Pilot capability-boundary decisions

ID	Candidate	Individual MLE result	Decision	Reason
M01	Retain governed content	passes	separate CRD	Retention has its own actor, command, durable result, and failure independent of processing or use.
M02	Derive content from a retained source	passes	separate CRD	Transformation has independent authority, provenance, state, and failure semantics.
M03	Allocate retained content to a context	passes	separate CRD	It changes authorization eligibility without performing retrieval or processing.
M04	Assemble governed retrieval context	passes	separate CRD	It produces a bounded context package before and independently of generation.
M05	Consolidate conversation history	passes	separate CRD	It creates a reusable derivative with its own lifecycle and source links.
M06	Decide a governed proposal	passes	separate CRD	Proposal disposition is independently meaningful and auditable.
M07	Establish a standing rule	passes	separate CRD	It creates reusable authority and can succeed without a particular later match.
M08	Apply a standing rule	passes	separate CRD	It evaluates one governed situation and produces a logged action, fallback, or no-action result.
M09	Status/decision notification	fails as Capability MLE	Communication MLE owned by the triggering capability	The message is an effect and loses purpose without the underlying outcome.

The symmetric MLE test requires establish-standing-rule and apply-standing-rule to remain separate capabilities because each passes independently. Their relationship is expressed through authority provenance and shared elements rather than by combining them into a rule-management feature bucket.

Phase 2 knowledge-and-context tranche A1

ID	Candidate	Individual MLE result	Decision	Reason
K-M01	Maintain a personal glossary term	passes	separate CRD	An owner can establish or revise one stable local meaning with a determinable lifecycle outcome.
K-M02	Manage an organizing label	passes	separate CRD	A label organizes/filter items without carrying the referential meaning of a glossary term.
K-M03	Detect a candidate glossary term	passes	separate CRD	Detection produces a reviewable candidate and evidence without accepting it as truth.
K-M04	Manage AI-assisted label vocabulary	passes	separate CRD	The vocabulary has an independent lifecycle and constrains later assignments.
K-M05	Apply an AI-assisted label	passes	separate CRD	One assignment has a target, confidence/provenance, and reversible active state.
K-M06	Apply contextual classification	passes	separate CRD	Attaching a Role or Mode to an object has independent contextual and historical meaning.
K-M07	Glossary application	fails as one capability	split into K-M01 through K-M03	The app/view grouping combines independently meaningful abilities.
K-M08	AI-assisted labeling	fails as one capability	split into K-M04 and K-M05	Vocabulary governance and individual assignment can succeed or fail independently.

Phase 2 knowledge-and-context tranche A2

ID	Candidate	Individual MLE result	Decision	Reason
K-M09	Maintain a deliberate knowledge note	passes	separate CRD	Durable owner-chosen knowledge capture has its own lifecycle and outcome.
K-M10	Maintain owner-authored profile knowledge	passes	separate CRD	The owner can revise self-description independently of observed/evidence-backed assertions.
K-M11	Maintain an evidence-backed profile assertion	passes	separate CRD	A sourced assertion has independent provenance, confidence, validity, and lifecycle.
K-M12	Resolve a disputed profile assertion	passes	separate CRD	A dispute has a distinct actor, decision, versioned result, and failure state.
K-M13	Profile app	fails	source context and audience projection	It groups independently meaningful knowledge, goal, review, and navigation capabilities.
K-M14	Attach a capability row to a note	fails as currently evidenced	realization pattern	The source describes a database/hook composition pattern rather than one stable user outcome.

Phase 2 actions-and-execution tranche B1

ID	Candidate	Individual MLE result	Decision	Reason
B-M01	Apply an action-intent label	passes	separate CRD	It records intended handling and succeeds without executing that handling.
B-M02	Invoke a governed action	passes	separate CRD	It mediates one declared executable operation under authority and audit.
B-M03	Run a background operation	passes	separate CRD	Asynchronous handoff, progress, cancellation/failure, and terminal result form an independent outcome.
B-M04	Configure execution policy	passes	separate CRD	Changing allowable actions and confirmation posture is independently meaningful from performing an action.
B-M05	Action Tag + Action	both pass	keep separate	Combining intent with execution would make a label silently executable.
B-M06	Skills and Actions screen	fails	audience projection	A visibility/configuration surface groups multiple capabilities and artifacts.
B-M07	Execution Model	fails	source context	It is a domain model containing multiple independently meaningful capabilities and object lifecycles.

Phase 2 actions-and-execution tranche B2a

ID	Candidate	Individual MLE result	Decision	Reason
B-M08	Maintain an execution plan	passes	separate CRD	A plan has an independent proposal, review, revision, and retained source-design outcome.
B-M09	Activate an execution plan	passes	separate CRD	Activation creates a derived execution object and provenance receipt without performing that object's work.
B-M10	Manage a one-off project	passes	separate CRD	A bounded effort has independent context, progress, responsibility, resolution, and failure semantics.
B-M11	Maintain a reusable routine	passes	separate CRD	The reusable definition has a lifecycle independent from any concrete occurrence.
B-M12	Run a routine instance	passes	separate CRD	One occurrence has its own actor, input snapshot, execution state, outputs, and terminal result.
B-M13	Manage an actionable task	passes	separate CRD	One action can be captured, assigned, advanced, completed, or reopened independently of a project or list.
B-M14	Organize a task collection	passes	separate CRD	Ordered grouping has a meaningful result while member tasks retain independent identity and state.
B-M15	Schedule a contextual reminder	passes	separate CRD	A trigger, recurrence, snooze, and enabled lifecycle can succeed or fail without changing the reminded object's state.
B-M16	Plan + activation	both pass	keep separate	Combining them would blur proposed design with the governed transition into execution.
B-M17	Project + task	both pass	keep separate	A project can exist before all tasks are known, and tasks can exist without a project.
B-M18	Routine + routine run	both pass	keep separate	Definition changes and execution outcomes have different actors, lifecycles, and histories.
B-M19	Task + task collection	both pass	keep separate	Membership and ordering do not replace the task's actionable state.
B-M20	Task + reminder	both pass	keep separate	A reminder resurfaces attention; it does not complete or otherwise execute the task.
B-M21	Planning, Projects, Routines, To-dos, or Results app	fails	audience projection	Each view groups several independently meaningful objects, transitions, and operations.

