FORKIT EVIDENCE NETWORK · COMMUNITY PREVIEW
PRODUCT PROOF.
NOT VANITY METRICS.
What Footprint's release tests support, what opted-in machines report, and what we cannot yet claim. Technical validation and real-world adoption are different measurements.
01 / REPRODUCIBLE ENGINEERING EVIDENCE
Validated on the release candidate.
As recorded in the project's September 2, 2026 validation log for v0.2.6. Test fixtures are not field-accuracy measurements.
Five evidence surfaces: recognized agent processes, installed tools, local runtimes, local model metadata and MCP configuration presence. Model available is not model loaded. Agent process presence is not proof of an action, model output or sandbox escape.
02 / REAL COMMUNITY EVIDENCE
Community Preview observations
No public contributor totals until a reviewed, consented, deduplicated aggregate is available. We never count npm downloads or page views as users.
Collecting opt-in evidence. No authorized, validated public aggregate snapshot is published. Sample size: not available. Data shown here never includes raw events, identities, prompts, code, paths or model names.
Community Preview signatures protect integrity and continuity, not unique hardware or independently verified activity. MCP/tool presence and multi-tool rates will appear only when an approved aggregate supplies those measurements with a meaningful denominator. Reconstruction demo clicks are marketing interactions, not real actions reconstructed.
03 / SEE THE LIMIT
Know what the evidence can actually prove.
Footprint
See supported AI systems active on your Mac, local model inventory and tool/MCP signals. Detailed history stays local.
Passport
A persistent system identity and governance context. Not an external certificate.
Reconstruct
Inspect source-backed example claims and evidence gaps, not live customer actions.
BE PART OF THE EVIDENCE NETWORK