rite-learnlisted

# /rite-learn: review durable lessons Use native session memory and reviewed project Markdown instead of an engine miner, score, nudge, or convention registry. This command proposes durable guidance; it never turns an inferred pattern into a rule on its own. ## Modes - `/rite-learn` reviews recurring evidence across shipped work. - `/rite-learn "<lesson>"` evaluates one proposed lesson or rejected direction. ## Workflow 1. Read the applicable `AGENTS.md` or `CLAUDE.md`, accepted ADRs, and relevant `.devrites/archive/*/{decisions,drift,review,seal}.md` files. Use native host memory or session history when available, but verify every durable claim against repository evidence. 2. For a broad review, ask the exact `devrites-retrospector` agent to inspect the bounded archive in a fresh read-only context. Wait for its result and reconcile it against the cited files. 3. Keep a candidate only when it is specific and supported by either: - the same correction in at least two distinct features; or - one explicit, durable product/architecture decision with its rationale. 4. Choose the smallest authoritative home: - `AGENTS.md`, `CLAUDE.md`, or a scoped standards document for an operating rule; - an ADR for a significant durable architecture decision; - the active feature's `decisions.md` for a feature-scoped decision; - no change for a one-off, stale, or generic observation. 5. Show the evidence and exact proposed edit. Apply it only after the