The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to search for existing artifacts using ls docs/plans/ 2>/dev/null | grep -i "<topic>". Directly interpolating user-provided topic strings into shell commands presents a risk of command injection if the input is not properly sanitized by the underlying agent or tool handler.
[PROMPT_INJECTION]: The resume protocol and stage transitions involve reading existing files from the docs/ directory. If these files were modified by an untrusted party or contain malicious instructions, they could influence the agent's behavior during the conversion or planning stages (Indirect Prompt Injection).
Ingestion points: Reads content from docs/sketches/, docs/specs/, and docs/plans/ (SKILL.md).
Boundary markers: None explicitly defined to delimit documentation content from instructions.
Capability inventory: Access to Bash, Write, Edit, and Task tools allows for file modification and command execution.
Sanitization: No explicit sanitization or validation of the ingested documentation content is performed before processing.

arch-flow