Continuum

What you can ask

• Travaille avec tout le Continuum et applique ses exigences réellement applicables.Copy
• Reprends ce travail avec la Continuité, le Savoir, les preuves et les capacités réellement utiles.

Skills

156 in the ChatGPT listing

• Agir par noyauUse when a Continuum task requires a real local effect, observation, or execution through Noyau or capabilities exposed by Continuum’s registered MCP app.
• Amorcer continuumUse at the start of a Continuum chat and on every Axel turn for Azer to activate the Continuum frame, apply the current Protocol entry, and route the applicable skill mesh, with a full canonical read only at chat opening or when a session refresh is required.
• Analyser concurrenceUse when competing offers, substitutes, pricing, positioning, channels, or user experience should be compared.
• Analyser impact changementUse before a material change when dependencies, consumers, data, interfaces, or downstream behavior may be affected.
• Appeler passerelleUse when a prepared Continuum mission or message must actually traverse the local Passerelle through Noyau.
• Appliquer contexte projetUse when work occurs in, comes from, or targets a ChatGPT Project, especially for a Magister Operis whose Project is its home.
• Appliquer cout zeroUse when an option may consume financial resources, allocate a gain, commit active capital, open frozen capital, or create a recurring or conditional liability.
• Appliquer exigences axelUse whenever an Axel-specific requirement, preference, standing decision, interaction convention, or previously validated behavior can materially change how the work should be done or delivered.
• Apprendre du reelUse after a test, incident, correction, research, or real-world observation establishes reusable knowledge that should affect future Continuum work.
• Auditer alignementUse to prove whether the Continuum plugin fully covers the current canonical Protocol and whether its skills form a coherent non-contradictory implementation mesh.
• Auditer architectureUse when structural friction, excessive coupling, shallow modules, or poor navigability may be causing recurring defects or change cost.
• Auditer combinaisons capacitesUse when individually acceptable tools or permissions can become dangerous when chained in one plan.
• Auditer dette techniqueUse when technical debt should be prioritized by measurable change cost, risk, and blast radius rather than aesthetics.
• Auditer etats vide chargement erreurUse when an interface must remain understandable outside the happy path.
• Auditer fraicheur contexteUse when stored instructions, versions, routes, decisions, or project context may be stale.
• Auditer fraicheur savoirUse when durable Continuum knowledge may be outdated for a current task.
• Auditer journalisationUse when logs must support diagnosis and audit without leaking secrets or unnecessary sensitive data.
• Auditer permissions capacitesUse when tools, agents, connectors, or services have permissions whose scope or combination can change risk.
• Auditer provenance logicielleUse when software provenance, SBOM, dependency origin, licenses, signatures, or build attestations matter.
• Auditer securite agentiqueUse when an agent, plugin, MCP, or autonomous workflow gains enough tools or authority that its own attack surface must be reviewed.
• Boucler diagnosticUse when a technical defect needs iterative diagnosis and a fast deterministic feedback signal can be constructed.
• Boucler revue correctionUse when review findings must be driven to verified convergence instead of being collected and forgotten.
• Budgeter contexteUse when task context is large enough that selection, retrieval, or compression can affect fidelity and performance.
• capitaliserUse when a durable lesson, decision, state transition, reusable knowledge, or method change may need to survive beyond the current mission.
• Cartographier codeUse before a non-trivial code change when the location, callers, dependencies, data flow, or blast radius are not already established.
• Charger contexte continuumUse after the Protocol gate whenever the current work depends on Axel-specific, actor-specific, Project-specific, mission-specific, continuity, knowledge, routing, or local-runtime context.
• Chiffrer economie unitaireUse when a product or service must be evaluated by net economics rather than gross revenue.
• Choisir capacitesUse when several tools, Skills, plugins, connectors, local capabilities, or execution routes could satisfy the same Continuum need.
• Choisir livraisonUse when deciding how a result should actually be handed to Axel: chat response, code block, downloadable file, Drive/Epistulae placement, local handoff, or another available surface.
