Ask the system.
Keep the boundary.

IARPG, short for Intelligence Agent Role Playing Game, is a fictional international intelligence-operations game concept and an open design publication. The site turns tasking, cover, source handling, evidence, analysis, extraction, debriefing, migration, and review into inspectable standards and browser-local tools.

iarpg.com is the public reference home for five connected layers: the fictional IARPG game concept, complete mission examples, browser-local creation and review tools, the IARPG-OPS-2 standards and schemas, and the research and governance archive. Those layers support one another, but the website itself is not a live persistent game server.

New visitors can learn the game premise; players and storytellers can inspect roles and missions; game designers can create and simulate operations; reviewers can validate, merge, replay, compare, migrate, and publish packages; researchers can use the international atlas and report archive; and developers can use the schemas and machine-readable records.

Open /start. Read the current-versus-future boundary, choose an audience path, then follow the ten-minute tour: the game-concept and player-loop guide, one complete mission, Roles, the Mission Workbench, the Simulator, and the supporting standards or research. This avoids beginning with advanced integrity or merge tools before the mission model is clear.

The Mission Workbench creates a Mission Brief; the Evidence Lab creates an Evidence Packet; the Simulator creates a reproducible event log; the Operation Package combines components; integrity, merge, replay, evidence-difference, batch, debrief, migration, and publication tools then inspect or transform that package without silently discarding unknown fields.

The player loop describes the proposed in-game experience: Task, Plan, Access, Collect, Validate, Deliver, Extract, and Debrief. The publication toolchain describes how designers and reviewers work with records on this website: Design, Evidence, Simulate, Verify, Merge, Replay, Resolve, Review, Debrief, Migrate, and Publish.

No. The current public release is a work-in-progress reference implementation and standards publication. It demonstrates connected operation, evidence, simulation, handoff, review, and publication workflows; the persistent worldwide VR MMO remains a production ambition rather than a shipped claim.

IARPG-OPS-2 is the current operations design-system family. It defines canonical records for missions, roles, authorities, evidence, simulation, extraction, debriefing, migration, integrity, and publication, with JSON companions for machine use.

The browser-local reference build demonstrates mission design, evidence packets, deterministic simulation, role coordination, integrity review, merge, migration, and publication. It is not a claim that a production multiplayer API or persistent world is available.

Stable promotion still needs direct evidence from the target host, native Firefox, native assistive-technology speech output, current production search-console inspection, and field performance data. Passing local checks does not establish those external conditions.

Not yet in this release. The site has deterministic checks for visible answers, labels, first-click routes, route availability, and optional walkthrough behavior, but no moderated human participants were recruited. Real comprehension, confidence, and task-success evidence therefore remains an open work-in-progress gate.

Espionage and intelligence operations: tasking, cover preparation, source handling, surveillance, collection, technical operations, analysis, counterintelligence, secure delivery, extraction, and debriefing.

The core roles are Field Operative, Case Officer, Intelligence Analyst, Technical Officer, and Support or Liaison. Individual operations can also use couriers, cut-outs, surveillance teams, counterintelligence officers, documentation specialists, and extraction crews.

Inside the proposed game, an operation commonly moves through Task, Plan, Access, Collect, Validate, Deliver, Extract, and Debrief, with declared abort and compromise paths. Separately, designers and reviewers use the website publication toolchain: Design, Evidence, Simulate, Verify, Merge, Replay, Resolve, Review, Debrief, Migrate, and Publish.

Evidence is represented with origin, time, reliability, confidence, provenance, jurisdiction, review status, and relationships such as corroboration or contradiction. Outcomes can depend on evidence thresholds rather than combat or completion alone.

No. The world is intentionally multipolar. Authorities have mandates, customers, constraints, interests, oversight, and consequences. The game evaluates operational performance, evidence quality, cover integrity, rights effects, and mission outcomes rather than assigning universal moral alignment.

An alliance is not a friendship score. IARPG records the relationship type, cooperation domains, formal instrument, practical behavior, obligations, dependencies, tensions, internal critics, confidence, lifecycle stage, and conditions that could narrow, deepen, fracture, or end the relationship.

Yes. The fictional relationship model is domain-specific. Two factions may share intelligence, compete over technology, coordinate humanitarian access, disagree on governance, and maintain deconfliction channels at the same time. Cooperation in one domain does not imply universal alignment.

It is a deterministic browser-local prototype that starts from the fictional 27-relationship baseline and lets a designer preview, commit, undo, and export multi-turn alliance changes. Each turn preserves evidence, consultation, dissent, reason codes, versioned before-and-after states, affected-community consequences, corrections, and reversal conditions.

