About Toolwitness
Publisher: toolwitness.ai, an Organization. Agents discover tools; deterministic Python gates, Codex verification and Grok refutation form the machine-checking process. No human review is implied. Corrections: hello@toolwitness.ai. Read the method and limits.
Toolwitness publishes snapshot exercise evidence for Claude Code hooks, skills and local MCP servers. 0 fictional examples; 0 reproduced results. 1 offline listings (excluded from exercised results).
Requests, configured workflows, reported completions and reproduced results are separate evidence states. Reproduced results, explicit examples, labeled component pilots and offline listings are admitted here. Offline listings cover initialize and tools/list only; no tool was called and no model ran. Approved records display Machine-checked and the review date; unapproved records are previews only.
Sponsor toolwitness
Builders of AI-agent tools and services can sponsor a labelled paid placement. Placement never changes evidence or ordering. Network sites are publisher links, separate from paid placements. Contact hello@toolwitness.ai to discuss a placement.
No traffic, revenue, ranking or citation outcome is claimed.
Get new snapshots
Follow the machine-checked snapshot feed for published evidence updates.