layer-context 0.2.0

Compression-powered Context infrastructure for Agentic Systems

layer-context

Python SDK for ContextLayer - Compression-powered Context infrastructure for Agentic Systems.

Installation

pip install layer-context

For compression features (context compression + prompt compression):

pip install layer-context[compression]

# Pre-download the embedding model (~80MB) to avoid first-call latency:
layer-context-download

Quick Start

Initialize

from layer_context import CL

cl = CL(api_key="cl_sk_your_key_here")

Create a Block

cl.create_block("system-prompt")

Get a Block Handle

block = cl.block("system-prompt")

Append Context

block.append("You are a helpful assistant.")
block.append("Always respond in markdown.")

print(block.get_context())
# Output: "You are a helpful assistant.\nAlways respond in markdown."

Update (Replace) Context

block.update("You are a coding assistant. Use Python examples.")

Delete Specific Content

block.delete("Use Python examples.")

print(block.get_context())
# Output: "You are a coding assistant."

Delete a Block

cl.delete_block("system-prompt")

Versioning

block.create_version()

block.append("V2 content", version=2)
print(block.get_context(version=2))

View History

entries = block.get_history()
for entry in entries:
    print(f"{entry['action']}: {entry['content'][:50]}...")

List All Blocks

blocks = cl.list_blocks()
for b in blocks:
    print(b['name'])

Context Compression

Compress your stored context to reduce token usage while preserving relevance to a given query. Uses sentence embeddings, TextRank, and TF-IDF — runs entirely on your machine.

block = cl.block("support_faq")
block.append("You are a customer support assistant for Acme Corp.")
block.append("Always greet the user warmly and ask how you can help.")
block.append("If the user asks about billing, explain our pricing tiers.")
block.append("We offer Basic ($9/mo), Pro ($29/mo), and Enterprise (custom).")
block.append("For password resets, direct users to settings > security.")
block.append("Our refund policy allows refunds within 30 days of purchase.")

compressed = block.get_compressed_context(
    query="How much does the Pro plan cost?",
    target_tokens=50,
    aggressive_pruning=True,
)

print(compressed)
# Original: ~80 tokens -> Compressed: ~30 tokens

Arguments

Argument	Type	Default	Description
query	str	required	Topic/question to optimize relevance for.
version	int	1	Which block version to compress.
target_tokens	int | None	None	Max token budget. Defaults to ~50% of original.
relevance_threshold	float	0.3	Min cosine similarity to keep a sentence.
dedup_threshold	float	0.85	Similarity above which sentences are deduplicated.
aggressive_pruning	bool	False	Strip filler words and verbose phrases.

Convenience Shortcut

compressed = cl.get_compressed("support_faq", query="billing question")

Prompt Compression

Compress any prompt at token level before sending it to your LLM — reducing costs and improving accuracy.

prompt = """You are a helpful customer support assistant for Acme Corp.
Always greet the user warmly and ask how you can help them today.
If the user asks about billing, explain our pricing tiers in detail.
We offer Basic ($9/mo), Pro ($29/mo), and Enterprise (custom pricing).
For password resets, direct users to the settings > security page.
Our refund policy allows full refunds within 30 days of purchase."""

# Compress the prompt by 50%
compressed = cl.compress_prompt(prompt, rate=50)

Arguments

Argument	Type	Default	Description
prompt	str	required	The prompt text to compress.
rate	float	40	Compression percentage (0-100). A rate of 40 removes ~40% of tokens.

Full Pipeline Example

# 1. Get compressed context
context = cl.block("support_faq").get_compressed_context(
    query="billing question",
    target_tokens=50,
)

# 2. Build prompt
prompt = f"You are a support agent. {context} Answer the user's question."

# 3. Compress the final prompt
compressed_prompt = cl.compress_prompt(prompt, rate=40)

# 4. Send to LLM - fewer tokens, lower cost, higher accuracy

License

MIT