Min-gated asset ranking, never weighted averagerecord 1
{
"id": "P15-I-001",
"title": "Min-gated asset ranking, never weighted average",
"problem": "Recommenders average, and an average hides one fatal gap",
"mechanism": "Asset rank is gated on the minimum evidence tier across families; a single weak family caps the asset regardless of other strengths",
"evidence_basis": "local: T0-T4 anti-averaging rule — 'A single T1 family pins the whole block at T1 regardless of eight T4s'",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 1,
"rationale": "Imports a local invariant that the entire external recommender literature contradicts, because recommenders are built to average. This is the lane's clearest first-principles divergence from prior art."
}
Rank by observed adaptation cost, not popularityrecord 2
{
"id": "P15-I-002",
"title": "Rank by observed adaptation cost, not popularity",
"problem": "Stars and downloads do not predict adaptation quality",
"mechanism": "Primary ranking signal is measured adaptation hours by category (branding, onboarding, settings, identity, navigation, data, ops)",
"evidence_basis": "local: A22 'GitHub stars predict production adaptation quality' is unproven/weak; A07 adaptation cost unknown",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 2,
"rationale": "Directly attacks the assumption the current metadata-only shelf rests on, and produces the distribution A07 needs."
}
Sparse-regime estimator selectionrecord 3
{
"id": "P15-I-003",
"title": "Sparse-regime estimator selection",
"problem": "Bandit and recommender methods assume thousands of observations; Actionist will have tens",
"mechanism": "Select estimators explicitly validated for tens of observations — Bayesian/Bradley-Terry over frequentist A/B",
"evidence_basis": "wave-1 P15 OSS census surfaced choix, PyMC, BoTorch, MABWiser as sparse-suitable",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 3,
"rationale": "The single most common way this design could fail is adopting a mechanism that needs 10k conversions per arm."
}
Cold-start priors from industry archetyperecord 4
{
"id": "P15-I-004",
"title": "Cold-start priors from industry archetype",
"problem": "New client or capability has no history",
"mechanism": "Prior initialised from the archetype and industry, updated as evidence arrives",
"evidence_basis": "local: 17 industries, 10 archetypes, 12 atoms",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 4,
"rationale": "Cold start is not an edge case here — it is the only case that currently exists (zero production data locally)."
}
Falsified-belief records as a first-class outputrecord 5
{
"id": "P15-I-005",
"title": "Falsified-belief records as a first-class output",
"problem": "Learning loops rerank items but never record which belief was wrong",
"mechanism": "Each run emits records of assumptions the outcome falsified, feeding the assumption ledger",
"evidence_basis": "local: AutoSaaS §8 lists 'failed assumptions' as an update target",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 5,
"rationale": "A genuinely local idea absent from the external census; converts the assumption ledger from a static document into a maintained artifact."
}
Evidence-class-weighted signalsrecord 6
{
"id": "P15-I-006",
"title": "Evidence-class-weighted signals",
"problem": "A vendor claim and a measured outcome get equal weight",
"mechanism": "Signals carry evidence class; only observed outcomes move rank materially",
"evidence_basis": "local: evidence discipline",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Separate capability rank from archetype rankrecord 7
{
"id": "P15-I-007",
"title": "Separate capability rank from archetype rank",
"problem": "A good capability in a bad recipe is demoted unfairly",
"mechanism": "Score capability, binding and archetype separately",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Rank is scope-boundrecord 8
{
"id": "P15-I-008",
"title": "Rank is scope-bound",
"problem": "Good-for-A read as good-everywhere",
"mechanism": "Ranks bind to industry and host scope",
"evidence_basis": "local: admitted_scope hole",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Explicit exploration budgetrecord 9
{
"id": "P15-I-009",
"title": "Explicit exploration budget",
"problem": "Exploitation freezes the shelf; the best asset is never discovered",
"mechanism": "Declared fraction of builds may use a non-top-ranked candidate under review",
"evidence_basis": "wave-1: bandit precedents",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 6,
"rationale": "Without a budget, the shelf is a ratchet — the first-ranked asset stays ranked because nothing else is ever tried."
