The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use destructive shell commands including git rm -r and rm -rf to remove project documentation directories in the docs/sdd/ path as part of the archiving cycle.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by ingesting untrusted content from documentation files to update the agent's global memory and project logs.\n
Ingestion points: The agent reads the file docs/sdd/{change-name}/verify-report.md, which is an external artifact that may contain malicious instructions.\n
Boundary markers: The instructions do not define boundary markers or delimiters to separate the ingested content from the agent's instructions.\n
Capability inventory: The skill has the capability to write files (Write), delete directories (rm -rf), modify agent memory (mem_save), and delegate tasks to a high-capability sub-agent (model: fable).\n
Sanitization: No validation, sanitization, or filtering is performed on the ingested report content before it is processed or stored in memory.\n- [PROMPT_INJECTION]: The skill performs dynamic instruction loading by reading and executing the /memory-manage update process from ~/.claude/skills/memory-manage/SKILL.md inline.

sdd-archive