Skip to content

v0.5.6

Choose a tag to compare

View all tags
@tn-pisama tn-pisama released this 28 Jul 20:13
v0.5.6
f17a80f
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Ships two fixes that were merged to main but could not reach users until a release carried them.

Fixed: pisama[mcp] install was broken. The extra declared mcp>=1.0.0 with no upper bound, so a fresh install resolved mcp 2.0.0 and pisama mcp-server crashed on startup:

File ".../pisama/mcp/server.py", line 138, in create_local_server
  @server.list_tools()
AttributeError: 'Server' object has no attribute 'list_tools'

pisama/mcp/server.py uses the mcp 1.x decorator API, which 2.0.0 removed. Now pinned to mcp>=1.0.0,<2. The bound should be lifted once the server is ported to the 2.x API.

Fixed: incorrect README claims. PyPI renders the README from the published artifact, so the project page carried three errors:

  • Four documented detectors do not exist. delegation, grounding, retrieval_quality and compaction_quality raise UnknownDetectorError. Replaced with four that do exist and were undocumented: propagation, citation, routing, mcp_protocol.
  • "Framework-specific detectors for n8n, LangGraph, Dify, and OpenClaw" was false. There are 32 core detectors, gated per platform; none is unique to a framework.
  • The TRAIL benchmark table lacked its claim boundary. The archived run is in-distribution and not held out (144 of 148 traces appeared in calibration material), and the public archive does not contain the prediction-level data needed to recompute joint accuracy. The table now reports only what the archive supports, with the caveat inline and macro-F1 0.7535 / micro-F1 0.7463 named as the reproducible figures.

No API changes.

Assets 2
Loading