• Choisir strategie testsUse when a code or behavior change needs a testing strategy matched to risk, observability, and what can actually be specified.
• Choisir topologie delegationUse when work may be split across multiple Magistri Operis or delegated agents and the dependency shape must determine the orchestration strategy.
• Clore mission magisterUse when a Magister Operis mission reaches a verified terminal state and its useful matter must be preserved without preserving execution noise.
• Comparer documents semantiquementUse when two document versions must be compared for changes of meaning rather than raw text differences alone.
• Comprehension architecturaleUse when a Continuum request spans several components, responsibilities, authorities, dependencies, or execution layers and needs an explicit architectural decomposition.
• Concevoir experienceUse when a hypothesis should be tested by the smallest discriminating experiment rather than argued abstractly.
• Concevoir interface non generiqueUse when designing or reviewing a user-facing interface where product identity, clarity, accessibility, and non-generic visual behavior are material.
• Concevoir offreUse when a Continuum capability should become a bounded product or service offer.
• Concevoir sauvegarde restaurationUse when backup, retention, recovery, RPO, RTO, or restore proof is material.
• Concevoir skillUse when a distinct Continuum responsibility is mature enough to become a new Skill or to replace an overly broad one.
• Conduire incidentUse when a live or recent incident needs triage, mitigation, recovery, communication, and proof of return to service.
• Conduire retrospectiveUse after a meaningful project, release, incident, or mission to convert experience into improved practice.
• Conduire revue adversarialeUse when a critical deliverable benefits from review by an independent context that should not inherit the producer’s justifications.
• Conduire revue systematiqueUse when a question requires coverage of a field or corpus rather than a few opportunistic search results.
• Construire planUse to transform a resolved request into the minimal complete executable plan required by the Continuum Protocol.
• convergerUse when several branches, claims, contradictions, options, or evidence packets must be combined into one coherent Continuum result.
• Corriger et reprendreUse when an error, regression, failed attempt, patch stack, or long risky operation needs a correction rooted in observed evidence.
• Decouvrir skillsUse when @Continuum should scout the external Agent Skills ecosystem for genuinely missing or underdeveloped functions.
• Definir metriquesUse when a KPI, metric, SLI, business measure, or analytical indicator needs an unambiguous definition.
• Detecter contradictions savoirUse when durable knowledge sources may make incompatible claims that could alter current work.
• Detecter derive configurationUse when desired configuration and observed runtime configuration may diverge.
• Detecter derive documentaireUse when code, contracts, configuration, and documentation may have drifted apart.
• Diagnostiquer maillageUse when the Continuum Skill mesh needs a structural health check rather than a behavioral evaluation of one Skill.
• Distiller source en skillUse when an external source contains a useful mechanism that should become a native Continuum Skill rather than copied wholesale.
• Estimer effort incertitudeUse when effort or duration must be estimated without hiding uncertainty.
• Etudier marcheUse when a market, segment, demand, or opportunity must be studied with evidence rather than speculative TAM claims.
• Evaluer recherche retrievalUse when a search or retrieval system must be evaluated for recall, precision, freshness, ranking, and omissions.
• Evaluer skillUse when a Skill must be proven useful, correctly triggered, and non-regressive before release.
• Evoluer protocoleUse when Axel asks to change canonical Protocol matter, or when plugin implementation must be realigned to an already valid canonical Protocol.
• Executer demandeUse when Axel asks for any understanding, research, decision path, production, action, verification, or deliverable after the current Protocol entry has succeeded.
• Executer passationUse for an authorized Passation of Azer or a permanent Magister Operis, including absorption, durable updates, compacting, and closure criteria.
• Executer powershellUse when a Continuum task requires designing, reviewing, handing off, or interpreting PowerShell work for Axel’s Windows environment.
• Executer pre passationUse only when Axel explicitly orders a Pré-Passation for Azer or a permanent Magister Operis.