}
Exploration only within a safety enveloperecord 10
{
"id": "P15-I-010",
"title": "Exploration only within a safety envelope",
"problem": "Exploration must not endanger a client",
"mechanism": "Exploration restricted to candidates passing all zero-tolerance gates; never on regulated workloads",
"evidence_basis": "local: zero-tolerance kills",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 7,
"rationale": "Makes exploration acceptable in a client-delivery business, which is the precondition for having any exploration at all."
}
Thompson-sampling-style selection over feasible setsrecord 11
{
"id": "P15-I-011",
"title": "Thompson-sampling-style selection over feasible sets",
"problem": "Deterministic top-1 selection never learns",
"mechanism": "Sample from the posterior over feasible candidates",
"evidence_basis": "wave-1: MABWiser/VW bandit precedents",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Decay old evidencerecord 12
{
"id": "P15-I-012",
"title": "Decay old evidence",
"problem": "Stale outcomes misrepresent current quality",
"mechanism": "Time-decay weighting with declared half-life",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Demote on incident, not on complaintrecord 13
{
"id": "P15-I-013",
"title": "Demote on incident, not on complaint",
"problem": "Anecdote drives ranking",
"mechanism": "Only structured incident records affect rank",
"evidence_basis": "cross-part with P14",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Revocation propagates to rank instantlyrecord 14
{
"id": "P15-I-014",
"title": "Revocation propagates to rank instantly",
"problem": "A revoked capability stays recommended",
"mechanism": "Revocation forces rank to floor and reopens dependent assemblies",
"evidence_basis": "local: invariant 6 revocation propagation",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Rank explanations are mandatoryrecord 15
{
"id": "P15-I-015",
"title": "Rank explanations are mandatory",
"problem": "Opaque ranking cannot be trusted or debugged",
"mechanism": "Every rank carries its contributing signals",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Human override with recorded reasonrecord 16
{
"id": "P15-I-016",
"title": "Human override with recorded reason",
"problem": "Model ranking contradicts operator knowledge",
"mechanism": "Override permitted, logged, and itself a learning signal",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Never rank on corpus sizerecord 17
{
"id": "P15-I-017",
"title": "Never rank on corpus size",
"problem": "Catalogue breadth read as quality",
"mechanism": "Explicitly excluded signal",
"evidence_basis": "local: A19 rejected; OpenConnector 1,445 providers with unsafe storage",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Reuse-shape-specific scoringrecord 18
{
"id": "P15-I-018",
"title": "Reuse-shape-specific scoring",
"problem": "An intact service and an extracted package fail differently",
"mechanism": "Separate score dimensions per packaging mode",
"evidence_basis": "local: packaging modes",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Two-sided rank: fit and burdenrecord 19
{
"id": "P15-I-019",
"title": "Two-sided rank: fit and burden",
"problem": "A capable asset with heavy maintenance is not better",
"mechanism": "Score capability fit and maintenance burden separately, both visible",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Rank floor for unproven assetsrecord 20
{
"id": "P15-I-020",
"title": "Rank floor for unproven assets",
"problem": "Unproven assets promoted by optimism",
"mechanism": "Assets without production evidence cannot exceed a declared ceiling",
"evidence_basis": "local: fail-closed default",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Accepted plan as the primary positive signalrecord 21
{
"id": "P15-I-021",
"title": "Accepted plan as the primary positive signal",
"problem": "No signal that a composition worked",
"mechanism": "Record the accepted AssemblyPlan with its capability set",
"evidence_basis": "local: composition lifecycle",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Client edit volume as a fit signalrecord 22
{
"id": "P15-I-022",
"title": "Client edit volume as a fit signal",
"problem": "Heavy editing means poor fit but is uncaptured",
"mechanism": "Edit counts by verb per capability feed rank",
"evidence_basis": "cross-part with P13 operation log",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 8,
"rationale": "Turns P13's operation log into P15's richest signal at zero extra instrumentation cost."
