Technical Research Pack

Serves: adopt a technology area, design an integration approach, ground an architecture in current practice, assess feasibility before committing a roadmap.

Dimensions (priority order — prune to the decision):

1. Landscape & maturity — dominant approaches, what's consolidating vs churning, what the current generation newly makes possible
2. Integration & interoperability — protocols, formats, auth patterns, where integrations actually hurt
3. Architecture patterns in practice — which named patterns dominate at what scale, and what the failures teach
4. Implementation reality — learning curve, tooling, operational burden, what teams say 6–12 months in
5. Ecosystem health — contributor/release vitality, backing durability, the five-year regret risk

Craft (the non-obvious): read the retrospective threads, not the launch threads; favor accounts with production numbers over advocacy; before citing a pain point, check whether it was since fixed — an old complaint against a current version is a false claim; read repository metrics over time, never snapshots; issue trackers reveal the gap between docs and reality.

Freshness: versions & compatibility ≤ 1 mo · ecosystem signals ≤ 6 mo · landscape ≤ 12 mo (AI-adjacent ≤ 3 mo) · patterns ≤ 2 yr.

Two-source classes: version/compatibility claims; performance or scale numbers a recommendation rests on; claims that a technology or pattern failed — one post-mortem is an anecdote.

Feeds (bmm): architecture spine (candidate paradigms, operational constraints) · brief (feasibility) · roadmap risk and estimates.