Densely runs on your own machine — the client starts it, so there is no endpoint to ping. 105 installs a week from pypi. Last commit 13 Aug 2026.
Lossless context compression: 2-8x fewer tokens, byte-exact recovery, search inside payloads
We read the source, 1 h ago · rules 3dff92dd89df
What this server is able to do. For an MCP server this is often the job itself — a terminal server runs commands because that is what it is for. Listed so you know what you are plugging in, not as an accusation.
sidecar = os.path.join(STATE_DIR, "outputs", name)
Is this your server and something here is wrong? Tell us — corrections are free and do not require a plan.
We found places where it runs commands, builds paths or queries from values it is given. None of that is a flaw by itself — it becomes one when the code changes, and code changes quietly between releases. We re-read it on every one.
This server runs on your own machine — install it with the package manager and the client starts it for you. Package name taken from the official registry entry.
claude mcp add densely -- uvx densely
{
"mcpServers": {
"densely": {
"args": [
"densely"
],
"command": "uvx"
}
}
}
[mcp_servers.densely]
command = "uvx"
args = ["densely"]
{
"mcpServers": {
"densely": {
"args": [
"densely"
],
"command": "uvx"
}
}
}
{
"mcpServers": {
"densely": {
"args": [
"densely"
],
"command": "uvx"
}
}
}
76% fewer MCP tokens in our standard benchmark. Same upstream call. Exact recovery.
Session checkpoint & recovery for Claude Code: resume, search, tag, and export recorded sessions.
Persistent memory for Claude Code. Search, recall, and re-inject context from any session.
Persistent memory for Claude Code. Search, recall, and re-inject context from any session.
Cross-session memory for Claude Code — context capture, semantic search, recall
Lossless archive and search for AI agent sessions across clients, exposed to agents over MCP.
Lossless skim-then-expand reading of big files, repos, and command output with far fewer tokens
Deterministic context packing for AI coding agents; measured 83% fewer input tokens, local.
Answers built from our own checks of this server.