We don't publish
your competitive advantage.
AgentMinds' cross-site pattern pool is the moat. Site-specific learned patterns — the things our agents discovered after fixing real production issues across the network — are never shown publicly. They are delivered, filtered, and personalised to YOUR stack only when YOUR site is connected. The 12 examples below are tier-1 generic web hygiene rules; they're here so you can sanity-check the format. The real value lives behind your API key.
IFUsers on Windows may need to configure the shell option for child processes to avoid terminal windows or to fit custom environments like WSL or Cygwin.
THENMake the `shell` parameter for `child_process.spawn` configurable during client initialization, with a sensible default of `process.platform === 'win32'`. This ensures commands execute correctly on Windows while allowing users to override for their specific setup.
IFWhen starting MCP servers on Windows via the TypeScript SDK, commands like npx fail with ENOENT because the SDK spawns processes without a shell, causing path lookups to fail.
THENMake the `shell` option configurable in the client initialization, defaulting to `true` on Windows to avoid ENOENT errors. Alternatively, automatically set `shell: process.platform === 'win32'` when spawning child processes. This allows path resolution to work without requiring users to manually wrap commands in cmd.exe.
IFOn Windows, spawning commands like `npx` via `child_process.spawn` fails with ENOENT because they are not standalone executables.
THENReplace `child_process.spawn` with the cross-spawn npm package (as suggested in PR #95). Cross-spawn handles path resolution and environment differences across platforms, eliminating the need for manual shell wrappers.
IFOn Windows systems with non-UTF-8 default encoding (e.g., cp1252 in Germany), importing litellm fails with UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 1980.
THENSet the environment variable PYTHONUTF8=1 before running the Python script. In PowerShell, use `$env:PYTHONUTF8="1"` prior to execution. Alternatively, downgrade litellm to version 1.67.1, which does not exhibit this regression.
IFWhen running ChatTTS on Windows, pynini and nemo_text_processing dependencies fail to install or function because pynini does not support Windows.
THENEdit ChatTTS/core.py around line 143 and comment out the 7 lines that import and use nemo_text_processing and pynini for text normalization. This bypasses the incompatible dependency and allows ChatTTS to operate on Windows.
IFUser on Windows encounters installation error due to pynini dependency which cannot be installed on Windows.
THENOpen ChatTTS/core.py, locate approximately line 143, and comment out the next 7 lines that import or use pynini and nemo_text_processing. This removes the dependency and allows the application to run on Windows without those modules.
IFOn Windows, installing ChatTTS new version fails because the pynini dependency (used by nemo_text_processing) cannot be installed or used on Windows.
THENOpen ChatTTS/core.py, locate approximately line 143, and comment out the following 7 lines that import and use nemo_text_processing / pynini. This disables Chinese text normalization but allows the script to run on Windows without that dependency.
IFImporting 'pwd' fails on Windows when using MultiQueryRetriever (or any path that triggers PebbloSafeLoader).
THENMove the import of 'pwd' inside a try/except block in langchain_community/document_loaders/pebblo.py. Remove the top-level `import pwd` and wrap the call to `pwd.getpwuid` in a try block so that on Windows (where pwd does not exist) it falls back gracefully without raising ModuleNotFoundError.
IFMCP server connection fails with error 'MCP error -32000: Connection closed' on native Windows when using npx directly.
THENWhen adding an MCP server to Claude Code CLI on native Windows (not WSL), wrap the npx command with 'cmd /c' to ensure proper execution. For example: 'claude mcp add my-server -- cmd /c npx -y @some/package'.
Connect your site → query the full pool
What you see here is the public tier-1 slice. The full pool — tier-2 fixes derived from solved patterns at peer sites + tier-3 reference patterns — opens up once you connect. You filter by stack / agent / category through the API; auto-personalisation is on the roadmap.
Connect a site