[pull] main from regenrek:main

.cursor/commands/ask-codefetch-pro.md

@@ -0,0 +1,49 @@
+# Ask Codefetch Pro - Open GPT/Gemini/Claude
+
+Generate codebase, copy to clipboard, and open AI chat in browser.
+
+## Instructions
+
+1. Research: Analyze PROMPT to identify relevant files/directories
+2. Scope: Decide what to include:
+   - `--include-files` for file patterns (globs, comma-separated): `"src/auth/**/*.ts,src/api/**/*.ts"`
+   - `--include-dir` for directories (comma-separated): `"src/auth,src/utils"`
+   - Combine both if needed
+3. Run: Execute codefetch open with scope
+4. Paste: Codebase is in clipboard, paste into AI chat and ask your question
+
+## Run
+
+```bash
+# Default: opens ChatGPT, copies codebase to clipboard
+npx codefetch open --max-tokens 50000 \
+  --token-encoder o200k \
+  --project-tree 3 \
+  --include-files "src/**/*.ts" \
+  --include-dir "src"
+
+# Gemini
+npx codefetch open --chat-url gemini.google.com --chat-model gemini-3.0 \
+  --max-tokens 50000 --include-files "src/**/*.ts"
+
+# Claude
+npx codefetch open --chat-url claude.ai --chat-model claude-3.5-sonnet \
+  --max-tokens 50000 --include-files "src/**/*.ts"
+
+# Copy only, no browser
+npx codefetch open --no-browser --max-tokens 50000 --include-files "src/**/*.ts"
+```
+
+## Options Reference
+
+- `--chat-url <url>` - AI chat URL (default: chatgpt.com)
+- `--chat-model <model>` - model parameter (default: gpt-5-1-pro)
+- `--chat-prompt <text>` - custom prompt message
+- `--no-browser` - copy only, skip browser
+- `--include-files "pattern1,pattern2"` - file globs
+- `--include-dir "dir1,dir2"` - directories
+- `--exclude-dir "test,dist"` - exclude directories
+- `--exclude-markdown` - exclude .md files
+- `-e .ts,.js` - filter by extension
+- `--max-tokens 50000` - token limit
+- `--project-tree 3` - tree depth

.cursor/commands/ask-codefetch.md

@@ -0,0 +1,38 @@
+# Ask Codefetch - Context Gather
+
+Gather codebase context for agent analysis.
+
+## Instructions
+
+1. Research: Analyze PROMPT to identify relevant files/directories
+2. Scope: Decide what to include:
+   - `--include-files` for file patterns (globs, comma-separated): `"src/auth/**/*.ts,src/api/**/*.ts"`
+   - `--include-dir` for directories (comma-separated): `"src/auth,src/utils"`
+   - Combine both if needed
+3. Generate: Run codefetch with scope, limit 50k tokens
+4. Analyze: Use the generated markdown for your response
+
+## Run
+
+```bash
+# Research codebase, determine relevant files/directories based on PROMPT
+# Then generate context (limit ~50k tokens):
+
+npx codefetch --max-tokens 50000 \
+  --output codefetch/codebase.md \
+  --token-encoder o200k \
+  --project-tree 3 \
+  --include-files "src/**/*.ts" \
+  --include-dir "src"
+```
+
+## Options Reference
+
+- `--include-files "pattern1,pattern2"` - file globs
+- `--include-dir "dir1,dir2"` - directories
+- `--exclude-dir "test,dist"` - exclude directories
+- `--exclude-markdown` - exclude .md files
+- `-e .ts,.js` - filter by extension
+- `--max-tokens 50000` - token limit
+- `--project-tree 3` - tree depth
+- `--copy` - copy to clipboard