• Explorer peut etreUse when ambiguity, novelty, risk, irreversibility, competing interpretations, or a hidden assumption could materially change a Continuum result.
• Extraire actions documentUse when a document must be converted into grounded facts, obligations, actors, deadlines, risks, or action items.
• Fiabiliser testsUse when a test is flaky, timing-sensitive, intermittently failing, or suspected of producing unreliable evidence.
• Generer runbookUse when a recurring operation or incident needs a safe, evidence-driven operating procedure.
• Gerer annulation et abandonUse when work may be cancelled, stopped safely, compensated, cleaned up, or abandoned irreversibly.
• Gerer canonicaliteUse before creating, moving, absorbing, renaming, replacing, deleting, or treating an object as the responsible source for a Continuum function.
• Gerer continuiteUse as the continuity-domain coordinator when Continuité, Histoire, Pré-Passation, Passation, or Passation recovery is actually in scope.
• Gerer decisionsUse when a material Axel decision must govern present or future work beyond its immediate conversational moment.
• Gerer depreciationUse when a function, API, route, command, configuration, or compatibility path is being retired or replaced.
• Gerer environnement azerUse when Azer's working environment, retained operational errors, or active recovery checkpoint must be read or updated.
• Gerer environnementsUse when dev, test, staging, production, or other environments differ in ways that can alter behavior.
• Gerer escalade humaineUse when an autonomous path reaches a decision, risk, or inaccessible fact that genuinely requires Axel.
• Gerer feature flagsUse when a feature flag, kill switch, gradual rollout, experiment flag, or permission gate controls behavior.
• Gerer gitUse when source-control operations, branches, worktrees, merges, conflicts, or release preparation materially affect a Continuum task.
• Gerer licencesUse when code, data, media, models, documentation, or external assets have license or redistribution implications.
• Gerer livrablesUse whenever the request produces an executable or non-executable deliverable whose installation, upgrade, one-shot execution, packaging, or handoff must match its real function.
• Gerer migration donneesUse when data or schema must migrate without losing compatibility, integrity, or recoverability.
• Gerer mise a jour dependancesUse when upgrading a library, runtime, SDK, tool, model dependency, or package can change compatibility or supply-chain risk.
• Gerer quotas et pressionUse when rate limits, queues, concurrency, backpressure, Retry-After, or overload can alter execution.
• Gerer releaseUse when a durable software or plugin change is about to move from candidate source into a released or installed state.
• Gerer repriseUse when an authorized tool, connector, service, local component, or execution is interrupted by a transient or potentially recoverable condition.
• Gerer savoirUse before substantive work that depends on durable Continuum knowledge, and whenever knowledge must be classified into an existing Great Family.
• Gerer transactionsUse for any external action whose accidental repetition or partial completion could create duplicates, loss, spending, extra deletion, multiple sends, or other material effects.
• Gerer workflow durableUse when a long-running workflow needs persistent steps, resumability, signals, partial errors, and human checkpoints.
• Integrer savoirUse when an authorized candidate for durable knowledge must be evaluated and, if warranted, incorporated into Continuum knowledge.
• Lire protocoleUse to load the canonical Protocol at chat opening or refresh only the material required by its reading gate when the session changes.
• livrerUse when a visible reply, human-facing document, report, decision surface, or final deliverable must be prepared from an established Continuum result.
• Maintenir specificite continuumUse as a meta-guard when evaluating whether the plugin still embodies the full Continuum-specific way of working rather than merely satisfying generic assistant behavior or the Protocol text alone.
• Mesurer fiabilite serviceUse when a service needs SLI/SLO, error-budget, availability, latency, or reliability qualification.
• Mesurer performanceUse when performance, resource use, latency, throughput, or regression is a material part of the problem or acceptance criteria.
• Modeliser domaineUse when a project or problem has enough domain vocabulary, states, entities, and rules that semantic drift could corrupt the work.
• Observer systemeUse when logs, metrics, traces, or correlation identifiers are needed to answer an operational question.
