The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a protocol where all terminal interactions, including package management and script execution, must be routed through the bash_exec tool. This ensures that every command is part of a managed session with durable logging and process control (list/read/kill capabilities).
[PROMPT_INJECTION]: The skill processes external, potentially untrusted data from the local environment, creating a surface for indirect prompt injection.
Ingestion points: According to the Preconditions and gate section in SKILL.md, the workflow ingests quest files, user-provided assets, checkpoints, and extracted texts to inform the design and execution of analysis slices.
Boundary markers: The instructions do not specify the use of clear delimiters or warnings to ignore instructions embedded within the ingested data files.
Capability inventory: The agent is instructed to use bash_exec for full terminal access and a variety of artifact tools for complex file-system operations and branch management.
Sanitization: The skill does not explicitly mention sanitizing or validating the content of external assets before they are used to drive code generation or execution strategies.

ds-analysis-campaign