.
Boundary markers: Absent. The workflow does not provide delimiters or specific instructions to the agent to disregard instructions that may be embedded within the source documents.
Capability inventory: The agent has the ability to read and write files, as well as execute shell commands such as find, grep, unzip, and npm.
Sanitization: Absent. The skill emphasizes Zod schema validation for output structure but lacks mechanisms to sanitize input text or prevent content from influencing agent logic.
[COMMAND_EXECUTION]: The skill utilizes shell commands to perform data inventory and validation tasks.
The instructions direct the agent to run find, grep, and unzip on the contents of the source-data/ directory, as well as npm run validate for content checking.
[DATA_EXFILTRATION]: The skill performs network operations to verify external links discovered during data processing.
The workflow instructs the agent to "extract and verify company/project URLs" to ensure they are active (not 404), which involves connecting to arbitrary domains found in the user-provided data.

cv-data-ingestion