Project · public demand probe

MCP Demand Proxy Collector

A zero-spend 7-day probe that snapshots public Glama recent/deep MCP search feeds and the Official MCP Registry, normalises repeated search intent, joins it to registry supply metadata, and emits JSON/CSV decision aids.

Boundary: Every output is explicitly Public search-intent proxy; not product-usage telemetry. It does not infer private usage, users, conversions, or revenue. All sources are public and require no authentication.

Sources

Glama recent searches: recent-searches.xml
Glama deep searches: deep-searches.xml
Official MCP Registry: registry API

Method

Daily snapshots are idempotent by UTC date at data/snapshots/YYYY-MM-DD.json; cumulative output is output/report.json + output/report.csv. Run once daily for seven dates. Success after the window means ≥3 recurring demand clusters with an operator decision (missing_tool, positioning_opportunity, crowded_category); otherwise treat the feeds as weak/noisy.

Verification

Run ./run.sh then python3 -m unittest discover -s tests -v. Inspect output/report.json for the normalised demand clusters and source URLs. All outputs are reproducible from the public feeds.