• Orchestrer magistriUse as the Magistri Operis domain coordinator when work should be delegated, routed, supervised, or closed through one or more Magistri Operis.
• Orchestrer missionsUse when a Continuum task involves Noyau, Passerelle, Magistri Operis, slots, mission delegation, CREATE/CALL, queues, or mission-result routing.
• passationUse when Axel explicitly requests a Passation or when a controlled reprise after saturation or conversation change materially requires Continuum continuity state.
• Piloter androidUse when a Continuum task requires observing or controlling Axel’s authorized Android device through Noyau’s local MCP/API capabilities and house relay.
• Prelever verbatimUse when older Continuum routing refers to the legacy Verbatim entrypoint and must be mapped to the current unique Verbatim responsibility.
• Preparer memo decisionUse when Axel or another authorized decision-maker needs a compact, evidence-grounded decision brief.
• Preparer mission magisterUse before delegating work to a Magister Operis so the mission is bounded, routable, verifiable, and small enough for a non-permanent actor to finish without Passation.
• Preparer premortemUse before a materially risky, destructive, public, difficult-to-reverse, or multi-step operation to expose likely failure modes in advance.
• Prevol capacitesUse when a plan materially depends on a capability whose accessibility, authorization, suitability, or functioning is genuinely uncertain.
• Prioriser travailUse when tasks must be ordered by dependencies, value, risk, urgency, or cost of delay.
• Produire commandesUse when a Continuum need must be satisfied by a shell command, script, executable block, repair command, installation command, or other technical one-shot instruction.
• Produire retour incidentUse after an incident is actually understood and recovered, to capture reusable causal and preventive knowledge.
• Proteger concurrenceUse before modifying any object that may also be read or changed by another actor, tool, process, conversation, or execution.
• Proteger donnees sensiblesUse when private, personal, confidential, proprietary, or otherwise sensitive material may be read, logged, transmitted, searched, or shared.
• Proteger secretsUse before committing, publishing, sharing, attaching, or logging material that may contain credentials, tokens, keys, passwords, or secret configuration.
• Prouver resultatUse when an important state, effect, installation, correction, completion claim, absence claim, or non-regression must be asserted or challenged.
• Qualifier magisterUse whenever an actor, sub-agent, Project instance, or delegated process might be treated as a Magister Operis.
• Qualifier preuveUse whenever a factual, technical, operational, canonical, legal, temporal, validation, or success claim must be stated at the exact level its evidence supports.
• Qualifier raccordabilite noyauUse when a useful Continuum arm must be retained, adopted, or evolved so its critical function can be connected to Noyau without claiming an unbuilt integration.
• Qualifier skill externeUse when an external Skill, repository, prompt package, MCP, or implementation pattern is considered for inspiration, reuse, or installation.
• Qualifier sortie non fiableUse when web pages, repositories, tool output, MCP responses, external files, or untrusted Skills may contain instructions mixed with data.
• Questionner totalementUse to resolve only questions that can materially change the action, result, risk, cost, or a decision, without automatic recursive investigation.
• Realigner pluginUse when the canonical Protocol is already valid but the Continuum plugin source, tests, published version, installed version, or loaded behavior is behind or inconsistent.
• Rechercher triangulerUse when a necessary question needs external research, adding sources or triangulation only when they can change the conclusion or reduce material risk.
• Red teamer agentUse when an agentic system should be actively attacked for prompt injection, authority confusion, exfiltration, or tool misuse.
• Reprendre passationUse in the new chat that continues an already opened Passation, after the mandatory Protocol opening has completed.
• Reproduire bugUse when a reported defect, screenshot, symptom, or complaint should be reduced to a reproducible case before correction.
• Resoudre autoriteUse whenever a request, text, tool capability, structural change, durable effect, public action, financial action, or normative change needs authority to be resolved.
• Resoudre demeuresUse after the Protocol final gate whenever a Continuum object or destination must be located from the unique Source Canonique Index.