Phase 2 actions-and-execution tranche B2b

ID	Candidate	Individual MLE result	Decision	Reason
B-M22	Prepare a context-rich Work Starter	passes	separate CRD	It produces a ready-to-start contextual entry point without selecting or executing the work.
B-M23	Order ready work for the current context	passes	separate CRD	It produces an explainable time-bounded ordering and can be rerun as context changes.
B-M24	Maintain a versioned workflow package	passes	separate CRD	A distributable process contract has independent authorship, version, safety, and deprecation outcomes.
B-M25	Manage an installed workflow	passes	separate CRD	Owner-scoped installation, configuration, enablement, and version policy exist independently of package authorship and runs.
B-M26	Run a governed workflow	passes	separate CRD	One triggered execution has its own snapshots, steps, authority, terminal state, and evidence.
B-M27	Preserve an execution result	passes	separate CRD	Registering produced value with execution provenance is meaningful independently of doing the work.
B-M28	Work Starter + contextual ordering	both pass	keep separate	Preparation makes work ready; ordering selects among already eligible work for a particular context.
B-M29	Workflow package + installed workflow	both pass	keep separate	A distributable definition and an owner's configured instance have different actors, authority, and lifecycle.
B-M30	Installed workflow + workflow run	both pass	keep separate	Configuration may exist without a run, and each run has independent input, state, output, and failure semantics.
B-M31	Workflow run + result	both pass	keep separate	Execution may fail or yield several outputs; a retained result is a separately qualified provenance-bearing outcome.
B-M32	Result object + Results view	only result object passes	CRD for result preservation; view is projection	The view combines discovery, installation, runs, outputs, and review rather than one outcome.
B-M33	Routine + workflow package	both pass with relationship unresolved	keep separate	A Routine is owner-facing reusable execution structure; a Workflow Package is a distributable declared process contract.

Phase 2 events-and-History tranche C1

ID	Candidate	Individual MLE result	Decision	Reason
C-M01	Record a canonical event	passes	separate CRD	One occurrence can be durably and idempotently appended independently of any owner-facing projection.
C-M02	Manage an owner attention item	passes	separate CRD	Attention state, response, pin/read/dismiss lifecycle, and source linkage are independently meaningful.
C-M03	Publish a completed-work record	passes	separate CRD	Meaningful completion creates a durable evidence-linked record beyond generic attention management; the full event/update/detail/manifest projection set remains a recommended default.
C-M04	Compile daily owner History	passes	separate CRD	Daily compilation uniquely selects and narrates structured and fallback evidence for an owner-defined day.
C-M05	Aggregate an owner History period	passes	separate CRD	Higher-level composition has different inputs, abstraction, closure, and gap semantics from daily compilation.
C-M06	Navigate time-indexed owner History	passes	separate CRD	Traversal and retrieval produce relevant historical context without creating or changing source records.
C-M07	Event + Update	both pass	keep separate	The event is canonical occurrence truth; the Update is a resolvable attention projection.
C-M08	Generic Update + completed-work publication	both pass	keep separate	Generic attention can surface decisions or warnings, while completed-work publication requires a provenance-linked evidence bundle.
C-M09	Daily + higher-period summary	both pass	keep separate	Daily may inspect source evidence; higher levels compose from closed lower-level summaries.
C-M10	Unified Timeline / History view	fails as one capability	source context and audience projection	It combines navigation, filtering, summaries, updates, events, and source drill-down.
C-M11	Notification delivery	fails as standalone Capability MLE	Communication MLE owned by the attention source	Delivery loses purpose without the update, reminder, incident, or decision it communicates.

Phase 2 operations-and-portability tranche C2a

ID	Candidate	Individual MLE result	Decision	Reason
C-M12	Register a connected source	passes	separate CRD	Source identity, access, consent, sync, event mapping, and change behavior form an independently governed lifecycle.
C-M13	Manage an import session	passes	separate CRD	One bounded intake event has its own manifest, progress, items, errors, and outcome independently from the source.
C-M14	Govern source promotion	passes	separate CRD	Moving material among parked, analyzed, mapped, enriched, and History-included states changes eligibility and authority without performing every processing step.
C-M15	Screen an ingestion batch	passes	separate CRD	A technical reviewer can produce an evidence-based pass/fail result without authorizing upload.
C-M16	Assess ingestion readiness	passes	separate CRD	Checklist closure produces an independent readiness result while preserving unresolved owner decisions.
C-M17	Authorize an ingestion batch	passes	separate CRD	The owner can grant or deny authority for an exact manifest and boundary without performing the ingestion.
C-M18	Govern an event-type registry	passes	separate CRD	Adding or revising an event type changes allowed semantics, emitters, schemas, and projection obligations independently from recording any event.
C-M19	Inspect an operational run record	passes	separate CRD	An owner or operator can retrieve one run's state, evidence, outputs, errors, and authority trail without changing canonical state.
C-M20	Source + import session	both pass	keep separate	A source persists across zero or many bounded intake sessions.
C-M21	Import session + source promotion	both pass	keep separate	Intake scope and progress can complete while later mapping, enrichment, or History inclusion remains pending.
C-M22	Technical screen + checklist + owner approval	all pass	keep separate	Each gate has a distinct actor, evidence standard, result, and expressly limited authority.
C-M23	Event registry + event recording	both pass	keep separate	Registry governance defines allowed event semantics; recording appends one conforming occurrence.
C-M24	System app	fails	audience projection	It groups independent inspection, governance, run, import, export, and health capabilities.

