The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from business process descriptions provided via user arguments or local file paths. This data is used to drive the modeling process and interview, making the agent susceptible to instructions embedded within those descriptions. * Ingestion points: User input in $ARGUMENTS and file content read from user-provided paths in SKILL.md (Step 1). * Boundary markers: There are no boundary markers or instructions to ignore embedded prompts in the ingested content. * Capability inventory: The skill possesses file read/write capabilities and can execute local shell commands like git. * Sanitization: No sanitization or content validation is performed on the ingested descriptions.
[DATA_EXFILTRATION]: The skill allows the agent to read from arbitrary file paths provided in the user arguments. While intended for process documentation, this functionality could be used to access sensitive local files if a sensitive path (e.g., .env or SSH keys) is supplied.
[COMMAND_EXECUTION]: The skill utilizes the git rev-parse --show-toplevel command to identify the repository root for defining the output location of the generated domain models.

domain-model