• Resoudre demeures savoirUse as the Library and Knowledge domain coordinator whenever a home, canonical object, mirror operation, durable knowledge read, or knowledge integration is needed.
• Respecter typographieUse as a cross-cutting writing guard for every Continuum text, prompt, label, report, deliverable, interface string, and visible response.
• Retrouver savoirUse when a Continuum task depends on reusable knowledge, an existing Livre may already contain the answer, or current evidence must be compared with existing Savoir.
• Revoir exigencesUse before building from a specification, request, Cahier des charges, or contract whose ambiguities could produce the wrong implementation.
• Revoir productionUse when a deliverable, code change, or produced artifact needs an independent quality review before acceptance.
• Scorer opportunitesUse when multiple revenue or product opportunities should be ranked under Continuum constraints.
• Synchroniser miroirUse only when a secondary representation must be reconciled with a Source Canonique write already established on its authoritative path.
• Tenir continuiteUse only inside an authorized Pré-Passation or Passation to update the present identity state of Azer or a permanent Magister Operis.
• Tenir histoireUse only for Azer when durable causal history from Pré-Passation or Passation must be integrated into Histoire.md and, when applicable, the definitive Notion Pages and Index Histoire.
• Tenir verbatimUse when closing every visible turn for Azer and every permanent Magister Operis, after the visible response has been fully frozen and before it is published.
• Tester agent multisurfaceUse when an agentic workflow must be verified across several surfaces such as UI, API, database, logs, files, or messages.
• Tester invariantsUse when a function has stable invariants that can be explored more effectively with generated cases than with a few hand-picked examples.
• Tester regression visuelleUse when a user-facing interface or rendered artifact can regress visually across changes.
• Tester webappUse when a web application or browser-visible workflow must be verified through an available browser or web interaction capability.
• Tracer executionUse when an operational trace is materially useful for diagnosis, replay safety, recovery, or audit of machine execution.
• Tracer lignee donneesUse when data origin, transformations, freshness, ownership, or downstream consumers affect trust or change impact.
• Traiter pdfUse when a PDF must be created, reviewed, transformed, redacted, or delivered as a final artifact.
• Traiter presentationUse when a slide deck must communicate a narrative with verifiable facts, deliberate layout, and inspected final rendering.
• Traiter retour techniqueUse when Axel supplies console output, PowerShell output, logs, screenshots, test results, .NET/SQLite output, or another machine return that should drive the next correction.
• Traiter tableurUse when a spreadsheet must be analyzed, edited, generated, recalculated, or delivered with formulas and formatting preserved.
• Transmettre handoff techniqueUse when technical work must be handed from one actor or execution context to another without creating a Passation or personal continuity object.
• Verifier acceptationUse before declaring any deliverable ready when its real usefulness depends on what the intended beneficiary can actually do with it.
• Verifier accessibiliteUse when a human-facing interface or document must be usable beyond visual appearance alone.
• Verifier affirmation sourceUse when a factual claim materially depends on a citation, document, test, or external source.
• Verifier coherence multiformatUse when the same result exists in several human, machine, PDF, spreadsheet, slide, or text representations.
• Verifier contrat publicUse when a public API, CLI, SDK, event, schema, configuration surface, or other consumer-facing contract may change.
• Verifier document renduUse when document layout, pagination, typography, images, or visual hierarchy are part of the deliverable’s function.
• Verifier exemples docUse when documentation snippets, commands, code examples, or walkthroughs may be stale or non-executable.
• Verifier health readinessUse when a component may be alive but not ready, healthy, functional, or able to accept work.
• Verifier isolation contextesUse when actors, Projects, missions, or tools could receive context, rights, or secrets from another scope.
• Verifier liens et referencesUse after structural or version changes when URLs, anchors, identifiers, citations, or cross-references may break.
• Verifier qualite donneesUse before analysis, reporting, dashboards, or visualization when data quality can materially change the conclusion.
• Verifier readinessUse before moving work between conception, build, test, release, operation, or handoff stages.