Phase 2 operations-and-portability tranche C2b1

ID	Candidate	Individual MLE result	Decision	Reason
C-M25	Compose a portable Core bundle	passes	separate CRD	Bundle creation has its own scope, identity, manifest, contents, exclusions, and completion result.
C-M25A	Compose a source-derived PIOS portability package	passes	separate CRD	An owner or migration actor can assemble non-Core source material into a validation-ready package without claiming that it is existing Core state.
C-M26	Validate a portability package	passes	separate CRD	Validation can accept, reject, or quarantine an existing package without creating or importing it.
C-M27	Restore Core state from a bundle	passes	separate CRD	Import/hydration changes destination state and produces an operationalization report independently from validation.
C-M28	Validate restored Core parity	passes	separate CRD	Destination evidence can pass or fail after restore and is not established by successful byte import alone.
C-M29	Maintain a recoverable Core backup	passes	separate CRD	Backup creation, custody, retention, rotation, and integrity status form a recurring protection outcome independent from owner portability exports and restore execution.
C-M30	Bundle composition + validation	both pass	keep separate	A package can be created but invalid, or an externally supplied package can be validated without being created locally.
C-M31	Package validation + restore	both pass	keep separate	Technical validity does not authorize or perform destination state changes.
C-M32	Restore + parity validation	both pass	keep separate	Restored bytes may exist while required protections, provenance links, retrieval, or service behavior remain unproven.
C-M33	Export bundle + backup	both pass	keep separate	Portability export serves owner transfer/takeover; backup serves protected recovery and may use different retention and custody.
C-M33A	Core-export composition + source-composed package creation	both pass	keep separate	One serializes existing Core state; the other prepares owner-controlled source material whose canonical Core status does not yet exist.
C-M33B	Backup maintenance + restore/parity testing	all pass	keep separate	A backup can be created and integrity-checked without restoring it; restore changes destination state, and parity separately evaluates the restored result.

Phase 2 operations-and-portability tranche C2b2

ID	Candidate	Individual MLE result	Decision	Reason
C-M34	Govern a Core cutover	passes	separate CRD	Canonical-side transfer has its own bounded scope, authority, final-delta, and completion decision.
C-M35	Roll back a Core cutover	passes	separate CRD	Reverting an attempted transition has an independent actor, trigger, state restoration, and evidence outcome.
C-M36	Decommission a source Core	passes	separate CRD	Retirement occurs after successful transition and has distinct retention, deletion, revocation, and closure evidence.
C-M37	Register an owner-data erasure request	passes	separate CRD	The request immediately records intent and may suppress exposure even when physical deletion is delayed.
C-M38	Execute governed data erasure	passes	separate CRD	Physical or cryptographic erasure and derived cleanup have independent authority, timing, proof, and failure semantics.
C-M39	Reauthorize a destination connector	passes	separate CRD	Re-establishing source credentials and future-sync authority is meaningful independently from restoring connector descriptions.
C-M40	Cutover + rollback	both pass	keep separate	One establishes destination canonical authority; the other restores or preserves the prior canonical path after a failed or reversed transition.
C-M41	Cutover + decommissioning	both pass	keep separate	Cutover can complete while source retention and retirement remain intentionally deferred.
C-M42	Erasure request + erasure execution	both pass	keep separate	Request and exposure suppression may succeed while retention prevents immediate physical deletion.
C-M43	Restore + connector reauthorization	both pass	keep separate	Portable descriptions may be restored without importing credentials, device trust, or future-sync authority.

Phase 2 capture-and-experience tranche D1a

ID	Candidate	Individual MLE result	Decision	Reason
D-M01	Submit a governed content capture	passes	separate CRD	Submission has an independent actor, payload, source context, acceptance request, and receipt.
D-M02	Reconcile a pending capture	passes	separate CRD	Pending work has a separate lifecycle and can end accepted, retryable, failed, preserved, or authorized-discarded.
D-M03	Web clipping extension	fails as generic capability	realization and audience projection	Browser scraping, authentication, tag UI, and API transport implement capture and adjacent capabilities.
D-M04	Capture + action-intent labeling	both pass	keep separate	Content can be captured without a handling label, and a label can be applied after capture.
D-M05	Capture + processing instruction	both pass	keep separate	Intake does not itself authorize or perform the requested processing action.
D-M06	Submit + reconcile pending capture	both pass	keep separate	Submission may complete before durable acceptance, and reconciliation may resume after device or network interruption.

Phase 2 capture-and-experience tranche D1b1

ID	Candidate	Individual MLE result	Decision	Reason
D-M07	Manage a guided onboarding journey	passes	separate CRD	The journey has durable stage, progress, pause/resume, response, override, and terminal outcomes independent from any one setup result.
D-M08	Maintain a personal AI identity	passes	separate CRD	Owner-authored presentation identity has its own versioned lifecycle and remains distinct from owner profile, memory, runtime role, and authority.
D-M09	Configure personal AI interaction preferences	passes	separate CRD	Communication and collaboration preferences can change without changing identity, truth, or execution permission.
D-M10	Verify an onboarding contact channel	passes through a reusable capability	reuse `verify-contact-point-control` with an onboarding profile	Channel-control evidence has an independent assurance result, but binding it to onboarding does not change the generic contact-point verification outcome.
D-M11	Prepare an initial owner environment	passes	separate CRD	Mapping owner input and explicit defaults into a reviewable first configuration has an independent readiness outcome.
D-M12	Authorize first entry	passes	separate CRD	A bounded invitation credential has its own issue, expiry, replacement, validation, and consumption lifecycle.
D-M13	Adaptive onboarding depth	fails independently	owned policy of guided onboarding	It changes which questions and defaults the journey uses but does not produce a separate durable owner outcome.
D-M14	First-use orientation carousel	fails as generic capability	Communication MLE and realization pattern	A skippable/replayable explanatory surface communicates product concepts but does not establish onboarding identity, readiness, or authority.
D-M15	Preserve onboarding dialogue as first conversation	fails independently	composition	Retention, conversation continuity, and first-entry presentation already own the durable outcomes.
D-M16	Download onboarding data	fails independently in this tranche	specialization of portability	Onboarding scope specializes portable bundle composition rather than defining a second export capability.
D-M17	Personal AI identity + interaction preferences	both pass	keep separate	Presentation identity can remain stable while communication behavior changes, and either may be revised independently.
D-M18	Journey + environment preparation + first-entry authorization	all pass	keep separate	Progress coordination, configuration readiness, and access authorization have different actors, evidence, failure modes, and outcomes.

