Proposals Directory

Type: kb/types/index.md

← Parent

• A reader-facing banner for unreviewed notes (note) - Proposal: warn readers of unreviewed notes with a standardized top-of-note banner derived from review-state, kept consistent with the canonical status by a validator check
• Automated note refinement as a search over a fixed source bundle (note) - Proposal: build the automated note-refinement loop on a source bundle that emits a set of notes, not a single identity-stable note — reframing non-convergence (split, drift, kill) as search outcomes
• Backlink surfacing (note) - Proposal: how Commonplace could surface inbound links — build-time web rendering for humans, on-demand inversion for agents, curated symmetric labels for tension; committed generated footers and full manual bidirectionality are foreclosed
• Gate learning from accepted edits (note) - Proposal: learn review gates from accepted note edits, with candidate mining, usefulness tracking, lifecycle promotion, and budgeted loading. Atomic gates shipped; learning has not
• Harness-orchestrated review sweeps (note) - Proposal: run review sweeps through a harness's sub-agent orchestration feature (workflow scripts) instead of the Python thread pool — wiring validated by one Claude Code experiment, adoption deferred until comparable orchestration surfaces exist in more than one harness
• Review configuration object (note) - Proposal: a ReviewConfig object owned by commonplace.review, consolidating the review subsystem's project-shape constants (scan roots, gates root, artifact root, db path, sweep parallelism) with per-project override — deliberately not a global ProjectConfig, so the experimental db-backed review subsystem stays decoupled from the stable core
• Structured-output codec for the review protocol (note) - Proposal: make the review protocol's output encoding a pluggable codec — sentinel-delimited markdown today, schema-validated structured output when harnesses expose it — so decision parsing stops depending on free-text conventions