Help and evaluation support
If you are evaluating or using Setsumei, start with the checks below for the generated HTML brief, analysis refresh or MCP. For support, include your package version if known and a high-level description of the task; keep source, analysis results, credentials and private paths out of the message.
Get to the right contact.
Setsumei product and evaluation questions
Describe what you want to do and your language/build context. We can discuss fit or supply package-specific setup guidance.
RunQC accounts and agent testing
RunQC has separate account, guide and support routes.
Open the generated HTML to explore your scan.
Check whether analysis has finished for the repository you intended to scan. Follow your package instructions to locate the generated HTML, then open it in a browser to view the GUI scan brief. Opening the brief is separate from connecting an MCP client.
Start with the overview and prioritized findings, then inspect coverage, relationships and supporting source for your task. Model-assisted investigation is optional and separate from deterministic analysis.
Cannot locate or open the brief? Ask about your scan or brief with your package version and the high-level step that did not work. Explore the example brief.
Match the analysis to the code you are changing.
Check whether relevant source changed after the analysis. If it did, refresh before relying on earlier relationships or findings; stored context does not update continuously on its own. Check the analyzed source state and scope alongside the result.
If analysis is in progress or an MCP answer reports a warning, use the freshness, query-bound and provenance checks. Some tools can return warned results during analysis; when completeness matters, wait for completion and re-query.
Still unsure which analysis to use? Ask an analysis question. Describe the task and whether relevant source changed.
Check the MCP context before relying on an answer.
Check your package instructions and compatible client setup. Confirm the intended repository has completed analysis, then inspect the response’s freshness, scope and result limits. MCP queries stored analysis through a local read-only interface; opening the HTML brief does not establish this connection.
Review the client’s independent tool permissions and model-provider data handling before connecting private context. Use the freshness and query checks if source changed, results are bounded or analysis is in progress.
Need package-specific setup help? Tell us your client, package version if known, intended task and whether analysis has completed.