}
Rejected offers loggedrecord 23
{
"id": "P15-I-023",
"title": "Rejected offers logged",
"problem": "Only accepted choices recorded, biasing learning",
"mechanism": "Log declined candidates",
"evidence_basis": "cross-part with P13",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Refusal log as demand signalrecord 24
{
"id": "P15-I-024",
"title": "Refusal log as demand signal",
"problem": "Unsupported requests are wasted information",
"mechanism": "Aggregate refusals into roadmap priorities",
"evidence_basis": "cross-part with P13",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Escape-hatch rate per capabilityrecord 25
{
"id": "P15-I-025",
"title": "Escape-hatch rate per capability",
"problem": "No measure of whether bounds fit reality",
"mechanism": "Fork rate is a bounded-editing health metric",
"evidence_basis": "cross-part with P13",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Adaptation hours by categoryrecord 26
{
"id": "P15-I-026",
"title": "Adaptation hours by category",
"problem": "Adaptation cost is a single opaque number",
"mechanism": "Record hours per adaptation category",
"evidence_basis": "local: A06 categories enumerated",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Glue size per bindingrecord 27
{
"id": "P15-I-027",
"title": "Glue size per binding",
"problem": "Integration cost invisible",
"mechanism": "Measure glue LOC and complexity per port binding",
"evidence_basis": "local: predicted_glue_loc field",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Failed assumptions per buildrecord 28
{
"id": "P15-I-028",
"title": "Failed assumptions per build",
"problem": "Wrong beliefs repeat across clients",
"mechanism": "Structured record per falsified assumption",
"evidence_basis": "local: AutoSaaS",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Workflow defect count post-releaserecord 29
{
"id": "P15-I-029",
"title": "Workflow defect count post-release",
"problem": "Quality measured by build success",
"mechanism": "Count defects in the accepted workflow after release",
"evidence_basis": "local: workflow acceptance",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Incident attributed per capabilityrecord 30
{
"id": "P15-I-030",
"title": "Incident attributed per capability",
"problem": "Incidents attach to the app, not the component",
"mechanism": "Requires the stable capability identifier",
"evidence_basis": "cross-part with P14",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Maintenance events per capabilityrecord 31
{
"id": "P15-I-031",
"title": "Maintenance events per capability",
"problem": "Maintenance burden discovered late",
"mechanism": "Log upgrade and repair events per capability",
"evidence_basis": "cross-part with P14",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Upgrade-break eventsrecord 32
{
"id": "P15-I-032",
"title": "Upgrade-break events",
"problem": "Upgrade safety is asserted, not measured",
"mechanism": "Record when an upgrade broke client edits",
"evidence_basis": "verified: Shopify settings-wipe reports show this failure is real on mature platforms",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Time-to-first-working-assemblyrecord 33
{
"id": "P15-I-033",
"title": "Time-to-first-working-assembly",
"problem": "No speed baseline",
"mechanism": "Measure spec-confirmed to approved preview",
"evidence_basis": "local: client-value measures",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Reuse ratio per buildrecord 34
{
"id": "P15-I-034",
"title": "Reuse ratio per build",
"problem": "Reuse is claimed, not measured",
"mechanism": "Ports satisfied by admitted assets over total ports",
"evidence_basis": "local: reuse ratio measure",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Human-touch minutes per accepted assemblyrecord 35
{
"id": "P15-I-035",
"title": "Human-touch minutes per accepted assembly",
"problem": "Automation claimed without measurement",
"mechanism": "Reviewer minutes recorded",
"evidence_basis": "local: client-value measures",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Escalation rate by triggerrecord 36
{
"id": "P15-I-036",
"title": "Escalation rate by trigger",
"problem": "Escalations unattributed",
"mechanism": "Count by named trigger",
"evidence_basis": "local: escalation triggers",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Model tokens per accepted assemblyrecord 37
{
"id": "P15-I-037",
"title": "Model tokens per accepted assembly",
"problem": "Cost per call is the wrong denominator",
"mechanism": "Cost per accepted outcome",
"evidence_basis": "local: C_success",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
UNDERDETERMINED results as contract defectsrecord 38
{
"id": "P15-I-038",
"title": "UNDERDETERMINED results as contract defects",
"problem": "Solver failures blamed on the model",
"mechanism": "Each names a missing contract field; aggregate drives contract evolution",
"evidence_basis": "local: 'the single most valuable output for lane 01'",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 9,
"rationale": "Redirects effort from burning model budget to fixing the contract layer, which is where the defect actually is."
