The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple shell scripts from a non-standard location in the user's home directory (${HOME}/.claude-octopus/plugin/scripts/). These scripts include state-manager.sh and orchestrate.sh, which perform workflow management and project orchestration.
[DYNAMIC_EXECUTION]: In Step 4.5, the skill dynamically constructs a shell command to spawn a sub-agent. This command includes the full contents of a user-provided specification file interpolated into a string literal, which is then executed via orchestrate.sh.
[INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it reads untrusted data from a specification file and injects it directly into a prompt for a 'QA adversary' sub-agent.
Ingestion points: Content of the file at <spec_path> (SKILL.md, Step 4.5).
Boundary markers: None. The content is placed directly under a 'SPECIFICATION:' header.
Capability inventory: The skill has capabilities for shell command execution, file system navigation, and interaction with external CLI providers (codex, agy).
Sanitization: None detected. The raw file content is assigned to a variable and passed to a sub-shell command.
[EXTERNAL_DOWNLOADS]: The skill requires and checks for the presence of external CLI tools codex and agy (Antigravity CLI), which are used for scenario generation and model challenges. These tools must be pre-installed and are not managed by the skill's own package structure.

skill-factory