← Files NeoSigmaARCHIVED FILE
.codex-plugin/plugin.json
5.95 KB · Nov 17, 2026 · 06:14 UTC
{
"name": "neosigma",
"version": "1.7.5",
"description": "NeoSigma skills and MCP tools for coding agents: integrate the NeoSigma SDK (agent tracing and product events), import a codebase's existing checks as verifiers, suggest verifiers that catch missed production failures, convert benchmarks into Harbor tasks, write and verify a workspace's setup script, and read traces, issues, datasets, and evaluation runs through the NeoSigma MCP server.",
"author": {
"name": "NeoSigma",
"email": "founders@neosigma.ai",
"url": "https://neosigma.ai"
},
"homepage": "https://neosigma.ai",
"repository": "https://github.com/neosigmaai/skills",
"license": "MIT",
"keywords": [
"neosigma",
"llm",
"observability",
"tracing",
"product-events",
"opentelemetry",
"agents",
"sdk",
"verifiers",
"evals",
"mcp",
"harbor"
],
"skills": "./skills/",
"mcpServers": "./.mcp.json",
"interface": {
"displayName": "NeoSigma",
"shortDescription": "Find and fix agent failures",
"longDescription": "Find out why your AI agent fails in production, and test the fix next to your code.\n\nUse it to:\n- Search production sessions and read traces to find failures, including runs that look successful but are not.\n- Read the issues NeoSigma found and the sessions behind each one.\n- Turn failing sessions into proposed evals, review them, and save them to a dataset.\n- Run experiments on a dataset and compare results across agent versions.\n- Add NeoSigma tracing to your agent, import your existing checks as verifiers, and set up the environment your evals run in.\n\nWho it is for: teams that build and run AI agents and already send traces to NeoSigma, or want to start.\n\nLimits: you need a NeoSigma account, and you sign in on first use. The plugin reads and changes data only in the NeoSigma projects your account can access. It does not change your agent in production. Running an experiment uses your organization's NeoSigma usage.",
"developerName": "NeoSigma",
"category": "Developer Tools",
"capabilities": [
"Search production sessions and read agent traces",
"Read issues and the sessions behind them",
"Propose evals from failing sessions and save them to datasets",
"Run experiments and read their results",
"Set up tracing, verifiers and eval environments"
],
"websiteURL": "https://neosigma.ai",
"supportURL": "https://docs.neosigma.ai",
"privacyPolicyURL": "https://neosigma.ai/privacy",
"termsOfServiceURL": "https://neosigma.ai/terms",
"defaultPrompt": [
"Find sessions this week that looked successful but weren't, and explain what went wrong",
"Turn my agent's recent production failures into evals and save them to a dataset",
"Compare my last two experiments: which tasks got better and which got worse?"
],
"brandColor": "#0090FF",
"brandColorDark": "#0090FF",
"composerIcon": "./assets/icon.png",
"logo": "./assets/icon.png"
},
"extensions": {
"com.openai": {
"review": {
"test_cases": {
"positive": [
{
"description": "Find recent production failures",
"prompt": "Which of my agent's recent sessions failed, and why?",
"tools_triggered": "list_projects, search_sessions, get_session",
"expected_behavior": "Finds the person's project, searches recent sessions for failures, reads a few failed sessions, and summarizes the main causes with links to the sessions."
},
{
"description": "Explain one failed session",
"prompt": "Walk me through my agent's most recent failed session step by step.",
"tools_triggered": "get_session, list_session_turns, list_trace_spans",
"expected_behavior": "Finds the most recent failed session, reads its turns in order, and explains what the agent did at each step and where it went wrong, with a link to the trace."
},
{
"description": "Compare two experiments",
"prompt": "Compare my two most recent experiments. Which tasks got better or worse?",
"tools_triggered": "list_experiments, list_experiment_trials",
"expected_behavior": "Opens the two most recent experiments, compares their pass rates on the tasks both ran, and names each task whose result changed, or says that none changed."
},
{
"description": "Read an experiment result",
"prompt": "How did my latest experiment do? Which tasks failed?",
"tools_triggered": "list_experiments, get_experiment, list_experiment_trials",
"expected_behavior": "Opens the most recent experiment, reports the overall score and the number of passed and failed tasks, and names the failed tasks with their trial details."
},
{
"description": "List datasets and their tasks",
"prompt": "What eval datasets do I have, and what is in the biggest one?",
"tools_triggered": "list_datasets, list_dataset_tasks",
"expected_behavior": "Lists the datasets in the project, picks the one with the most tasks, and summarizes what its tasks test."
}
],
"negative": [
{
"description": "Unrelated general question",
"prompt": "What is the capital of France?"
},
{
"description": "Unrelated coding task with no agent data",
"prompt": "Write a Python function that reverses a string."
},
{
"description": "Request for another product's data",
"prompt": "Show me my Datadog APM error rate for the checkout service."
}
]
}
},
"publication": {
"release_notes": "A shorter first line for the directory description. No change to skills or tools."
}
}
},
"apps": "./.app.json"
}
SHA-256: 736366538eaf959bac0aa3bd4ec91e481b4bd6588ab4504489efdb74a88fa15b