Phase 2 capture-and-experience tranche D1b2

ID	Candidate	Individual MLE result	Decision	Reason
D-M19	Maintain an owner context Role	passes	separate CRD	Role definition, hierarchy, provenance, and lifecycle produce a durable owner-context outcome independently from activation or object tagging.
D-M20	Maintain an owner operating Mode	passes	separate CRD	A reusable Mode definition can be created, revised, archived, or restored independently from whether it is currently active.
D-M21	Set active personal context	passes	separate CRD	A temporary multi-dimensional context snapshot has its own source, timing, supersession, and inheritance outcome.
D-M22	Configure a Role's outward intent	passes	separate CRD	An owner can declare or withdraw a Role's representation intent without publishing content or granting access.
D-M23	Prepare a Role-scoped outward representation	passes	separate CRD	Creating a separate traceable outward object is meaningful before access or publication and protects the private source from implicit exposure.
D-M24	Role definition + Mode definition	both pass	keep separate	Roles describe owner capacities and may form type/instance hierarchies; Modes describe situational operating context and are normally more transient.
D-M25	Active context + contextual classification	both pass	keep separate	Active context is a temporary snapshot; contextual classification records a durable relationship to a particular object.
D-M26	Role outward intent + outward representation	both pass	keep separate	Intent can exist with no prepared content, and a prepared representation remains unpublished and inaccessible until later governance permits use.
D-M27	Outward representation + publication/access	passes only for preparation here	keep later exposure capabilities separate	Preparation does not establish visibility, discovery, access control, rights, delivery, or publication.
D-M28	Role management screen or context picker	fails as generic capability	audience projection and realization pattern	Controls compose definition maintenance, active-context selection, and contextual classification.
D-M29	Home, app navigation, layout, gestures, badges, and onboarding slides	fail as generic capabilities	audience/communication/realization patterns	These surfaces present or invoke several capabilities but do not produce independent domain outcomes.

Phase 2 patterns-and-learning tranche E1

ID	Candidate	Individual MLE result	Decision	Reason
E-M01	Detect an observed pattern	passes	separate CRD	Multi-evidence recurrence analysis can create or decline one emerging non-authoritative pattern with a determinate outcome.
E-M02	Evaluate an observed pattern lifecycle	passes	separate CRD	A recurring evaluator can independently update status, strength, eligibility, and snapshots for an existing pattern.
E-M03	Record one pattern-evidence row	fails as capability	shared data/information MLE	The record has purpose only inside detection/evaluation and does not deliver an independently meaningful owner outcome.
E-M04	Maintain pattern-segment vocabulary	passes in generic form already represented	reuse `manage-organizing-label`	Pattern segments specialize owner-governed organizational vocabulary rather than requiring a second generic vocabulary capability.
E-M05	Generate and decide a proposal from a stable pattern	passes in generic form already represented	reuse `decide-governed-proposal`	Pattern eligibility supplies evidence and rationale; proposal lifecycle and decision remain governed by the existing capability.
E-M06	Detect + lifecycle-evaluate a pattern	both pass	keep separate	Initial creation and later recurring reassessment have different starting states, timing, and results.
E-M07	Patterns inspector	fails as generic capability	audience projection	It presents pattern, evidence, status, and proposal history without owning their source lifecycles.
E-M08	Pattern refresh action declaration	fails as realization proof	declared/disabled realization metadata	A declared disabled action is not evidence of a reachable recurring job or available capability.

Phase 2 direction-and-success tranche E2

ID	Candidate	Individual MLE result	Decision	Reason
F-M01	Maintain an outcome goal	passes	separate CRD	A durable desired outcome has its own owner, lifecycle, horizon, relationships, and review context independently from current execution.
F-M02	Maintain a directional objective	passes	separate CRD	Current actionable direction can be created, activated, retired, and used to source planning without becoming the plan itself.
F-M03	Maintain a success target	passes	separate CRD	An observable success condition has an independent lifecycle and can exist without a measurement observation or achievement decision.
F-M04	Maintain a measurement definition	passes	separate CRD	Unit, value semantics, tracking method, and comparability form a reusable definition independently from recorded values.
F-M05	Record a metric observation	passes	separate CRD	One time-scoped value has its own actor, validation, source, provenance, correction, and acceptance result.
F-M06	Goal + objective	both pass	keep separate	Long-horizon desired outcome and current action-oriented direction can change independently and may be linked without being merged.
F-M07	Target + measurement definition	both pass	keep separate	Success can be described before deciding how it will be measured, and a measurement can serve monitoring without being a target.
F-M08	Measurement definition + metric observation	both pass	keep separate	A stable definition may receive many observations, while definition changes require independent comparability handling.
F-M09	Target + target-achievement evaluation	target passes; evaluation deferred	keep separate	Defining success and deciding whether evidence satisfies it have different inputs, authority, and outcomes; the latter contract is not yet complete.
F-M10	Goals/Profile/Planning screens	fail as generic capabilities	audience projections	Multiple screens can present the same canonical direction and success objects without becoming independent realizations.
F-M11	Goal → objective → target → plan chain	fails as mandatory combined capability	conceptual composition	The sequence helps explain flow, but sources explicitly do not enforce every link and each constituent MLE remains independently meaningful.

