The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on the camp command-line tool (specifically camp workitem and its aliases) to perform data discovery, create directories, and manage metadata.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from workflow directories and design documents, which represents a surface for indirect prompt injection.
Ingestion points: Data is ingested via camp workitem --json which reads from directories like workflow/<type>/<slug> and metadata files.
Boundary markers: The instructions do not define specific boundary markers for the AI when processing the JSON output of the tool.
Capability inventory: The skill has the capability to write to the file system through the create and adopt subcommands.
Sanitization: The skill includes explicit validation rules for slugs and types, requiring them to be path-safe (no control characters, specific length limits, and character restrictions).
[SAFE]: No obfuscation, malicious exfiltration patterns, or unauthorized persistence mechanisms were identified. The tool operates within defined project paths (workflow/, .campaign/).

camp-workitems