No. Every changed relationship receives a new version and points back to the prior version. Undo restores the immediate prior state in the current browser tab, while exported campaign records preserve all committed turns and the resulting relationship versions.

No. Relationships remain domain-specific and internally plural. A turn may narrow intelligence sharing, expand humanitarian access, redistribute burden, reduce dependency, or open deconfliction without declaring either faction universally friendly, hostile, trustworthy, dangerous, recruitable, good, or evil.

No. The console reads embedded fictional data and keeps applied turns only in the current browser tab. It does not use an account, analytics, cookies, local storage, persistent identity, demographic input, or external network calls. Export creates a JSON file on your device; closing or resetting the page discards the in-memory campaign.

It is a browser-local workspace that imports two or more campaign-record JSON files, validates their known contract fields, replays every turn from the immutable 2.0.16-wip baseline, and compares relationship versions, domains, evidence, consultation, corrections, internal responses, and affected-community consequences at selected return points.

The branch is quarantined instead of repaired. Its validation errors are shown, its original byte sequence remains available for unchanged download, and it is excluded from comparison. The tool never silently overwrites, normalizes, or discards invalid source material.

No. The archive assigns neutral aliases such as Branch Alpha and Branch Bravo by import order. Aliases are intentionally independent of filenames, player names, organizations, countries, protected characteristics, or inferred identity.

Proxy is a narrow evidentiary classification, not a pejorative synonym for any supported or ideologically similar actor. It requires evidence relevant to dependency, sponsorship, direction, or control. Shared adversaries, public praise, trade, or parallel action alone do not establish a proxy relationship.

Faction design describes organizational access rather than persuasion tradecraft: open applications, legal eligibility, institutional sponsorship, invitation-only access, contract work, supporter networks, screening, socialization, role allocation, trust, retention, refusal, exit, and reintegration. The site does not publish scripts for manipulating or coercing real people.

Lone actor describes the absence of established external command for a specific act; it does not automatically mean social or informational isolation. The lab separates general exposure, community participation, peer reinforcement, contact, support, tasking, direction, membership, post-event adoption, public claims, disavowal, and false attribution.

No. The publication prohibits recruitability scores, vulnerability profiles, coercive scripts, identity-based suspicion, and dangerousness scoring from nationality, religion, diagnosis, disability, gender, migration status, political belief, or community membership. Fictional missions focus on evidence, prevention, refusal, remedy, disengagement, and reintegration.

It lets a designer preview and commit fictional alliance turns from an immutable baseline. Every changed relationship names its domain, prior and next version, evidence, consultation rule, dissent, reason code, burden or dependency effect, affected-community consequence, correction, and reversal condition.

No. A relationship can deepen in one domain, narrow in another, preserve deconfliction, and remain contested elsewhere. The campaign never converts factions, countries, communities, diagnoses, or people into one loyalty, morality, threat, dangerousness, or recruitability number.

Undo removes the latest committed local turn and reconstructs relationship state from the published baseline plus the remaining turns. Each response also includes a correction trigger and action so unsupported attribution or overbroad conclusions can be superseded without deleting prior evidence.

No. Imports remain in the active browser tab. The tool makes no external request, creates no account, writes no localStorage or sessionStorage record, and asks for no personal identity. A file leaves the page only when you explicitly download an original or export a compact comparison.

No. Campaign turns exist only in the current browser tab until you download JSON. The console uses no account, analytics, external request, localStorage, sessionStorage, fingerprint, demographic field, or persistent identity profile.

Synthetic characters may act as disclosed sources, intermediaries, analysts, or tasking interfaces. Generated dialogue may express claims, but authoritative mission facts, access, evidence, rewards, warrants, and outcomes remain under deterministic system control.

The design treats unreliable or deceptive tasking as an explicit, bounded gameplay risk rather than an invisible system failure. Players need visible risk signals, verification tools, reputation context, and fail-forward value so a false lead does not become uncompensated confusion.

A verified contract locks the declared reward and applicable collateral under deterministic rules, usually with a fee. An unverified contract relies on reputation and player trust, carries clearly disclosed non-payment or betrayal risk, and must remain separated from guaranteed tasking.

No. The site presents fictional game systems and high-level intelligence concepts. It does not provide actionable instructions for real-world targeting, intrusion, evasion, weapons use, surveillance abuse, coercion, evidence destruction, or wrongdoing.