}
Zero-tolerance events never averaged awayrecord 39
{
"id": "P15-I-039",
"title": "Zero-tolerance events never averaged away",
"problem": "A leak becomes a statistic",
"mechanism": "Recorded separately; never contribute to an aggregate score",
"evidence_basis": "local: zero-tolerance classes",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Signal provenance recordedrecord 40
{
"id": "P15-I-040",
"title": "Signal provenance recorded",
"problem": "Cannot audit why rank changed",
"mechanism": "Every signal carries source, date and class",
"evidence_basis": "local: evidence discipline",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Negative results retainedrecord 41
{
"id": "P15-I-041",
"title": "Negative results retained",
"problem": "Only successes are recorded",
"mechanism": "Failed builds retained with reasons",
"evidence_basis": "local: preserve failing receipts",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Client outcome signal, not just technical successrecord 42
{
"id": "P15-I-042",
"title": "Client outcome signal, not just technical success",
"problem": "Technical success is not client value",
"mechanism": "Requires a client-agreed outcome rubric — currently absent",
"evidence_basis": "local: outcome fit is U",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Cross-client aggregation is opt-in and contractualrecord 43
{
"id": "P15-I-043",
"title": "Cross-client aggregation is opt-in and contractual",
"problem": "Learning across clients without consent is a breach",
"mechanism": "Aggregation requires explicit contractual permission per client",
"evidence_basis": "local: client/legal gate G7",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "top10",
"rank": 10,
"rationale": "Client-private work makes this a precondition for the whole cross-client learning premise, not a feature."
}
Aggregate only structural facts, never contentrecord 44
{
"id": "P15-I-044",
"title": "Aggregate only structural facts, never content",
"problem": "Client data leaks through learning",
"mechanism": "Learn from capability IDs, edit verbs, timings and outcomes — never client content, schema names or copy",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
k-anonymity threshold before any aggregaterecord 45
{
"id": "P15-I-045",
"title": "k-anonymity threshold before any aggregate",
"problem": "Small cohorts identify a client",
"mechanism": "Aggregates suppressed below a declared client count",
"evidence_basis": "wave-1 privacy-aggregation category",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Differential-privacy noise on published aggregatesrecord 46
{
"id": "P15-I-046",
"title": "Differential-privacy noise on published aggregates",
"problem": "Repeated queries reconstruct individuals",
"mechanism": "Calibrated noise with a declared budget",
"evidence_basis": "wave-1: OpenDP/Google DP precedents",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Per-tenant learning isolated by defaultrecord 47
{
"id": "P15-I-047",
"title": "Per-tenant learning isolated by default",
"problem": "Cross-tenant learning is the default in most systems",
"mechanism": "Tenant-local learning unless aggregation is authorised",
"evidence_basis": "local: tenancy invariants",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
No client identifiers in the learning storerecord 48
{
"id": "P15-I-048",
"title": "No client identifiers in the learning store",
"problem": "Identifiers leak through joins",
"mechanism": "Pseudonymous cohort keys only",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Retention limit on learning signalsrecord 49
{
"id": "P15-I-049",
"title": "Retention limit on learning signals",
"problem": "Signals accumulate indefinitely",
"mechanism": "Declared retention and deletion path",
"evidence_basis": "local: retention requirements",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Client deletion propagates to learningrecord 50
{
"id": "P15-I-050",
"title": "Client deletion propagates to learning",
"problem": "Deletion request cannot be honoured",
"mechanism": "Signals removable and aggregates recomputable",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Learning store is read-only to composition agentsrecord 51
{
"id": "P15-I-051",
"title": "Learning store is read-only to composition agents",
"problem": "Agents could poison ranking",
"mechanism": "Write path restricted to the deterministic pipeline",
"evidence_basis": "local: models cannot grant authority",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Poisoning resistancerecord 52
{
"id": "P15-I-052",
"title": "Poisoning resistance",
"problem": "A single bad client skews the shelf",
"mechanism": "Per-client influence caps",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
No client-identifiable data in promptsrecord 53
{
"id": "P15-I-053",
"title": "No client-identifiable data in prompts",
"problem": "Learning context leaks into models",
