MCP server and proxy that compresses LLM prompts, tool output, and replies to cut token cost.
Llmtrim MCP server exists for a simple reason — assistants are far more useful when they can act on Llmtrim directly instead of describing what you should do. MCP server and proxy that compresses LLM prompts, tool output, and replies to cut token cost.
Local proxy that compresses LLM API traffic so you pay less. Same answers, smaller bill.
Being a remote server, there is no local install. You register the endpoint with your client, authorise it once, and the tools appear.
Once Llmtrim is connected, these are the calls the assistant has available:
Situation — CommandDiagnose — llmtrim doctor · repair with doctor --fixWaste — What happensStage — What it doestool-output — Lossless template fold first, then window logs · diffs · grep · dumps down to errors / changes / matches; shell-capable agents can restore omittedskeletonization — tree-sitter keeps relevant function bodies, drops the rest to signatures (14 languages)dedup — Collapse duplicate + near-duplicate lines (prose only)multimodal — Downscale images to the provider's resolution capFeature — What you getGuard — Blocks one turn if a cold-cache resume would rewrite a huge context (and bill for it)Language — InstallRust — cargo add llmtrim-coreYou will need one environment variable: ANTHROPIC_AUTH_TOKEN. The server will not start without them, which is usually why the tools fail to appear on a first run. Keep credentials in your client's env block or a secrets manager rather than in a file you might commit.
This sits in the monitoring and observability group, where several servers overlap in what they claim to do but differ sharply once you actually set them up. Llmtrim's toolset — Situation, Diagnose, Waste and 11 more — is a fair guide to whether it matches your workflow. It is maintained by fkiene; worth a glance at recent repository activity before you build anything load-bearing on it.
This entry was verified against Llmtrim's own documentation before publication; SyncDev keeps the directory reviewed rather than auto-generated.
| Tool | What it does |
|---|---|
| Situation | Command |
| Diagnose | llmtrim doctor · repair with doctor --fix |
| Waste | What happens |
| Stage | What it does |
| tool-output | Lossless template fold first, then window logs · diffs · grep · dumps down to errors / changes / matches; shell-capable agents can restore omitted first-arrival results with llmtrim recall |
| skeletonization | tree-sitter keeps relevant function bodies, drops the rest to signatures (14 languages) |
| dedup | Collapse duplicate + near-duplicate lines (prose only) |
| multimodal | Downscale images to the provider's resolution cap |
| Feature | What you get |
| Guard | Blocks one turn if a cold-cache resume would rewrite a huge context (and bill for it) |
| Language | Install |
| Rust | cargo add llmtrim-core |
| Ruby | gem install llmtrim |
| Kotlin | implementation("io.github.fkiene:llmtrim:0.12.3") |
```json
{
"mcpServers": {
"llmtrim": { "command": "llmtrim", "args": ["mcp"] }
}
}Configuration as documented by the project. Restart the client after saving.
| Variable | Description | Required |
|---|---|---|
| ANTHROPIC_AUTH_TOKEN | Credential the server authenticates with. | Yes |
Give your coding agent the full DevTools toolbox: traces, network, console, heap snapshots and Lighthouse.
Dashboards, Prometheus and Loki queries, incidents and alerts — observability by conversation.
Errors with full context — stack traces, issue triage and AI-powered root-cause analysis from Sentry's server.
Enables enhanced web research capabilities for large language models through intelligent search queuing and advanced content extraction.
Automates browser interactions and enables Large Language Models (LLMs) to interact with web pages through Playwright and Chrome DevTools Protocol
Guides tool usage by providing recommendations for MCP tools at each problem-solving stage.