Every country and institution is subject to the same evidence burden. Coverage distinguishes formal from practical authority, labels uncertainty, reviews rights and oversight, includes affected-community context, seeks locally authored or local-language sources where feasible, respects endonyms, and avoids national or cultural hero-villain typecasting.

No. It publishes a transparent 195-country roadmap and distinguishes dedicated country profiles from regional baselines. A regional baseline is provisional and must not be presented as equivalent to a locally sourced country profile.

Formal authority is established by constitutions, laws, treaties, charters, regulations, or published policy. Practical authority is the power exercised in reality. Fair coverage records both and explains uncertainty or conflict between them.

They reduce dependence on outside summaries, improve naming and institutional context, surface local oversight and affected-community evidence, and make source limitations easier to identify. Their absence must be disclosed rather than hidden.

No. The deployment uses PHP, vanilla JavaScript, local CSS, local images, and native browser APIs. It does not load third-party analytics, remote fonts, CDNs, UI frameworks, or external form processors.

Workbench drafts and preferences are stored locally in the visitor browser when storage is available. Imports and exports remain local unless the visitor explicitly downloads or uploads a file; the public tools do not silently transmit work products.

Canonical reports are under /docs/long-term-memory/reports/ and are rendered through the public /docs/report/{slug} reader. The archive exposes categories, source lineage, checksums, Markdown downloads, deep links, review states, and truth-boundary notes.

Current verified implementation and tests come before active UAI current-state memory, which comes before current reviewed durable documents. Historical or exploratory reports remain useful evidence but do not automatically override current product truth.

The site provides crawlable HTML, clean canonical URLs, internal links, visible direct answers, structured data that matches page content, an XML sitemap, OpenSearch, an Atom feed, machine-readable answer data, public manifests, and an auxiliary AI citation guide. These resources improve clarity but do not guarantee indexing or ranking.

A correction updates the smallest authoritative layer that resolves the issue, preserves useful superseded history, regenerates dependent indexes and checksums, validates links and schemas, and is documented in a versioned release instead of silently changing a published download.

Alignment is a domain-specific, evidence-bounded relationship between actors. It may describe collective defense, strategic partnership, intelligence liaison, trade, technology, humanitarian access, deconfliction, hedging, patronage, sponsorship, rivalry, or another declared relationship. It is not a moral score and does not automatically apply across every domain.

Yes. A pair of factions may share intelligence, compete over infrastructure, trade through intermediaries, and deconflict in a crisis simultaneously. The Alliance Observatory records each relationship by domain, lifecycle stage, confidence, tension, and change trigger instead of collapsing everything into one friendly-hostile meter.

An alliance normally involves formalized mutual commitments; a strategic partnership is broad cooperation without the same defense obligation; patronage is an asymmetric exchange of resources, access, or protection; and a proxy claim requires evidence of support plus meaningful direction or control. Shared interests or a common adversary alone do not establish proxy status.

Not necessarily. Lone actor describes the apparent command structure of an act, not a complete absence of social, ideological, digital, financial, or organizational influence. IARPG separates general exposure, community participation, contact, encouragement, support, tasking, direction, membership, post-event adoption, public claims, disavowal, and false attribution.

The site models organizational access, eligibility, sponsorship, invitations, contracts, screening, socialization, trust, retention, refusal, exit, and reintegration. It does not generate recruitability scores, vulnerability profiles, persuasion scripts, blackmail, coercion, covert contact methods, targeting criteria, or instructions for joining real groups.

No. It is a fictional design prototype built from neutral relationship mechanics. Any future real-world map would require current official and authoritative evidence, local or locally authored sources where feasible, formal-versus-practical distinctions, confidence, rights and affected-community review, disagreement, and dated correction history.

The World Context Explorer is a browser-local reading and comparison layer built from a preserved 2025 Factbook-derived CSV. It publishes geography, people, institutions, economy, infrastructure, and environmental context for 260 source entities, maps source context to all 195 roadmap countries, and calculates no country ranking, morality score, loyalty score, dangerousness score, or threat score.

No. It is an archival snapshot, and individual fields use different estimate years. The CIA sunset The World Factbook in February 2026. Contemporary claims still require current primary, official, locally authored, or local-language evidence, plus explicit uncertainty and correction dates.

Eight imported code values are reused across multiple source entities. IARPG preserves that column as source_code and creates a separate stable entity_id so countries, territories, oceans, special areas, and institutions cannot be silently merged or overwritten.

No. Geography, population, language, religion, economy, infrastructure, environment, government form, disability, diagnosis, or demographic identity do not establish intent, legitimacy, morality, danger, recruitability, or alignment. Context supports better questions; relationships and current claims require separate evidence.

