The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands such as mkdir and find to manage an internal cache in /tmp, and git commands to identify the repository context for hashing. These are standard operations for codebase analysis tools provided by this vendor.
[EXTERNAL_DOWNLOADS]: The skill triggers the ce-web-researcher sub-agent to gather external information. This is a documented, integral part of the ideation workflow and is subject to user-provided skip phrases.
[DATA_EXFILTRATION]: The skill reads project metadata such as directory structure and strategy documents to ground its suggestions. Persistence involves writing results to the local docs/ideation/ directory or uploading them to a review service (ce-proof) within the same vendor ecosystem.
[PROMPT_INJECTION]: The skill ingests untrusted user input via the $ARGUMENTS parameter to guide its ideation process.
Ingestion points: User-supplied arguments are interpolated into the focus hint context in SKILL.md.
Boundary markers: The arguments are delimited by <focus_hint> tags for sub-agent processing.
Capability inventory: The skill can write markdown files to the repository and interact with other tools in the compound-engineering suite.
Sanitization: No explicit sanitization is performed on the input.
Risk Assessment: The risk is minimal as the skill's output is limited to ideation artifacts that undergo multi-agent critique and human review before further action.

ce-ideate