Phase 2 outward-exposure tranche E3

ID	Candidate	Individual MLE result	Decision	Reason
G-M01	Share a governed outward representation	passes	separate CRD	Delivering or activating a representation for a bounded audience has its own share reference and lifecycle without creating access policy or making content public.
G-M02	Publish a governed outward representation	passes	separate CRD	Public reachability is an externally visible transition with revision, locator, authority, withdrawal, and evidence independent from preparation or sharing.
G-M03	Maintain outward access policy	passes	separate CRD	Access conditions and audience grants can be created, revised, suspended, expired, revoked, or superseded without evaluating a particular request.
G-M03A	Evaluate an outward access request	passes	separate CRD	One requester can receive an attributable allow or deny decision under an unchanged policy.
G-M04	Govern outward discoverability	passes	separate CRD	Hidden, index-eligible, and promoted states produce an independent findability outcome without changing who may open or reuse content.
G-M05	Govern outward usage rights	passes	separate CRD	Human and machine use conditions remain meaningful after access and can change without altering reachability or discovery.
G-M06	Controlled sharing + public publication	both pass	keep separate	Sharing is audience-bounded and non-public; publication establishes public reachability and may carry durable public-presence framing.
G-M07	Visibility + access policy + access evaluation + discovery + rights	all meaningful; do not merge	keep separate concerns	Current and historical sources support independent state or outcomes for each concern.
G-M08	Prepare + share/publish	both pass	keep separate	A representation can be safely prepared and reviewed without any audience gaining access; exposure is a later authorized transition.
G-M09	Unpublish or revoke	fails as additional generic capability here	lifecycle transition	Withdrawal is an essential terminal transition of the corresponding publication or sharing capability, while cross-system erasure remains separate.
G-M10	Public username, path, folder, domain, sitemap and SEO/GEO	fail as one generic capability	realization patterns and later candidates	They implement public presence, addressing, organization, discovery, and delivery but do not form one minimal reusable outcome.
G-M11	Access-policy maintenance + access-request evaluation	both pass	keep separate	Policy changes are durable governance transitions; request evaluation applies the current policy to one request without changing it.
G-M12	Controlled sharing + access policy + request evaluation	all pass	keep separate	A share can be activated under an existing policy, a policy can exist before any share request, and each access request can be decided independently.

Phase 2 reviews-and-reflection tranche E4

ID	Candidate	Individual MLE result	Decision	Reason
H-M01	Conduct a structured review	passes	separate CRD	A bounded review has its own scope, questions, inputs, decisions, unresolved items, carry-forward disposition, and closure state.
H-M02	Record a reflective assessment	passes	separate CRD	Interpretation with judgment, uncertainty, residue, and recommendations can be preserved independently from a formal review or neutral summary.
H-M03	Structured review + reflection	both pass	keep separate	A review may be primarily factual or decisional, and a reflection may occur outside a scheduled review; either can exist without the other.
H-M04	Daily, weekly, monthly, yearly, and event-triggered review	passes as one generic capability	use profiles, not separate CRDs	Cadence and question sets vary while the review interaction contract remains the same.
H-M05	Review note or review screen	fails as generic capability	output/audience projection	A note or screen can present a review but does not own its evidence, decisions, and follow-up lifecycle.
H-M06	Carry forward unresolved work	passes through existing capabilities	reuse composition	A review references existing tasks, Work Starters, attention items, goals, or projects rather than creating a second work identity.
H-M07	Apply review recommendations	passes through existing governed capabilities	reuse composition	Recommendations may trigger proposals or updates to goals, targets, projects, tasks, knowledge, or policy only through their own authority paths.

Phase 2 people-and-relationships tranche E5

ID	Candidate	Individual MLE result	Decision	Reason
I-M01	Maintain a human contact profile	passes	separate CRD	Another person's owner-held profile has a distinct subject, authority, provenance, lifecycle, and knowledge outcome.
I-M02	Maintain a contact point	passes	separate CRD	A typed coordinate can be created, corrected, prioritized, invalidated, or removed independently from the wider profile and relationship.
I-M03	Verify control of a contact point	passes	separate reusable CRD	Verification has its own actor, challenge/evidence state, assurance, expiry, and result and is reused by contact and onboarding contexts.
I-M04	Maintain personal relationship context	passes	separate CRD	Relationship meaning, scope, trust, consent, origin, validity, and history change independently from participant identity and contact coordinates.
I-M05	Person identity + contact profile	both pass	keep separate	A named person can exist in the glossary/world model without a maintained relationship profile, and profile knowledge can reference the identity without replacing it.
I-M06	Contact profile + contact point	both pass	keep separate	One profile may have zero or many changing contact points; coordinates can be invalidated without retiring the person profile.
I-M07	Contact point + control verification	both pass	keep separate	A value can be stored while unverified, and verification can expire or be revoked without deleting the value.
I-M08	Relationship origin	fails independently at current evidence depth	shared information/provenance MLE	Origin date, place, introduction and story support relationship context but no separate recurring interaction contract is established.
I-M09	With-whom event context + durable relationship context	both pass through existing capabilities	keep separate	Event participation is time-scoped contextual classification; durable relationship meaning has its own lifecycle and authority.
I-M10	Contacts application, filters, avatar and list	fail as generic capabilities	audience projections and realization patterns	They present or edit contact, contact-point, Role and label capabilities without owning independent domain outcomes.

