The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the agent to execute various shell commands for environment diagnostics and state management. Examples include running node --version, python3, and date -u to populate the .story-deployed sentinel file, as well as executing framework-specific commands like npx skills add for component updates (found in SKILL.md and references/diagnostics.md).
[DYNAMIC_EXECUTION]: The skill orchestrates the execution of multiple internal scripts provided within the skill package to handle configuration merging and agent generation. Key scripts mentioned include merge-claude-settings.py, generate-antigravity-agents.mjs, and deploy-antigravity-skills.py. It also calls external scripts from related skills, such as tracking_commit.py (found in SKILL.md and references/diagnostics.md).
[INDIRECT_PROMPT_INJECTION]: The skill presents a significant indirect prompt injection surface.
Ingestion points: It reads and processes arbitrary user fiction projects, including manuscripts, directory structures, and configuration files (specified in SKILL.md).
Boundary markers: It mentions using specific markers for configuration merging but lacks explicit instructions to disregard natural language instructions embedded within the ingested manuscripts.
Capability inventory: The agent is granted the ability to write to the filesystem, execute local scripts, and run shell commands based on the state of the ingested project data.
Sanitization: There is no evidence of sanitization or filtering for the external project content before it is processed by the orchestration logic.

story-setup