Tigerless Labs
The GitHub organization that publishes autoharness, the self-maintaining Claude Code skill layer.
A thin node: the only evidence so far is the repo itself, so nothing is asserted about the org
beyond its authorship of autoharness. Retracted 2026-09-03 (absence-claim drain). There is more
than the one repo now: the GitHub organization carries a second project, cost-xray ("see what
Claude Code and Codex actually send to the API — and what each part costs"), an X account
(@Tigerless_labs), and a self-description as an open lab working on engineering research and tooling
for AI workflows. A third-party write-up of autoharness appeared in September 2026 on a personal
technical blog — coverage rather than evaluation, and tiered accordingly.
Still asserted narrowly: nobody has evaluated anything the org ships, and the second repo is the same kind of evidence as the first. What changed is that the org is no longer a single artifact, so the page can say what it works on.