Borrowable Features

Decision

Do both experiments in parallel: try the full ARIS-shaped Research Wiki inside a workshop, and separately borrow individual lifecycle patterns into commonplace.

The full subsystem trial protects against premature decomposition: some value may come from the whole loop, not any one artifact. The incremental path protects against overfitting commonplace to an ML-research schema when a smaller convention would solve the general problem.

Borrow Now

1. Dynamic working README. Generate or refresh a bounded orientation document for active workshops. ARIS calls this query_pack.md; in commonplace the better concept may be a separate working README that includes the question, current gaps, active claims, failed ideas, selected sources, experiments/probes, and open decisions. This is the most directly useful feature because it changes what future agents load.

2. Failed ideas as first-class memory. Failed ideas should not disappear into logs. For design work, knowing what not to retry is often more valuable than another source summary.

3. Claim/experiment loop. Active claims need states and links to probes. This fills the gap between ingest reports and durable structured claims.

4. Lifecycle lint. Add checks for stale claims, dead ideas, orphan source cards, open gaps without active ideas, and promoted artifacts that still appear as active.

5. Mutation log. Every helper-driven change should leave a receipt. This is cheap and improves recovery after compaction or handoff.

Adapt Carefully

1. Edges as source of truth. A JSONL edge file is useful for workshop-local operational relationships, especially for generating query packs. It should not replace markdown links in durable library artifacts.

2. Paper cards. ARIS paper pages are useful, but in commonplace they should be workshop cards pointing to kb/sources/ snapshots and ingest reports, not replacements for them.

3. Gap map. Useful for active investigations, but gaps should not become a permanent parallel taxonomy unless they promote into indexes or notes.

4. Re-ideation triggers. The triggers are promising: enough new sources, enough failed ideas, a contradiction, or an unaddressed gap. They should start as lint warnings or review prompts, not autonomous task creation.

5. Stats command. Counts for active claims, failed ideas, experiments, sources, edges, and stale items would help workshop reviews. Avoid making metrics into quality proxies.

Do Not Borrow Directly

1. Do not replace kb/sources/. ARIS paper pages mix capture and interpretation. Commonplace intentionally separates immutable source capture from ingest analysis.

2. Do not create a top-level research-wiki/ peer to the KB. The workflow belongs inside kb/work/<workshop>/ unless it proves general enough to become a reference subsystem.

3. Do not copy ARIS entity names as permanent collection types. Idea, Experiment, and Claim are good workshop states, but durable commonplace artifacts already have their own type system.

4. Do not rely on prose-only lifecycle rules. ARIS's strongest implemented parts have helpers and generated receipts. If we borrow the lifecycle, it needs at least a small helper or deterministic lint.

5. Do not promote every workshop claim. The workshop needs a graveyard as much as a promotion path. Retired and invalidated claims are local memory, not necessarily library content.

Feature Priority

Priority	Feature	Why
1	Dynamic working README	Directly improves context loading for active work without making the canonical README volatile.
2	Failed idea ledger	Prevents repeated waste and captures negative learning.
3	Claim/experiment statuses	Gives lifecycle shape to claims before promotion.
4	Lifecycle lint	Makes stale or orphaned work visible.
5	Mutation log	Improves handoff and recovery with low cost.
6	Workshop edges file	Useful if working-README generation needs structured relationships.
7	Re-ideation triggers	Useful after the first manual lifecycle has evidence.

Strongest Design Constraint

The workshop map must be disposable. Its purpose is to help an active investigation complete and promote durable outputs. If it becomes a permanent shadow library, it has failed the commonplace workshop model.