Skip to content

Releases: ericmaddox/compressor-reflex-mcp

Release list

v0.2.2 - Output preservation at limits and graceful MCP error handling

v0.2.2 - Output preservation at limits and graceful MCP error handling Latest
Latest

Choose a tag to compare

View all tags
@ericmaddox ericmaddox released this 03 Oct 02:32
v0.2.2
7b3bdc6

Compressor Reflex MCP (v0.2.2)

Maintenance and reliability release preserving complete tool outputs when approaching resource thresholds and hardening MCP protocol error responses.

What's New in v0.2.2

  • Preserve Output Tails: Decoupled neural scoring budget truncation from the returned line buffer in serializer.py, preventing truncation of end-of-log error traces.
  • Fail-Open Limit Bypasses: Inputs exceeding size limits (5MB or 10,000 lines) now return 100% of raw content unaltered with structured bypass metadata (input_char_limit / input_line_limit).
  • Resilient MCP Error Handling: Catches internal tool exceptions and returns clean MCP {isError: true} payloads rather than terminating the stdio pipe.
  • Automated Regression Suite: Added 14 model-independent regression tests and CI regression workflow.

Installation

# Direct run via uvx
uvx compressor-reflex-mcp

# Or install from PyPI
pip install compressor-reflex-mcp
Assets 2
Loading

v0.2.1 - Compressor Reflex MCP

v0.2.1 - Compressor Reflex MCP

Choose a tag to compare

View all tags
@ericmaddox ericmaddox released this 02 Oct 02:37
v0.2.1
a33fece

Compressor Reflex MCP v0.2.1

Model Context Protocol (MCP) server & transparent proxy for the Compressor Reflex model. Cuts tool-output token consumption by up to 90% in Cursor, Antigravity IDE, Claude Desktop, and other MCP environments while preserving 100% of critical anchors, test failures, and decisive lines.

Highlights

  • 89.7% Empirical Tool Compression: Validated over 181 tool outputs across real IDE coding sessions.
  • 100.0% Anchor Retention: Mathematically calibrated threshold tau* = 0.50 ensures zero loss of test failures or error tracebacks.
  • Fail-Open Bypass Policy: Automatically bypasses outputs with <= 5 physical lines or <= 64 tokens with zero overhead.
  • Dual-Mode Integration: Callable directly via MCP tools (compress_tool_output, compress_file) or as a transparent proxy (compressor-reflex-mcp proxy -- ).
  • Hugging Face Model Hub: Pre-quantized INT8 ONNX weights automatically downloaded and cached from �ialchemist-dev/compressor-reflex.
  • Security Hardened: Path traversal prevention, workspace containment, payload DoS limits, and cryptographic SHA-256 integrity verification.

Quick Installation

�ash pip install git+https://github.com/ericmaddox/compressor-reflex-mcp.git

Assets

  • compressor_reflex_mcp-0.2.1-py3-none-any.whl
  • compressor_reflex_mcp-0.2.1.tar.gz
Loading