The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data (revenue forecast packs) using the scripts/check-output.py validator, which creates a potential surface for indirect prompt injection if the source data is untrusted.
Ingestion points: The validation script scripts/check-output.py reads content from a file path provided as a command-line argument.
Boundary markers: The skill does not implement specific delimiters or instructions to ignore embedded commands within the forecast files it processes.
Capability inventory: The script is restricted to reading local files and checking for the presence of specific markdown headers and placeholders. It does not have capabilities for network access, file writing, or command execution beyond its own logic.
Sanitization: The script uses basic encoding error replacement (errors="replace") but lacks sanitization or validation logic to filter out potential malicious instructions within the text.

revenue-forecasting