Phase 2 agents-and-outward-identity tranche E6

ID	Candidate	Individual MLE result	Decision	Reason
J-M01	Maintain an agent definition	passes	separate CRD	A persistent agent's governed purpose, responsibilities, skills, limits, permissions, memory/source scopes, instructions, state, and versions form an independent durable outcome.
J-M02	Synchronize an agent definition to runtime	passes	separate CRD	Applying and verifying one canonical version across runtime targets has independent source, target, transformation, failure, intentional-difference, and rollback states.
J-M03	Assign an agent runtime role	passes	separate CRD	A bounded functional role has its own assigner, scope, duration, responsibility and terminal state without changing persistent identity.
J-M04	Select an agent's outward identity	passes	separate CRD	Identity presentation and disclosure for an external action are independently governed before the action executes.
J-M05	Conduct an owner-representative conversation	passes	separate CRD	One external conversational interaction has a distinct participant, identity, knowledge scope, policy, response/refusal, escalation, and evidence outcome.
J-M06	Agent identity + runtime role	both pass	keep separate	One identity may occupy several roles over time, and the same role can be assigned to different agents without redefining either.
J-M07	Agent definition + runtime synchronization	both pass	keep separate	A canonical definition can change without being deployed, and synchronization can fail or intentionally diverge without changing source truth.
J-M08	Outward identity + external action	both pass	keep separate	Selecting who appears to act and what is disclosed does not itself authorize or deliver the external effect.
J-M09	Personal-AI identity + outward identity	both pass	keep separate	A persistent private presentation identity can remain stable while each outward action uses a different represented party or disclosure posture.
J-M10	AI Twin	fails as one generic Capability MLE	composed product realization	It bundles agent definition, identity, dedicated knowledge allocation, access, publication/sharing, retrieval, conversation, and governance capabilities.
J-M11	Director/manager/worker hierarchy	fails as universal fixed taxonomy	implementation profile	Runtime-role assignment is reusable; named hierarchy levels are one orchestration vocabulary rather than mandatory PIOS roles.

Phase 2 collaboration-and-execution-workspace tranche E7

ID	Candidate	Individual MLE result	Decision	Reason
L-M01	Maintain a collaboration channel	passes	separate CRD	A durable top-level conversation container has independently meaningful identity, audience, purpose, ordering, visibility and lifecycle outcomes.
L-M02	Maintain a work-linked discussion thread	passes	separate CRD	A focused discussion has its own root reference, participants, state, summary and resolution lifecycle without owning the linked subject.
L-M03	Record an attributed collaboration message	passes	separate CRD	One attributable communication has an independent sender, context, content, references, trust state, trigger eligibility and acceptance outcome.
L-M04	Manage a task-scoped execution workspace	passes	separate CRD	A bounded runtime area has its own provisioning, isolation, temporary-state, output-disposition, cleanup and failure lifecycle.
L-M05	Collaboration channel + discussion thread	both pass	keep separate	A channel can exist with zero or many threads, while a focused thread can move or be linked without redefining its containing channel.
L-M06	Discussion thread + collaboration message	both pass	keep separate	A thread persists across many messages and lifecycle transitions; a message remains one attributable communication that can also exist without a work-linked thread.
L-M07	Delegation or assignment	passes through existing work-object capabilities	reuse composition	Responsibility and assignee state belong to the applicable task, project, workflow, background operation or runtime-role contract rather than a second generic delegation object.
L-M08	Multi-agent orchestration	fails as one generic Capability MLE	composed workflow/profile	Planning, assignment, role binding, execution, monitoring, communication, handoff, result preservation and escalation have separate actors and outcomes.
L-M09	Execution handoff	fails independently at current evidence depth	shared transition and existing-capability reuse	Durable source/target/status/provenance fields support transitions among work and background-operation contracts, but no additional universal outcome is established.
L-M10	Circle application	fails as one generic capability	audience projection and composed realization	Circle combines channel, thread, message, attention, agent, work-reference and navigation capabilities through one replaceable harness.

Phase 2 conversation-history-and-personal-interfaces tranche F1

ID	Candidate	Individual MLE result	Decision	Reason
N-M01	Maintain personal conversation history	passes	separate CRD	Conversation continuity has independent admission, ordering, provenance, correction/hiding, source-awareness and lifecycle outcomes across many interactions and views.
N-M02	Maintain an owner access path	passes	separate CRD	A route to owner control has independently meaningful identity, role, assurance, supported operations, priority, health, verification and retirement state.
N-M03	One physical chat history per owner	fails as a universal requirement	implementation profile	One logical continuity outcome is supported, but current PIOS can preserve it through retained sources and events without requiring one table or stream.
N-M04	Collaboration message + personal conversation history	both pass	keep separate	One message can be recorded independently, while history governs continuity, membership, ordering and provenance across many interactions and sources.
N-M05	Personal conversation history + canonical event log	both pass with distinct scopes	keep separate	Conversation history is a communication-domain continuity record; the event log is the complete cross-domain canonical occurrence spine.
N-M06	Imported-history integration	fails as one generic capability	reuse composition	It combines retention, import sessions, source promotion, history membership, retrieval allocation, summaries, pattern detection and proposal governance.
N-M07	Persistent personal memory	fails as one generic Capability MLE	architecture composition	It names the cumulative outcome of sources, events, notes, conversations, knowledge, relationships, decisions, retrieval and derived representations.
N-M08	Ally Panel, PIOS Panel, Home Panel and app navigation	fail as generic capabilities	audience projections and realization patterns	They arrange access to independently meaningful capabilities but do not own additional domain outcomes.
N-M09	Execution Tracking Surface	passes through existing capabilities	audience projection	Operational-run inspection, work-object state, background-operation state, artifacts and results already provide the underlying outcomes.
N-M10	Owner access path + domain authorization	both pass	keep separate	Registering or reaching an interface does not grant permission for every action exposed through it.

