Automatic analysis: identify encoding and architecture errors for DollhouseMCP PR #2997

Hello Sonar team,

Please investigate the provider-hosted automatic-analysis logs for this public project and PR analysis:

Repository: GitHub - DollhouseMCP/mcp-server: A Free, Open Source MCP server for dynamic custom persona management with public a GitHub collection of personas, skills, templates, and other elements for AI models. · GitHub
Pull request: fix(security): refuse guarded background validation before mutation by mickdarling · Pull Request #2997 · DollhouseMCP/mcp-server · GitHub
Sonar project key: DollhouseMCP_mcp-server
Organization: dollhousemcp
Analyzed commit: 79468188a31f4816326432b48340b8f946f16b8a
Background task ID: AaENUUfw8NE9f5k1r6yN
Analysis ID: 9d72c813-2430-4215-91c3-71092fae3030
Task execution: 2026-10-05 18:26:33–18:26:36 UTC
PR scope: SonarQube Cloud

This is the PR’s published head at the time of this request. The exact task’s Show Warnings dialog was checked in the authenticated UI after completion.

The task reports SUCCESS and the quality gate passes, with zero reported PR issues and hotspots. However, this exact task’s Show Warnings action displays:

  1. “There are problems with file encoding in the source code. Please check the scanner logs for more details.”
  2. “Architecture analysis executed with errors. Analysis results may be incomplete or invalid.”

The same warning texts were verified in the earlier PR analyses:

  • Commit 97e23b477f63d7e8b0cf6d77574f4cb29b7b6205, background task AaENK8rx8NE9f5k1r3_M, analysis 68e42ebe-cb28-400b-8d86-e51f0ab26486.
  • Commit 0ae948715e35fe8fd18a0afc9578b364a3f7a957, background task AaEKP9ZBTbi2tNkwrn2P, analysis 95b72525-3d4c-4264-9d1f-3f834440cddc.
    This establishes recurring wording, not a known common cause.

The accessible scanner context for the retained 97e analysis confirms sonar.autoscan.enabled=true. That task reports logs:false; the background-task UI exposes context and warnings but no scanner-log download, and the inspected repository has no configured Sonar scanner job in its GitHub workflows. Your guidance explains that automatic-analysis scanner logs run on Sonar infrastructure: Where can I find the "scanner logs" in the SonarQube Cloud UI? - #7 by stevan.vanderwerf

As a bounded local check of retained commit 97e, we strictly decoded its 2,631 tracked nonbinary/nonempty text candidates after applying the existing .sonarcloud.properties file-scope exclusions. All decoded as UTF-8. We also inspected the remaining 45 tracked files’ names, extensions, and MIME types: five PNG images and 40 WOFF2 font assets, with no source/text-extension candidates. This check was not repeated for 79468188 and is not Sonar’s indexed-file manifest, so it does not identify or clear the encoding warning.

Could you provide:

  • The exact affected filenames, scanner-selected encoding, byte/line positions, and underlying encoding diagnostics.
  • The exact architecture exception/error and the relevant analyzer stage/version or origin.
  • Which files or analysis results were omitted, incomplete, or potentially invalid, and whether either warning affects PR #2997’s security/reliability analysis or quality-gate result.
  • The smallest supported repository/configuration correction, or provider-side correction if applicable, and how to verify complete analysis afterward.

We are holding merge pending a concrete disposition. We have not changed exclusions, disabled architecture analysis, suppressed findings, or treated the passing quality gate as proof of complete analysis. Please distinguish confirmed causes from hypotheses.

Thank you.

Hey @mickdarling, welcome to the Community!

Regarding encoding warnings:

The scanner logged 46 Invalid character encountered ... for encoding UTF-8 warnings:

  • 40 .woff2 fonts under src/web/public/fonts/ and src/web-console/ui/fonts/
  • 5 .png files:
    • docs/assets/portfolio-browser.png
    • docs/assets/dollhouse-logo.png
    • docs/assets/dollhouse-logo-orange.png
    • src/web-console/ui/dollhouse-logo.png
    • src/web/public/dollhouse-logo.png
  • tests/unit/web-console/modules/IntegrationOperationCatalog.test.ts, line 52 (the line contains a literal U+FFFD replacement character: expect(result.generatedSkill?.content).not.toContain('�');)

The impact is that these files were not analyzed, the solution is to exclude from analysis files that are not supposed to be analyzed (add **/*.woff2 and **/*.png to sonar.exclusions in .sonarcloud.properties) and to fix the characters with broken encoding in the typescript file.

Regarding architecture anlaysis:

Two files caused an error when processed by the architecture analyzer: src/config/safeConfigDeletion.ts and tests/security/framework/RapidSecurityTesting.ts. This looks more like a bug on our side, so I’m going to flag it for the team to investigate.

The impact is that these files are excluded from the architecture graph, but they’re still analyzed for normal issues and the quality gate and metrics are not affected.