docs: refresh recent changes and companion repositories
Experiment quality + classical-ML demo rows; E-023 REPL-docs resolution
demos: Data Forge learning path + errata E-021 (data-forge feature note)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
docs: audit glossary coverage and README accuracy
Document the glossary's builtin coverage guard, all-match search, new running_sum and running_product scan vocabulary, and added ML concepts including ICRL, MTP, mHC, reasoning architectures, and energy models. Add errata for resolved builtin/concept discoverability gaps and the remaining README inaccuracies: deprecated names, obsolete paths, workspace-invalid commands, development-status prose, count-policy contradiction, and demo-coverage overclaim. Preserve the Resources page links to repository planning, status, changelog, and history records.
docs: separate user guidance from development history
Rewrite public wiki pages around present capabilities, usage, and explicit unsupported boundaries. Remove saga, roadmap, milestone, release-status, and planned/deferred language from user guidance; update MLX resident-training and CUDA fallback descriptions; replace deprecated iota examples with range; and limit Engram and mHC wording to implemented behavior. Expand the errata taxonomy and record the wiki sweep. Keep repository planning, status, changelog, and history links in a clearly labeled development-records section on Resources and Reference.
demos: note the Thinking in Arrays loop-pedagogy demo (c909864f); errata E-016 feature note
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
docs: build comprehensive sw-MLPL wiki
Add onboarding, language, use-case, execution-surface, architecture, learning, and resource pages with Mermaid diagrams. Document browser, CPU, MLX, CUDA, and remote-server capabilities and limitations. Add a source-documentation errata audit, clarify Engram as in progress, and identify mHC as planned soon.