library-backend-adoption-evallisted
Install: claude install-skill williamblair333/Uncle-J-s-Refinery
## When to use
When a new library emerges as a candidate to replace an existing backend. Especially valuable when:
- The current backend has caused recurring operational pain
- The replacement inverts a known architectural weakness (e.g., source-of-truth inversion)
- Benchmarks exist but scale validation does not
## Steps
1. **Prior art check** — search MemPalace for any prior analysis of the candidate library or migration topic.
2. **Fetch the discussion** — use `gh` CLI to pull the PR, issue, or community discussion. Prefer `gh` over WebFetch for GitHub URLs.
3. **Read the migration/handoff doc** — if a migration guide exists, read it and flag where it is more aggressive than the implementation (e.g., "opt-in" PR vs. "make it default" doc).
4. **Review implementation quality** — read the actual PR code, not just the description.
5. **Structure the analysis with these sections:**
- **What this is** — one paragraph: what the library does, its source-of-truth model, and any traction signal (stars, age, adoption).
- **Why the architecture matters for our specific situation** — connect the library's design to the *specific failure mode* of the current backend. Name the pain it would have prevented.
- **The numbers — honest assessment** — table with: metric | claim | credibility. Rate credibility as high/medium-high/medium/low. Flag where benchmark conditions differ from your scale.
- **What's unproven / risky right now** — numbered list. Include: library age,