docs(skills): record agent report release evidence

[Aymericr]

Records the immutable 0.1.8 skill release, exact public commit and hashes, package checks, and the limits of the evidence.

The record explicitly preserves package-checks-only: the local Claude OAuth session could not refresh, so this does not claim a native host cohort, marketplace listing, hosted deployment, adoption, or retention.

Validation:

• downloaded tag archive matched the recorded .mcp.json and skill hashes
• bun scripts/validate-skills.ts
• claude plugin validate . --strict
• git diff --check

---

Note

Low Risk
Documentation-only update to validation bookkeeping; no runtime, auth, or skill behavior changes in this diff.

Overview
Updates skills/VALIDATION.md to document bundle 0.1.8 and bump the recorded release metadata (package 0.1.8, furniture-fit 0.1.4, September 10, 2026).

Adds a new Bundle 0.1.8 agent-report footprint pre-check section that captures what shipped: optional consent-gated Pascal calculator links from furniture-fit, strict omission rules for bad or scene-dependent inputs, immutable tag/commit/SHA-256 evidence, CI and package eval scope, and Fable source review. It explicitly keeps package-checks-only because native Claude host validation did not run (OAuth refresh failure), and states no claims for marketplace, hosted deployment, adoption, or retention.

Older sections (0.1.7 MCP release, MCP tool-annotation candidate, etc.) are unchanged aside from the new block inserted above them.

^{Reviewed by Cursor Bugbot for commit bc4b798. Bugbot is set up for automated code reviews on this repo. Configure here.}

[pascal]

I hit an error while handling your request (Model unavailable on AI Gateway free tier: Free tier users do not have access to this model. Upgrade to paid credits at https://vercel.com/d?to=%2F%5Bteam%5D%2F%7E%2Fai%3Fmodal%3Dtop-up for unrestricted…).

Please try again, rephrase, or reach out if it keeps failing.

Error id: 7c0be2de-c420-43b5-896d-c1f78f35095b