What does provenance-aware priority handling do?

It prevents untrusted RAG, tool, and assistant content from gaining protected priority solely through imperative or safety-critical wording.

Does LLMSlim prevent prompt injection?

No. It mitigates compression-induced instruction elevation and applications still need defense in depth.

The default extractive strategy runs locally. Rewrite and hybrid strategies use a caller-supplied provider.

LLM ProviderPython integration

Prepare long Python-managed context before an Anthropic SDK request.

Use the same Python LLMSlim API with your Anthropic client.

Install LLMSlim and the official Anthropic SDK using your package manager:

terminal

pip install llmslim

STEP 01

1. Retrieve or construct application context.

STEP 02

2. Label provenance accurately and compress with LLMSlim.

STEP 03

3. Send the resulting text to the provider SDK.

STEP 04

4. Apply application-level policy and output validation.

Complete, runnable implementation wrapper for Anthropic:

anthropic_app.py

from llmslim import compress

result = compress(long_context, target_ratio=0.5)
# Pass result.compressed_text to your Anthropic SDK request.

Use LLMSlim in your Python application before constructing the provider request. Measure target workloads before choosing a compression ratio.

01.Use compress_documents() for retrieved documents; it defaults to RAG provenance.
02.Leave system messages unchanged by default in compress_chat_messages().
03.Do not treat compression as a complete prompt-injection defense.

Metric Dimension	Uncompressed Payload	LLMSlim Compressed	Recorded Impact
v0.4.0 release gate	N/A	489 tests passed	90.93% coverage; 375 schemas across 18 catalogs

No. LLMSlim is provider-agnostic and does not bundle provider SDKs.

Issue: Unexpected compression result

Solution: Check target_ratio, role labels, and whether the input is long enough to benefit from compression.