PR #2112 — README Capability Claims Now Carry Their Evidence Threshold

The Broader Goal at the End of the Previous PR

The empirical-routing policy now sets a five-merged-job minimum for each harness and task type, and keeps the configured allocation as an interim default until that evidence exists. The capability report and cross-workspace measurement still depend on #1993 and canonical state work in #1926. The README should explain that evidence boundary wherever it describes harness routing.

Strategic Shifts in This PR (if any)

No strategy change. This is a correction to the product description so it matches the policy already committed in .synlynk/policy.json and explained in the empirical capability policy article.

What This PR Shipped

• Replaced the README's vague “best-measured harness” and “minimum sample size” wording with the explicit requirement of at least five completed, merged jobs for the relevant task type.
• Stated that routing follows the documented interim policy below that threshold.
• Linked the README claim to the empirical capability policy and the routing policy file.
• Audited the source website/ tree; it had no matching external claim to update.
• No code or runtime behavior changed. The check reviewed Markdown links, searched the README and website source for the old claim forms, and checked the diff for whitespace errors.

Design: docs/superpowers/specs/2026-10-07-readme-capability-evidence-design.md. Plan: docs/superpowers/plans/2026-10-07-readme-capability-evidence-copy.md.

Brainstorm Visuals Used

None. The wording follows the approved design spec and the existing empirical policy.

What This Achieved on the Path to Autonomy

Readers can distinguish measured harness allocation from the interim policy used while task-specific evidence is sparse. The copy no longer presents a future report as a shipped capability.

Strategic Note: The Goal at the End of This PR

The external routing claim now matches the five-sample policy. The next measurement milestone remains canonical state consolidation (#1926), followed by the capability report (#1993). This PR does not claim those measurements are already available.