The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation instructs users and agents to install the official @daimonia/stc package from the npm registry and adds the skill using standard agent skill managers. These are standard installation procedures for developer tools from the known vendor daimonia-ai.
[COMMAND_EXECUTION]: The skill provides a Node.js command-line interface (tools/cli.mjs) for initializing the project environment and performing automated text checks. It includes scripts that perform file system operations (read/write) within the project directory to manage configuration files like AGENTS.md and CLAUDE.md, and it uses spawnSync for build and test workflows.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process external Chinese text for review and rewriting purposes. As a utility that analyzes user-provided content, it has a potential surface for indirect prompt injection.
Ingestion points: The skill processes user-provided Chinese text passed via standard input or file paths to the stc check command (described in SKILL.md and tools/cli.mjs).
Boundary markers: While instructions in SKILL.md advise the agent to preserve facts and remove filler, no cryptographic delimiters or structural isolation are used for the processed text.
Capability inventory: The skill can write to project configuration files via the init command and read arbitrary project files through the check command scripts.
Sanitization: The extraction logic in tools/check/src/extract.mjs uses regex to mask code blocks and URLs, providing basic isolation, though it does not provide full sanitization of natural language instructions within the analyzed text.

stc