"mechanism": "Prompt packets carry structural facts only",
"evidence_basis": "local: retrieval packet rules",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Audit log of learning updatesrecord 54
{
"id": "P15-I-054",
"title": "Audit log of learning updates",
"problem": "Rank changes unexplainable after the fact",
"mechanism": "Immutable log of rank changes and causes",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Regulated-industry exclusionrecord 55
{
"id": "P15-I-055",
"title": "Regulated-industry exclusion",
"problem": "Health/finance/legal carry stricter obligations",
"mechanism": "Excluded from cross-client aggregation by default",
"evidence_basis": "local: contract-pressure column",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Published methodologyrecord 56
{
"id": "P15-I-056",
"title": "Published methodology",
"problem": "Opaque ranking cannot be defended to a client",
"mechanism": "Ranking methodology documented and versioned",
"evidence_basis": "wave-1: documented-methodology registries are the credible ones",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Refresh cadence without assuming the large corpusrecord 57
{
"id": "P15-I-057",
"title": "Refresh cadence without assuming the large corpus",
"problem": "The 1.3M/850k corpus has no authoritative path",
"mechanism": "Design refresh against the assets actually in hand; treat the corpus as an optional feed",
"evidence_basis": "local: A20/A21 unknown",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Refresh triggered by events, not only schedulerecord 58
{
"id": "P15-I-058",
"title": "Refresh triggered by events, not only schedule",
"problem": "Scheduled refresh misses urgent changes",
"mechanism": "Upstream release, advisory or licence change triggers refresh",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Source-bearing vs bundle-only distinction preservedrecord 59
{
"id": "P15-I-059",
"title": "Source-bearing vs bundle-only distinction preserved",
"problem": "Bundles counted as reusable supply",
"mechanism": "Refresh tracks the distinction explicitly",
"evidence_basis": "local: 3,506 source-bearing of 8,515 identities",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Deduplicate before countingrecord 60
{
"id": "P15-I-060",
"title": "Deduplicate before counting",
"problem": "Duplicate identities inflate supply",
"mechanism": "Dedupe key per asset",
"evidence_basis": "local: corpus join practice",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Deprecation on abandonmentrecord 61
{
"id": "P15-I-061",
"title": "Deprecation on abandonment",
"problem": "Abandoned assets stay recommended",
"mechanism": "Maintenance signals trigger deprecation review",
"evidence_basis": "wave-1: Daytona went unmaintained while holding 71,868 stars",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Licence re-check on refreshrecord 62
{
"id": "P15-I-062",
"title": "Licence re-check on refresh",
"problem": "Licences change between versions",
"mechanism": "Rights re-verified per refreshed revision",
"evidence_basis": "local: rights discipline",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Successor naming on deprecationrecord 63
{
"id": "P15-I-063",
"title": "Successor naming on deprecation",
"problem": "Deprecated assets leave clients stranded",
"mechanism": "Deprecation names a successor or an explicit none",
"evidence_basis": "local: supersedes/superseded_by",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
New-asset probationrecord 64
{
"id": "P15-I-064",
"title": "New-asset probation",
"problem": "New assets enter at full rank",
"mechanism": "Probation period with a rank ceiling",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Taxonomy drift detectionrecord 65
{
"id": "P15-I-065",
"title": "Taxonomy drift detection",
"problem": "Categories quietly diverge",
"mechanism": "Periodic reconciliation with a versioned taxonomy",
"evidence_basis": "local: 68-vs-144 taxonomy conflict unresolved",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Coverage gaps become research prioritiesrecord 66
{
"id": "P15-I-066",
"title": "Coverage gaps become research priorities",
"problem": "Gaps invisible until a build fails",
"mechanism": "EMPTY port results aggregate into a discovery backlog",
"evidence_basis": "local: EMPTY as coverage information",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Stop rules on breadthrecord 67
{
"id": "P15-I-067",
"title": "Stop rules on breadth",
"problem": "Breadth expands without closing unknowns",
"mechanism": "Explicit breadth/depth stop rules",
"evidence_basis": "local: decision ledger stop rules",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Refresh receiptsrecord 68
{
"id": "P15-I-068",
"title": "Refresh receipts",
"problem": "Refresh claimed without proof",
"mechanism": "Dated receipt per refresh cycle",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Corpus provenance retainedrecord 69