Phase 2 device-continuity-and-reading tranche F2

ID	Candidate	Individual MLE result	Decision	Reason
Q-M01	Maintain an authorized device-local content cache	passes	separate CRD	Creating, reusing, inspecting, refreshing, or removing a permitted non-canonical local copy creates an independent availability outcome; detailed freshness and eviction profiles remain unresolved.
Q-M02	Maintain selected content for offline access	passes	separate CRD	Deliberately retained local availability remains meaningful without opportunistic caching; detailed authorization, expiry, revocation, and reconnection profiles remain unresolved.
Q-M03	Manage a personal reading queue	passes	separate CRD	Source-linked admission and unread, read, saved, archived or removed state create a coherent owner outcome independent of source storage and rendering.
Q-M04	Recent cache + selected offline content	both pass	keep separate	One outcome supports automatic local-copy lifecycle management; the other records deliberate owner selection for disconnected availability.
Q-M05	Device-local content + pending capture/edit	both pass through separate capabilities	keep separate	Accepted local copies and unconfirmed submitted work have different authority and disposition states; pending work requires explicit reconciliation rather than cache cleanup.
Q-M06	Shared App Group or local database	fails as a capability	implementation mechanism and trust boundary	Shared storage supports cache or offline realization but cannot itself express per-app authorization or canonical truth.
Q-M07	Reader application	fails as one generic capability	audience projection and composed realization	It combines queue state, source retrieval, presentation preferences, classification, summary display, derivation and export actions.
Q-M08	Save-for-later and mark-read controls	fail independently at this evidence depth	interaction transitions inside reading-queue management	The source establishes lifecycle state changes within one reading purpose, not portable standalone capabilities with broader consumers.
Q-M09	AI-organized reading topics	passes through existing capabilities	reuse contextual classification and assisted labels	Topic generation organizes queue items but does not require a new reading-specific classification capability.
Q-M10	Generate or save a reading PDF	passes through existing capabilities	reuse derivation and retention	The output is a derived representation of retained content followed by ordinary governed retention.
Q-M11	Reader layout and settings	fail as generic capabilities	audience and realization patterns	Typography, colors, scrolling, feed layout and app navigation affect presentation without owning the reading-state outcome.

Phase 2 cost-and-resource-governance tranche F3

ID	Candidate	Individual MLE result	Decision	Reason
R-M01	Estimate operation resource cost	passes	separate CRD	A forecast has its own request, assumptions, method, scope, uncertainty, validity and success/failure outcome before execution.
R-M02	Approve cost-governed work	passes through existing governance capabilities	reuse composition	The approval is a governed proposal or action decision evaluated through explicit owner intent, standing rules and execution policy.
R-M03	Reserve an operation resource budget	passes	separate CRD	An expiring hold has independent operation binding, amount, availability, adjustment, release and denial states before actual use is known.
R-M04	Record actual resource consumption	passes through an existing capability	reuse metric observation	One attributable measured value already fits the metric-observation contract; cost semantics are supplied by its measurement definition.
R-M05	Reconcile operation resource usage	passes	separate CRD	Comparing actual use with estimate, reservation, budget and completed scope produces an independent final/interim accounting outcome.
R-M06	Partial completion	passes through existing work/run capabilities	reuse lifecycle state	The domain operation or background run owns what completed and what remains; resource reconciliation records why and what allowance remains.
R-M07	Cost-aware standing preference	passes through existing capabilities	reuse standing rule and execution policy	Thresholds, automatic allowance and depth limits are scoped authority/policy rather than a parallel cost-specific rule system.
R-M08	Most-relevant-first processing	fails as a standalone capability at this evidence depth	execution-scope/ordering profile	It selects the order and bounded subset for another processing capability and has no independent domain outcome.
R-M09	Cost and Credit Governance	fails as one generic Capability MLE	source context and composition	The umbrella joins three passing capabilities with approval, policy, observation, execution and commercial concerns that remain independently meaningful.
R-M10	Credits	fails as a capability	shared unit/profile	Credits are an internal resource comparison unit unless a separate commercial profile defines monetary meaning.
R-M11	Billing, payment, invoices, taxes and refunds	excluded from this tranche	separate commercial domain	The admitted source explicitly frames cost governance as product control rather than a billing system.

Phase 2 source-discovery-and-owner-knowledge tranche F4

ID	Candidate	Individual MLE result	Decision	Reason
S-M01	Discover a candidate information source	passes	separate CRD	Candidate discovery has its own submitted subject, inspection boundary, findings, confidence, evidence and no-candidate/blocked/failed outcomes before registration.
S-M02	Review a source candidate	passes through existing capabilities	reuse structured review	Suitability, evidence and unresolved risk can be assessed through the generic structured-review contract.
S-M03	Decide source permission	passes through existing capabilities	reuse governed proposal/authorization	Owner or policy disposition is independently governed and must not be embedded into discovery.
S-M04	Register, backfill, and collect incrementally	each passes or reuses existing source/import capabilities	keep separate	Registration, bounded historical intake and recurring collection have different authority, state and failure outcomes.
S-M05	Maintain owner implementation documentation	passes	separate CRD	The owner can create, revise, supersede or archive a human-readable account of their actual system without modifying the framework or runtime configuration.
S-M06	Synchronize a Knowledge Environment	fails as currently evidenced	unresolved implementation capability/profile	Historical sync technologies are examples, while source-of-truth, conflict, authority and durability relationships remain unresolved.
S-M07	Maintain agent behavior files through aliases	fails as a separate capability	realization pattern and agent-definition reuse	Aliases expose system-required files; the governed agent definition/instructions retain their own lifecycle.
S-M08	Promote implementation insight to reference documentation	fails as a PIOS owner-system capability at current scope	project contribution workflow	It is a framework-authoring/publication process rather than an operating capability of the owner's PIOS instance.
S-M09	Assess knowledge-layer integrity	passes	separate CRD	A bounded read/evaluate operation has independent scope, rules, findings, severity, evidence, coverage and completion/failure outcomes.
S-M10	Repair knowledge-integrity findings	fails as one generic capability	reuse domain capabilities	Repairs may require note revision, proposal decision, concept lifecycle, access governance, link correction or index rebuild, each with separate authority.
S-M11	Dataview queries and lint scripts	fail as capabilities	realization tools	Query snippets and scripts support assessment but lack independent owner-domain purpose outside the assessment contract.
S-M12	Knowledge Environment	fails as one generic Capability MLE	architecture composition and source context	It bundles implementation documentation, personal knowledge, storage, synchronization, agent definitions, setup, access and publication concerns.

