Semantic Cache MCP
An MCP file cache for AI coding agents that reduces token usage by caching file contents. Re-reads cost approximately 40 tokens instead of sending the whole file, and changed files return diffs. It supports tools like read, grep, BM25 search, and atomic edits, working with clients such as Claude Code, Cursor, Codex, and VS Code.
Package and access
Local package with source-attributed installation metadata. Koskamo does not execute packages during review.
What you can do with it
- Reduce token costs for re-reading unchanged files
- Return diffs for files with minor edits
- Perform grep and BM25 search over cached files
- Execute atomic edits with dry_run capability
Source and maintenance
- Maintenance checked
Packages and installation
Source-provided package data. Koskamo has not executed or live-probed this package.
semantic-cache-mcp 0.7.0
{
"version": "0.7.0",
"transport": "stdio",
"identifier": "semantic-cache-mcp",
"runtimeHint": "uvx",
"registryType": "pypi"
}Tools
Reported by the attributed source; not independently observed by Koskamo.
Connect from your client
Each snippet configures the same source-provided package; replace placeholders before use.
Cursor
{
"mcpServers": {
"semantic-cache-mcp": {
"command": "uvx",
"args": [
"semantic-cache-mcp==0.7.0"
]
}
}
}Cline
{
"mcpServers": {
"semantic-cache-mcp": {
"command": "uvx",
"args": [
"semantic-cache-mcp==0.7.0"
]
}
}
}VS Code
{
"servers": {
"semantic-cache-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"semantic-cache-mcp==0.7.0"
]
}
}
}Claude Desktop
{
"mcpServers": {
"semantic-cache-mcp": {
"command": "uvx",
"args": [
"semantic-cache-mcp==0.7.0"
]
}
}
}Let your agent find its tools
Find servers like Semantic Cache MCP from your agent. Add Koskamo once to search the index for endpoints, access requirements, and client setup. Public discovery needs no key.