sift_onboardLABSCO SUMMARY
It spends CPU so the model does not spend context.
The engines doing the compression are downloaded and run on your machine — nothing goes out to a compression service, and there is no key. That is also the cost, which is why sift_warmup exists: the first call pays for loading the models. The honest touch is sift_extraction's diff mode, which shows exactly what was thrown away rather than asking you to trust a rate.
What it is
A Python server that sits between raw data and the context window. It routes each input to one of three engines — a heuristic sieve for logs, LLMLingua-2 semantic compression for prose, and a local BGE cross-encoder for ranking — and returns the distilled text with an audit header showing token reduction and latency.
What you get
- A local file read through the right pipeline instead of raw, with
.pdf,.docx,.xlsx,.pptx,.zipand.htmlconverted to Markdown first —sift_read_file - A noise estimate for a file or a string before you spend context on it, ending in a recommendation for which sifter to use —
sift_analyze_file,sift_analyze - Repetitive log noise stripped while errors and stack traces survive —
sift_logs - Prose and chat history condensed at a rate you choose, with a low-fidelity warning appended when vocabulary overlap falls below the threshold —
sift_chat - Long documentation put through two stages: structural sieve first, then semantic compression —
sift_doc - OCR and PDF extractions cleaned of metadata debris, optionally with a
--- REMOVED CONTENT ---section so you can check what was cut —sift_extraction - A list of chunks ranked against a query by a local cross-encoder, returning the top results with confidence scores —
sift_rank - Token-saving and latency telemetry, a one-shot project setup that writes IDE hooks, and a model warm-up —
get_sift_stats,sift_onboard,sift_warmup
Requirements
No API key: every model runs locally. pip install "semantic-sift[neural]", then run semantic-sift. Reads outside the workspace root are blocked unless SIFT_ALLOW_GLOBAL_READS=true. SIFT_RANK_TOP_N sets the default result count for sift_rank (default 3), and SIFT_COMPACTION_FIDELITY_THRESHOLD (default 0.3) decides when a compaction is flagged as low fidelity. The package is semantic-sift 0.3.8; the [multi-modal] extra adds MarkItDown for binary formats.
Setup effort
One command — pip install "semantic-sift[neural]"
USE CASES
01
Reading a log file that would fill the context window
Route the file through the heuristic sieve so the errors and stack traces survive and the repeated boilerplate does not.
02
Deciding whether a file even needs sifting
Get the noise ratio and a recommendation first, so a short clean file is read natively instead of being compressed for nothing.
03
Feeding a long PDF spec to an agent
Convert the binary to Markdown, strip headers, footers and page numbers, then compress the prose at a rate you pick.
04
Narrowing search results before the model reads them
Rank the candidate chunks against the question locally and pass on only the top ones, with their scores.
11 tools
Listed in project docs. Your client may show a different list.
sift_warmupsift_read_filesift_analyze_filesift_analyzesift_logssift_chatsift_docsift_ranksift_extractionget_sift_statsBEFORE YOU INSTALL
What to know first
Tools you get
11 tools named in the project's documentation.
Requires
semantic-sift on PATH
INSTALLATION
Connect your client
Add the block to your client, or run the command directly.
Server setup
Configuration file
JSON
{
"mcpServers": {
"semantic-sift": {
"command": "semantic-sift"
}
}
}pip
TERMINAL
pip install "semantic-sift[neural]"
