Structured reasoning via API. For agent builders. No signup.
curl -X POST https://hummbl-api.hummbl.workers.dev/v1/try \
-H "Content-Type: application/json" \
-d '{"problem": "My agent makes shallow decisions"}'Improve RAG quality by retrieving reasoning frameworks alongside documents. When your agent pulls context, inject the right mental model for that problem type.
Models: P1 Stakeholder Mapping · DE3 Modularization · CO4 SynthesisGive your planner structured approaches to decomposition and sequencing. Stop shallow task lists—force rigorous breakdown of complex goals.
Models: DE1 First Principles · IN2 Premortem · RE1 KaizenBuild critics that reason systematically. Instead of vibes-based scoring, evaluators apply formal models to assess completeness and quality.
Models: IN3 Problem Reversal · SY1 Systems Thinking · P3 Red Teamimport requests
r = requests.post(
"https://hummbl-api.hummbl.workers.dev/v1/try",
json={"problem": "My agent needs better reasoning"}
)
models = r.json()["recommended"]// Add to claude_desktop_config.json
{
"mcpServers": {
"hummbl": {
"command": "npx",
"args": ["@hummbl/mcp-server"]
}
}
}We're mapping which models chain together for multi-step reasoning—7,140 possible pairs, 138 validated so far. The graph grows weekly as we score synergies, tensions, and sequencing between models.