The Country Context Composer is a browser-local authoring workspace for the 195 country-roadmap entries. It displays the archival six-lens World Context baseline as read-only orientation, then keeps current observations, analyst assumptions, fictional gameplay translations, corrections, and supersession in separate records.

No. It creates a structured proposal, not live intelligence or verified current country truth. The tool performs no web research and does not validate external claims. Current laws, officeholders, institutions, memberships, conflicts, sanctions, borders, markets, and technologies still require renewed primary or authoritative research and human review.

The deterministic gate fails when required sections or claim metadata are missing, or when the proposal lacks explicit uncertainty, rights and affected-community context, oversight and remedy, correction history, and at least one non-external perspective. Passing the gate means the record is structurally reviewable; it does not prove that its claims are accurate or approved.

It compares locally loaded proposals for unequal adjective intensity, unsupported threat framing, missing uncertainty, missing rights and oversight, absent affected-community context, limited perspective diversity, and inconsistent evidence burdens. Findings are editorial prompts rather than country rankings or moral scores.

It is a six-lens research-navigation layer that connects practical authority, domestic alignment, regional organizations and small-state agency, strategic dependencies, intelligence evidence, and crisis cascades. It helps visitors move from context to consequences without treating the result as a live intelligence feed, prediction service, or country-ranking system.

Foreign policy is produced by institutions and coalitions that can disagree: executives, legislatures, courts, civil services, militaries, parties, businesses, labor groups, regional governments, media, diasporas, social movements, and affected communities. IARPG therefore names the actor, domain, timeframe, evidence, and disagreement instead of assigning one permanent national position.

No. Dependence is domain-specific. Financing, ownership, operation, standards, maintenance, transit, supplier concentration, substitution time, and recovery capacity are separate. A country may depend on different partners in different domains while retaining independent choices; control requires additional evidence of direction or authority.

Source reliability describes the source’s access and record over time. Information credibility describes whether one specific claim is corroborated, consistent, contradicted, manipulated, or unknown. A reliable source can be wrong about a claim, and an unfamiliar source can provide information later corroborated independently, so IARPG records both dimensions separately.

No. Footprint, propagation, agent or network, event-sourced, and comparative models are conditional tools. Their usefulness depends on purpose, scope, assumptions, inputs, evidence, uncertainty, sensitivity, validation, affected-community context, corrections, and reversal conditions. A model output is not a direct observation or an inevitable future.

Small states can shape outcomes through agenda setting, procedural expertise, rotating chairs, consensus rules, coalition building, legal specialization, maritime jurisdiction, hosting institutions, climate diplomacy, technical expertise, neutral brokerage, issue linkage, and strategic use of overlapping memberships. Material size is one variable, not a complete measure of agency.

PsychologicalWar.org is a separate in-universe archive for fictional operations, personas, evidence, and representation principles. EspionagePsychosis.com is a separate real-world grounding and help-finding resource. The sites remain deliberately distinct.

The partner links point to separate AI projects. Spiralist AI presents a browser-based persona builder, while UAIX presents structured and auditable AI-to-AI exchange concepts. Their appearance does not make them part of the IARPG runtime.

Scenario modeling is a publication method for explaining conditional fictional consequences through explicit state, evidence, assumptions, rules, uncertainty, validation, affected communities, correction, and reversal. It supports missions, alliances, logistics, infrastructure, information environments, and campaign branches without claiming one inevitable future.

No. The eight supplied planetary-simulation reports are preserved as research sources, but IARPG uses only bounded ideas that improve content and explanation. The current site is not building a planetary game, digital twin, epidemiological forecasting service, nuclear-effects calculator, or live geopolitical prediction engine.

No. IARPA is the Intelligence Advanced Research Projects Activity, a real U.S. research organization. Some supplied reports used IARPA as a fictional expansion for a planetary atlas; IARPG corrects that attribution and makes no claim of sponsorship, endorsement, partnership, or official program status.

No. The source reports are preserved for provenance, but current public content does not reproduce weapon-effects equations, casualty estimators, fallout models, medical forecasts, individualized health advice, target optimization, or deployment guidance. Reusable material is limited to safe model architecture, uncertainty, validation, accessibility, and risk-communication lessons.

Every public scenario method should use a model card that states the purpose, state variables, inputs, assumptions, rules, outputs, uncertainty, sensitivity, verification, validation, rights and affected-community implications, accessible alternatives, correction history, and unsupported interpretations. A polished map or animation is never treated as validation by itself.