LLM Context.py MCP Server
Project codebase file selection and context packing via reusable rules.
Data last scanned today · Reviewed today
Overview
LLM Context.py automates the tedious process of manual file curation for coding agents. By installing a project-specific skill, the tool teaches agents to autonomously identify, verify, and pack relevant file contents into composable rules. This helps prevent token limit exhaustion while ensuring the model has exactly what it needs for specific refactoring tasks. The system utilizes a rule-based schema to define file selections, which agents can then author and reuse. Users interact with the tool via a set of CLI commands, including lc-preview for verifying selection scope and lc-context for piping content to other agents or LLM sessions.
Our verdict
With a security score of 43/100, this server is classified as high-risk. This rating stems primarily from the fact that it is a community-maintained project that has not yet undergone an editorial review, leaving its read-only and authentication security models unverified. While it features zero direct dependencies and recent activity, the lack of third-party security vetting means it should be used cautiously in environments handling sensitive data or proprietary codebases.
- Zero direct dependencies reduces supply chain attack surface
- Supports automated, rule-based context curation for coding agents
- Provides visibility into selection size and missing dependency symbols
- Licensed under the permissive Apache-2.0 license
- High-risk security profile due to lack of editorial review
- Authentication methods and read-only behaviors are not yet verified
- Community-maintained without official vendor oversight
Tools
| Tool | Description | Risk |
|---|---|---|
| lc_missing | Checks for missing files related to a context pack. | low |
| lc_changed | Identifies files that have changed relative to a context pack. | low |
| lc_outlines | Provides outlines of files included in the context. | low |
| lc_preview | Previews the file selection for a specific rule. | low |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›How do I install the LLM Context server?
You can install it using 'uv tool install "llm-context>=0.6.0"'. After installation, run 'lc-init' within your project root to set up the necessary configuration and skill files.
›Where can I find the full documentation for this server?
Full documentation is contained within the 'lc-curate-context' skill files, located at '.claude/skills/lc-curate-context/' after running the initialization command.
›How is the MCP server configured?
You can add it to your MCP configuration file using the command 'uvx' with the arguments '--from', 'llm-context', and 'lc-mcp'.
›What is the purpose of a rule in this server?
A rule is a YAML-formatted file that defines which files and information a coding agent needs for a specific task. These rules help agents manage context, verify selections, and reuse instructions for future tasks.
›Can I check what files a rule selects before using it?
Yes, you can use the 'lc-preview' command followed by the rule name to see the exact file list, total size, and any dependencies identified via the code graph.
›What happens if I update the package?
Run 'uv tool upgrade llm-context' to upgrade the package. You should then execute any 'lc-*' command to refresh the skill, rules, and templates in your project.
Alternatives
Badge
Maintain this server? Add the live badge to your README.