Skip to content
#explainable ai Open access

MattArtzAnthro/gephi-ai: v1.21.2 — filters, statistics and preview settings that do what they say

Oct 2026 · Zenodo (CERN European Organization for Nuclear Research)
Scientific Computing and Data Management

Abstract

MCP server 1.21.2 · Gephi plugin 1.5.2 · Claude and Codex workflow packages 1.20.2 Update the Gephi plugin with the .nbm below (Tools › Plugins › Downloaded › Add Plugins, then restart Gephi). It needs Gephi 0.11.3. The server updates itself through uvx; Claude Desktop users can install the .mcpb. Fixed Packaging. The wheel lists what it ships (the modules, caveats.json, the viewer) instead of packaging the whole directory, which wrote every path as ./x, a form PyPI now refuses, and installed the tests, pyproject.toml and .gitignore into site-packages. scripts/verify-artifact.sh fails on either. A statistic already running is not started a second time. A second request for the same statistic now returns "already running" instead of starting another run alongside it, which repeated the work and kept the graph's read lock held for both runs. The degree and edge-weight filters choose what to remove after taking the write lock, so another write landing in between cannot leave them removing from a list that no longer matches the graph. Failures inside Gephi leave a stack trace in Gephi's log. Every request that fails with an exception now logs it before returning the error. A PNG export crash had come back as a one-line message with nothing to report. A statistic no longer waits an hour on a Statistics panel that cannot run it. Until Gephi's Statistics window has set up the workspace, the panel accepts a run and never reports back, so the request held for the full one-hour wait. The statistic now runs directly in that case. Filtering into a column or a new workspace now writes what it reports. action="column" wrote no column and action="new_workspace" copied no nodes, while both reported success, because the filter was made Gephi's current query before it had a result. Graphs with integer ids are found by id. After importing a file whose nodes have integer ids, every tool that takes a node id (get, label, edge, path, ego network and more) answered "Node not found". Preview settings report what they could not set. A misspelled property name was stored as a new property and counted as set, and a value that did not fit its setting ("yes" for a true/false setting, 30 for a whole-number setting sent as JSON) was dropped or read as false. The reply now lists unknown_properties and unapplied_properties, whole numbers set whole-number settings, and background.color reads back under the name it is written with. A refused path mark adds no column. gephi_find_shortest_path with a mark_column that already holds text refused, but had already added a column of that name to the other table. gephi_export finds an exporter by its file extension, so gml and net work; the tool description had said gml could not be exported. gephi_create_regex_column counts its matches and warns when nothing matched. Gephi tests the whole value, so ^Dept- matches nothing where Dept-.* matches; the description now says so. A request whose values have the wrong JSON type gets an error that names the request. Gson's own message was a bare type name such as "JsonObject", or empty. A probe run no longer fails on a local overlay that is not a JSON object. It is replaced, as an unreadable one already was. Changed The workflow packages pin MCP server 1.21.2 and name Gephi AI plugin 1.5.2 as the minimum, so the fixes above reach anyone who updates. Documentation The skill, tool reference and version notes explain the "already running" refusal and say to wait or use gephi_stop_statistic, never to retry in a loop. The README says where Gephi's log is for bug reports. A test keeps the skill's wording tied to the plugin's message. Development Coverage floors in CI. JaCoCo checks the plugin's line and branch coverage in mvn test (floors in pom.xml), and pytest-cov checks the MCP server's on Python 3.12 (fail_under in pyproject.toml). pytest-cov and pyyaml are dev dependencies; pyyaml is what the Codex package test reads skill frontmatter with. The release scripts and the Gephi check hook are tested. check-drift.sh and verify-artifact.sh run against throwaway repositories with stand-in network tools, and CI now runs the hook test and shellchecks the plugin's scripts. mypy checks the viewer package. CI ran mypy ./*.py, which reaches gephi_mcp_viewer only through the editable install, so mypy treated it as a site package and silenced its errors. It is now named on the command line in CI and in scripts/check.sh, and the twelve type errors it had hidden are fixed (annotations only; no behaviour changes). Tests read the shipped caveat register only, so a local probe run cannot make a test pass on one machine and fail on another. Plugin unit tests stay off port 8080 and run headless. NetBeans runs the module's installer inside the test JVM, and it started the API server on 8080. With a Gephi already holding that port, the bind failure opened an error dialog on the desktop mid-run.

View source

Similar papers

#artificial intelligence Conference Open access Apr 2020

ECCOLA - a Method for Implementing Ethically Aligned AI Systems

The method, ECCOLA, is presented, which aims at making the high-level AI ethics principles more practical, making it possible for developers to more easily implement them in practice.

Ville Vakkuri, Kai-Kristian Kemell, P. Abrahamsson · 64 citations · ⚡6
#computer vision Review Apr 2024

AI-powered Code Review with LLMs: Early Results

The goal is to not only refine the accuracy of the LLM-based tool but also to underscore its potential in streamlining the software development lifecycle through proactive code improvement and education.

Z. Rasheed, Malik Abdul Sami, Muhammad Waseem et al. · 62 citations · ⚡3
#computer vision Open access Mar 2024

LLM-based agents for automating the enhancement of user story quality: An early report

The use of large language models to automatically improve the user story quality in Austrian Post Group IT agile teams is explored, with a reference model for an Autonomous LLM-based Agent System developed and implemented at the company.

Zheying Zhang, M. Rayhan, Tomas Herda et al. · 48 citations · ⚡4
#computer vision Review Mar 2024

System for systematic literature review using multiple AI agents: Concept and an empirical evaluation

This paper introduces a novel multi-AI-agent system designed to fully automate SLRs, and demonstrates how it substantially reduces the time and effort traditionally required for SLRs while maintaining comprehensiveness and precision.

Abdul Malik Sami, Z. Rasheed, Kai-Kristian Kemell et al. · 44 citations · ⚡2
#computer vision Feb 2024

Can Large Language Models Serve as Data Analysts? A Multi-Agent Assisted Approach for Qualitative Data Analysis

The proposed LLM-based multi-agent system automates qualitative data analysis process, creating opportunities for researchers and practitioners, and future improvements focus on enhancing multilingual performance and integrating continuous expert feedback.

Z. Rasheed, Muhammad Waseem, Aakash Ahmad et al. · 41 citations
#artificial intelligence Conference Open access Jun 2018

The Key Concepts of Ethics of Artificial Intelligence

It is suggested that the focus on finding keywords is the first step in guiding and providing direction for future research in the AI ethics field.

Ville Vakkuri, P. Abrahamsson · 39 citations · ⚡2

Related blog posts

MIT News · Artificial Intelligence Oct 15, 2026

AI model decodes the language cells use to communicate

A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.