The Agent Skills Directory

[COMMAND_EXECUTION]: The skill deploys and runs several scripts to manage project infrastructure. This includes merging project settings via merge-claude-settings.py and merge-codex-hooks.py, enforcing path safety during installation with copy-path-safety.py, and providing runtime guardrails for writing projects through story_hook_core.js and story_codex_hook.py. These operations are consistent with the skill's purpose of environment setup and management and do not exceed the necessary scope.
[EXTERNAL_DOWNLOADS]: In the session-start.sh hook, the skill uses curl to fetch the latest release version from the author's official GitHub repository (zenstory-ai/oh-story-claudecode). This is a standard update check from a vendor-owned source and is documented neutrally as a safe operation.
[PROMPT_INJECTION]: The skill functions as a writing assistant that processes user-provided markdown files. Because the agents are granted file-system access and command-execution permissions to perform analysis and linting, there is an inherent surface for indirect prompt injection if malicious instructions were embedded in the story text. However, the sub-agents are provided with robust structural instructions and format constraints to mitigate such risks, and these capabilities are required for the skill's primary functionality.
Ingestion points: The agents read markdown files from project directories like 正文/ (Body) and 设定/ (Settings) using Read, Glob, and Grep tools.
Boundary markers: Comprehensive developer instructions are provided to agents to enforce specific formatting and analysis boundaries, reducing the likelihood of accidental instruction following.
Capability inventory: The agents utilize Write, Edit, Bash, Read, Glob, and Grep tools to manage the writing environment.
Sanitization: The skill relies on structured data extraction and specific agent prompts to manage content, rather than arbitrary text execution.

story-setup