Evidence base
What the system has actually read
Every number elsewhere in this application is a count over these documents. Categories below are derived from what was collected — not from a menu of sources the system wishes it had.
Evidence records
5,090
entity mentions
Documents
2,783
collected and normalised
Categories
3
with real data behind them
By category
Where the evidence comes from
| Category | Evidence | Documents | Last updated | Sources |
|---|---|---|---|---|
| Open-Source Repositories | 4,341 | 1,849 | 2026-08-18 | github |
| Job Postings | 727 | 421 | 2026-08-18 | gmail_linkedin_jobs, job_rss |
| News & Articles | 22 | 15 | 2026-08-18 | rss |
By source
Reliability, with a stated basis
First-party developer activity from the GitHub API. Leading indicator, not a hiring signal.
First-party job alerts delivered to the operator's inbox. Narrow by construction.
Real job postings from public RSS feeds (Indeed, LinkedIn). No API key required.
Publisher-controlled feeds. Timely, but headline-led and not first-party hiring data.
Ratings are stated editorial assessments with their reasoning attached — not computed scores. Nothing here is inferred from popularity or volume.
Coverage gap
Source types this system does not read
- Research PapersWe don't yet read academic research databases like Google Scholar or PubMed. Academic citations are not included in these trends.
- Company Career PagesWe don't yet pull directly from company websites or job application systems (like Greenhouse or Workday). Only the connected job feeds are read.
- Government DataWe don't yet include official labour statistics from sources like the US Bureau of Labor Statistics. Government employment data is not reflected here.
- SEC FilingsWe don't yet read company financial filings. Hiring plans mentioned in earnings reports or 10-K filings are not captured.
- User Uploaded DocumentsThere is no way to upload your own documents yet. Only the configured data sources are read automatically.
These are listed rather than hidden. Showing them as empty categories would imply coverage that does not exist; each one is a collector still to be built.