Tags

Browse the KB by topic. Each tag has one head here, <tag>-README.md, that says what the tag gathers and picks its load-bearing entries; the published site appends the full member listing across the participating collections. See COLLECTION.md for which collections' tags count and how to write a head.

For current-state documentation about the live Commonplace system rather than transferable theory, start at ../reference/README.md.

Tag Indexes

  • Self-improving systems — evidence-responsive operative change to a system's own organization; reflective versus non-reflective self-improvement is the central distinction, and the stake is a conjectured sample-efficiency payoff: reflection may cut target data under structured shifts
  • Theory builder — the theory-builder definition and its companions; the research program Commonplace runs
  • Software factory — factories and houses: family-specific production machinery and factory learning
  • Improvement loop — the proposal-selection architecture: search control, evaluation, retention
  • Reflection — reflective systems: addressability, second-order lessons, graded coverage
  • Warranted autonomy — which decisions a computational actor is warranted to take over, and how far
  • Continual learning — how a deployed system keeps learning outside model weights
  • Architecture — how Commonplace is structured and installed: repo layout, control-plane design, file-based storage
  • Evaluation — what works, what doesn't, what needs testing
  • Learning theory — how systems learn, verify, and improve; covered by six child tags (enforced), so its README routes rather than enumerates:
  • Deploy-time learning — the framework: adaptation through durable inspectable artifacts
  • Constraining — narrowing interpretation space, codification, relaxing
  • Discovery — positing generals, recognizing instances; explanatory-reach
  • Artifact analysis — the four-field vocabulary for retained artifacts
  • Agent memory — memory architecture and policy (also a top-level area)
  • Document system — document types, writing conventions, and validation rules for structured notes
  • Computational model — programming-language framing for LLM instructions and orchestration
  • Context engineering — getting the right knowledge into bounded context at the right time
  • Failure modes — characteristic ways knowledge can exist without changing agent behavior
  • Links — link semantics, navigation behavior, and link-management methodology
  • Type system — why documents have types, what roles they serve, how structured writing improves quality
  • LLM reliability — why LLM output deviates from intent (underspecification, interpreter failure, indeterminism) and the machinery for catching and correcting it: oracle theory, error correction, architectural separation
  • Observability — making hidden state, hidden failure, and quality drift visible enough for operators and maintenance loops to act on
  • KB maintenance — operations, audits, and maintenance methodology
  • Review system — gates, verdicts, freshness, protocol, and the review proposals
  • Claims and grounding — title as claim, modality, quotes, source grounding, ground truth
  • Curation — indexes and tag heads, quality signals, hygiene, capacity, retirement
  • Methodology — how an agent selects, borrows, and is controlled by a methodology
  • Trace-learning — external systems that learn from their own agent traces through a raw-to-distilled loop
  • Related systems — external systems tracked for comparison and convergence signals

Workshop Layer

Gaps

Complete file listing (generated at build time)