The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell commands for operational tasks: it employs curl to send status notifications to a local server (http://localhost:31337/notify) and echo to append JSON-formatted execution logs to ~/.claude/LIFEOS/MEMORY/SKILLS/execution.jsonl.
[COMMAND_EXECUTION]: The TechnicalCreativityGemini3 workflow utilizes the llm CLI tool to offload specialized technical reasoning tasks to external models.
[PROMPT_INJECTION]: The SyntheticDataExpansion workflow presents a surface for indirect prompt injection by ingesting and processing user-provided seed corpora.
Ingestion points: User-supplied 'Seed corpus' in the SyntheticDataExpansion.md workflow.
Boundary markers: The workflow uses clear text headers (SCHEMA:, SEED CORPUS:) to delineate instructions from untrusted input data.
Capability inventory: The skill has the ability to write to the local filesystem (echo) and execute specific CLI utilities (llm, curl).
Sanitization: Risk is mitigated through mandatory JSON schema validation and semantic deduplication of the generated output.

BeCreative