{
"id": "P15-I-069",
"title": "Corpus provenance retained",
"problem": "Asset origin lost over refreshes",
"mechanism": "Immutable provenance chain",
"evidence_basis": "local: provenance fields",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Manual curation lane preservedrecord 70
{
"id": "P15-I-070",
"title": "Manual curation lane preserved",
"problem": "Full automation drifts from judgement",
"mechanism": "Named human curation decisions recorded as signals",
"evidence_basis": "local: automate evidence, preserve decisions",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipes version independently of assetsrecord 71
{
"id": "P15-I-071",
"title": "Recipes version independently of assets",
"problem": "Recipe changes silently alter builds",
"mechanism": "Versioned recipes with pinned asset ranges",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipe success rate trackedrecord 72
{
"id": "P15-I-072",
"title": "Recipe success rate tracked",
"problem": "Recipes assumed good",
"mechanism": "Track accepted-build rate per recipe",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Successful compositions promote to recipesrecord 73
{
"id": "P15-I-073",
"title": "Successful compositions promote to recipes",
"problem": "Good combinations are rediscovered each time",
"mechanism": "Repeated successful sets become candidate recipes",
"evidence_basis": "local: template registry improvement loop",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recurring edits promote to defaultsrecord 74
{
"id": "P15-I-074",
"title": "Recurring edits promote to defaults",
"problem": "Every client makes the same edit",
"mechanism": "Recurring edits become capability defaults or variants",
"evidence_basis": "cross-part with P13",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Archetype fit measured by workflow comprehensionrecord 75
{
"id": "P15-I-075",
"title": "Archetype fit measured by workflow comprehension",
"problem": "Shell assumed universal",
"mechanism": "Test archetypes against task success",
"evidence_basis": "local: A15 unproven",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Industry deltas recorded separately from archetypesrecord 76
{
"id": "P15-I-076",
"title": "Industry deltas recorded separately from archetypes",
"problem": "Industry specifics pollute shared recipes",
"mechanism": "Delta layer per industry",
"evidence_basis": "local: A02/A03",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipe deprecation on repeated failurerecord 77
{
"id": "P15-I-077",
"title": "Recipe deprecation on repeated failure",
"problem": "Bad recipes persist",
"mechanism": "Failure threshold triggers review",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Contract evolution driven by UNDERDETERMINED aggregatesrecord 78
{
"id": "P15-I-078",
"title": "Contract evolution driven by UNDERDETERMINED aggregates",
"problem": "Contracts evolve by opinion",
"mechanism": "Missing-field counts prioritise contract work",
"evidence_basis": "local: solver output",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Binding patterns learned per port pairrecord 79
{
"id": "P15-I-079",
"title": "Binding patterns learned per port pair",
"problem": "Same glue rewritten repeatedly",
"mechanism": "Successful bindings become reusable patterns",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Anti-pattern registryrecord 80
{
"id": "P15-I-080",
"title": "Anti-pattern registry",
"problem": "Known-bad combinations retried",
"mechanism": "Record combinations that repeatedly failed",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipe cost modelrecord 81
{
"id": "P15-I-081",
"title": "Recipe cost model",
"problem": "Recipe cost unknown until built",
"mechanism": "Predict cost from historical builds",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipes carry their own falsifiersrecord 82
{
"id": "P15-I-082",
"title": "Recipes carry their own falsifiers",
"problem": "Recipes become dogma",
"mechanism": "Each recipe states what would retire it",
"evidence_basis": "local: falsifier discipline",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Cross-industry pattern detectionrecord 83
{
"id": "P15-I-083",
"title": "Cross-industry pattern detection",
"problem": "Shared patterns missed",
"mechanism": "Detect atoms recurring across industries",
"evidence_basis": "local: 12 atoms",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
Recipe provenance to first successful buildrecord 84
{
"id": "P15-I-084",
"title": "Recipe provenance to first successful build",
"problem": "Recipe origin unknown",
"mechanism": "Link recipe to the build that proved it",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: ranking on stars/downloadsrecord 85
{
"id": "P15-I-085",
"title": "REJECTED: ranking on stars/downloads",
"problem": "Does not predict adaptation quality",
"mechanism": "Observed adaptation cost and outcomes",
