/freesearch — Zero-Token Research via Argus

Uses 0 Claude Code tokens. Calls Argus in cheap mode (SearXNG).

Usage

/freesearch [topic]

Process (IMPORTANT: Follow this order)

1. Check global docs-cache FIRST:

   cat ~/github/docs-cache/docs/cache/.index.md | grep -i "[KEYWORD]"
   

   If found → Return cached doc path immediately

2. If NOT in cache → Ask 2-3 clarifying questions (goal, depth, audience)

3. Search via Argus (cheap mode = SearXNG only):

   curl -s -X POST http://100.112.130.100:8270/api/search \
     -H "Content-Type: application/json" \
     -d '{"query": "[TOPIC]", "mode": "cheap"}'
   

4. If results include official docs → Add to global cache:

   cd ~/github/docs-cache
   mkdir -p docs/cache/{category}/{name}
   # Write README.md with content
   # Update .index.md
   git add docs/cache/
   git commit -m "Add cache: {name}"
   git push
   

5. Create docs/research/{date}_{topic}_final.md in current project

Output

[CACHED] Found in docs-cache: ~/github/docs-cache/docs/cache/tools/anthropic/README.md

OR

[Searched Argus: cheap mode] Key findings:
- [finding 1]
- [finding 2]

[Added to cache] ~/github/docs-cache/docs/cache/tools/{name}/README.md
Full research: docs/research/YYYY-MM-DD_{topic}_final.md

Fallback

If Argus is unreachable:

ssh homelab-ts "curl -s http://localhost:8270/api/health" >/dev/null 2>&1

...then check config/search.yaml for cheap mode providers and call directly.

Notes

• Research takes 5-15 seconds
• Check cache BEFORE searching
• Only add official docs to cache (not random blog posts)
• Cache location: ~/github/docs-cache/ (global, not per-project)