The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to execute a platform script setup-superpower-loop.sh. The execution string is dynamically constructed by incorporating a <resolved-design-path>, which is sourced from user-supplied $ARGUMENTS or discovered via file system search in docs/plans/. Constructing shell commands with external or unvalidated strings creates a risk of command injection if the input contains shell metacharacters.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection. It is instructed to read bdd-specs.md as the 'source of truth' and incorporate its content into generated task files.
Ingestion points: Reads bdd-specs.md during Initialization and Phase 1.
Boundary markers: Absent. The skill does not define clear delimiters or instruction-override protections for the ingested content.
Capability inventory: The skill can execute shell scripts (Superpower Loop), write files (task generation), perform git operations, and launch parallel sub-agents.
Sanitization: Absent. Instructions mandate reading the file completely and replicating its scenarios in detail without filtering or validation.
[REMOTE_CODE_EXECUTION]: While no remote code is downloaded from the internet, the skill depends on an external executable script provided by the environment at ${CLAUDE_PLUGIN_ROOT}/scripts/setup-superpower-loop.sh. The safety of the skill relies on the integrity and security of this platform-level script.

writing-plans