"evidence_basis": "local: A22",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: corpus size as supply proofrecord 86
{
"id": "P15-I-086",
"title": "REJECTED: corpus size as supply proof",
"problem": "Discovery is not qualified supply",
"mechanism": "Qualification funnel metrics",
"evidence_basis": "local: A19 rejected",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: weighted-average asset scoresrecord 87
{
"id": "P15-I-087",
"title": "REJECTED: weighted-average asset scores",
"problem": "Hides fatal gaps",
"mechanism": "Min-gated ranking",
"evidence_basis": "local: anti-averaging",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: cross-client learning by defaultrecord 88
{
"id": "P15-I-088",
"title": "REJECTED: cross-client learning by default",
"problem": "Client-private work forbids it",
"mechanism": "Opt-in contractual aggregation",
"evidence_basis": "local: G7",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: learning from client contentrecord 89
{
"id": "P15-I-089",
"title": "REJECTED: learning from client content",
"problem": "Leaks private data",
"mechanism": "Structural facts only",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: frequentist A/B on tens of buildsrecord 90
{
"id": "P15-I-090",
"title": "REJECTED: frequentist A/B on tens of builds",
"problem": "Underpowered; will mislead",
"mechanism": "Bayesian/sparse estimators",
"evidence_basis": "wave-1 sparse-fit analysis",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: assuming the 1.3M corpus feed existsrecord 91
{
"id": "P15-I-091",
"title": "REJECTED: assuming the 1.3M corpus feed exists",
"problem": "No authoritative path in current receipts",
"mechanism": "Design without it",
"evidence_basis": "local: A20/A21",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: automatic promotion on green metricsrecord 92
{
"id": "P15-I-092",
"title": "REJECTED: automatic promotion on green metrics",
"problem": "Metrics are not admission",
"mechanism": "Human admission decision retained",
"evidence_basis": "local: admission rule",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: model-authored ranking changesrecord 93
{
"id": "P15-I-093",
"title": "REJECTED: model-authored ranking changes",
"problem": "Models cannot grant authority",
"mechanism": "Deterministic pipeline owns writes",
"evidence_basis": "local: invariant",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: treating vendor claims as loop evidencerecord 94
{
"id": "P15-I-094",
"title": "REJECTED: treating vendor claims as loop evidence",
"problem": "Marketing is not methodology",
"mechanism": "documented_methodology class required",
"evidence_basis": "wave-1 classification",
"evidence_class": "observed_behavior",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: single global rankrecord 95
{
"id": "P15-I-095",
"title": "REJECTED: single global rank",
"problem": "Ignores scope and industry",
"mechanism": "Scope-bound ranks",
"evidence_basis": "local: admitted_scope",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: silent truncation of candidate setsrecord 96
{
"id": "P15-I-096",
"title": "REJECTED: silent truncation of candidate sets",
"problem": "Reads as full coverage",
"mechanism": "Record candidates_dropped",
"evidence_basis": "local: no-silent-caps rule",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: retiring assets on age alonerecord 97
{
"id": "P15-I-097",
"title": "REJECTED: retiring assets on age alone",
"problem": "Age is not quality",
"mechanism": "Maintenance and outcome signals",
"evidence_basis": "inference",
"evidence_class": "inference",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: one learning model across all reuse shapesrecord 98
{
"id": "P15-I-098",
"title": "REJECTED: one learning model across all reuse shapes",
"problem": "Shapes fail differently",
"mechanism": "Shape-specific scoring",
"evidence_basis": "local: packaging modes",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: contextual-bandit SaaS dependencyrecord 99
{
"id": "P15-I-099",
"title": "REJECTED: contextual-bandit SaaS dependency",
"problem": "A flagship service in this category was retired",
"mechanism": "Own the mechanism; treat vendors as replaceable",
"evidence_basis": "wave-1: Azure Personalizer retirement — verify status",
"evidence_class": "secondary",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}
REJECTED: claiming a learning loop exists todayrecord 100
{
"id": "P15-I-100",
"title": "REJECTED: claiming a learning loop exists today",
"problem": "Zero production data exists",
"mechanism": "Design for cold start; state absence plainly",
"evidence_basis": "local: zero accepted plans/incidents/outcomes",
"evidence_class": "first_party_docs",
"observed_date": "2026-08-27",
"limitations": "design hypothesis; unimplemented and unmeasured; no production learning data exists locally",
"disposition": "register",
"rank": null,
"rationale": null
}