The Synthetic Context Generator

Generate tailored context windows with relevant code patterns, pitfalls, StackOverflow solutions, and local project context for AI coding tasks.

This skill appears to do what it says: it will read and index any project path you give it, combine results with its built-in knowledge base, and fetch relevant StackOverflow answers. Before using it: (1) do not point it at directories that contain secrets (private keys, .env files, credential stores) because those files can be read and included in outputs; (2) consider running it in a non-sensitive workspace or with a copy of your repo; (3) be aware outputs may include snippets from your codebase that you then paste into other tools or LLMs — avoid leaking sensitive data that way; (4) if you need token-accurate counts, ensure tiktoken is installable in your environment (it may require native bindings). If you want stronger assurance, request the full project repository URL or a signed release so you can verify package provenance and review the complete package-lock before installing.

Type: OpenClaw Skill Name: scg Version: 1.0.2 The Synthetic Context Generator (SCG) is a developer tool designed to reduce context bloat for AI coding assistants by retrieving only relevant code patterns, local project snippets, and StackOverflow solutions. The codebase is well-structured and uses standard libraries like 'commander' for CLI management, 'fuse.js' for local fuzzy searching, and 'tiktoken' for token estimation. Analysis of 'src/sources/localSearch.js' and 'src/sources/webSearch.js' confirms that local file reading is restricted to the user-specified directory and web queries are limited to keywords derived from the user's goal. There are no signs of data exfiltration, malicious execution, or prompt injection attempts in the documentation or instructions.