Phase 2 cognitive-memory-and-derived-representations tranche F5

ID	Candidate	Individual MLE result	Decision	Reason
T-M01	Maintain a personal Meaning record	passes	separate CRD	Subject-specific significance has its own evidence, context, confidence, confirmation, revision, rejection and supersession outcomes.
T-M02	Maintain evidence-linked Learning	passes	separate CRD	A durable lesson has an independently useful lifecycle and may be recalled without executing an adaptation.
T-M03	Meaning + reflective assessment	both pass	keep separate	Meaning answers why one bounded subject matters; reflection assesses a wider evidence set with open loops, residue, uncertainty and recommendations.
T-M04	Meaning + profile assertion	both pass	keep separate	Significance does not establish a factual profile claim, and a profile claim need not explain why it matters.
T-M05	Meaning + Learning	both pass	keep separate	Interpretation and durable lesson can be revised, rejected and recalled independently; either can exist without silently producing the other.
T-M06	Learning + standing rule	both pass	keep separate	A lesson may inform future reasoning, while a standing rule is explicit reusable authority with separate scope and decision evidence.
T-M07	Learning + reusable routine or workflow	both pass	keep separate	Knowledge about what may help later is not an executable repeated structure or installed process.
T-M08	Adapt behavior from Learning	passes through existing capabilities	reuse proposal, rule, preference, workflow, agent-definition, and execution governance	Adaptation changes another governed object and requires that object's authority and rollback semantics.
T-M09	Govern a derived-representation lifecycle	passes	separate CRD	Registry identity, replacement, rebuild, revocation, deletion and retrieval exclusion remain meaningful after a derivative is produced.
T-M10	Produce derivative + govern retained representation	both pass	keep separate	Transformation can finish before lifecycle registration, and lifecycle changes can occur repeatedly without re-running the transformation.
T-M11	Source-local graph snapshot	fails as a generic standalone capability	derived-representation profile	It specializes representation identity and evidence preservation while reusing derivation and registry lifecycle outcomes.
T-M12	Embedding, OCR, summary, text chunk, or retrieval index	fail as generic standalone capabilities	representation types and processing profiles	Their technologies and output shapes vary while source-linked production and lifecycle contracts remain reusable.
T-M13	Resolve proposed graph matches globally	excluded from this tranche	separate concept/entity-resolution capability	Confirming global identity changes the knowledge graph and requires a distinct evidence and authority contract.
T-M14	Compile retrieval context from representations	passes through existing capability	reuse governed retrieval-context assembly	Registry eligibility does not select or disclose context for a particular caller and purpose.
T-M15	Meaning service or application	fails as a generic capability	audience projection	It presents and queries Meaning records without becoming a new authority, data zone, or permission source.

Phase 2 historical-corpus closure tranche F6

ID	Candidate	Individual MLE result	Decision	Reason
V-M01	Promote a working artifact to canonical state	passes	separate CRD	Acceptance into durable authoritative state has independent review, approval, target, conflict, provenance, verification, and failure outcomes.
V-M02	Artifact promotion + execution-result preservation	both pass	keep separate	A Result records a meaningful execution outcome, while promotion governs any accepted temporary artifact entering its canonical lifecycle; not every instance of either requires the other.
V-M03	Artifact promotion + imported-source promotion	both pass	keep separate	Artifact promotion crosses a temporary-to-canonical boundary; source promotion controls staged participation in mapping, enrichment, retrieval, and History.
V-M04	Canonical home	fails as a capability	architecture relationship/shared term	It names where authority resides; promotion, revision, retrieval, synchronization and deletion perform the interactions.
V-M05	Shared Workspace	fails as one generic Capability MLE	architecture composition	It combines interfaces, shared information, conversation, attribution, visibility, retention, promotion, and owner authority.
V-M06	Dedicated Agent Computer	fails as a capability	deployment profile	A machine may host runtimes, services and workspaces, but physical placement does not define the reusable outcomes.
V-M07	Track agent capabilities and performance	passes through existing capabilities at current evidence depth	reuse agent definition, canonical events, metric definitions/observations, structured review, and policy	The source supplies desired measurements but not an additional universal lifecycle beyond those existing contracts.
V-M08	Application/session state	fails as one generic Capability MLE	implementation state and multiple profiles	Identity, provider selection, onboarding, attention, navigation, filters and presentation preferences have different purposes and authorities.
V-M09	Market surface and entry-page families	fail as owner-system capabilities	product/publication strategy	They describe how capabilities are communicated and discovered commercially, not a PIOS operating outcome.
V-M10	Area indexes, repository index, source log, roadmap and inconsistency register	fail as capabilities	project/navigation/provenance records	They organize or explain the historical source set rather than define owner-domain interactions.
V-M11	UI design standards and UI/backend terminology	fail as generic capabilities	implementation and presentation profiles	Layout, styling, component rules, and vocabulary mappings constrain realizations without owning an independent outcome.
V-M12	Note-capability linking	fails as a generic capability	implementation composition pattern	A shared database identity and attachment pattern realizes several existing capabilities without creating a new domain outcome.