Studio explorer.
One operational index for every current Studio surface. Follow production, inspect Bench, or read the agent and API map without changing state.
Production
9 entriesCurrent production state and the next human gate.
HomeMars HouseWorld rules, telemetry, and story pressure.
WorldEpisodesAll episodes and their production state.
IndexSelection TapesThe active episode control room.
EpisodeCharactersCharacter identities, beliefs, and relationships.
CastCanonObjective facts and approved story history.
RecordAudienceAggregate audience signals and Crossroads proposals.
SignalNadia interviewScene intent, continuity inputs, and current coverage.
SceneProduction reviewHuman review for the active production run.
ReviewBench
8 entriesEvaluation state across suites, runs, and providers.
BenchSuitesAll versioned benchmark suites.
IndexMars House dialogue videovideo benchmark with 1 case.
Version 1Bench run run-mars-house-dialogue-1Stored planned run with 2 candidates.
RunBench review run-mars-house-dialogue-1Blinded candidate review for this stored run.
ReviewModelsPlanned benchmark candidates. Performance evidence is not yet available.
Planned coveragePipelinesEvaluation workflow stages and contracts.
SystemInsightsPatterns found across completed Bench runs.
AnalysisAgents
21 entriesCheck the exact media request against the current screenplay, shot, and canon.
media-canon-reviewerTurns the selected treatment and frozen scene evidence into exact scene-level writing contracts.
scene-contract-architectWrites the sole authoritative screenplay from committed-order Scene Contracts.
screenplay-drafterCreates evidence-bound recurring-character voice guidance without rewriting screenplay prose.
voice-profile-architectProposes anchored before-and-after dialogue changes while preserving the authoritative screenplay.
dialogue-editorProtects canon, world rules, long arcs, pacing, and episode objectives.
world-directorBuilds causal escalation, earned reversals, setup and payoff, and strong endings.
causality-and-structure-writerBuilds concrete objectives, conflicting incentives, subtext, tactics, and earned relationship movement.
character-and-relationship-writerMakes prior choices visibly consequential while advancing mysteries with earned surprises.
audience-consequence-and-mystery-writerDiagnoses canon, knowledge, time, resource, relationship, and obligation conflicts without rewriting.
continuity-prosecutorTests principal-character objectives, knowledge, incentives, tactics, voice, agency, and earned change.
character-advocatesTests visible Crossroads consequence, durable choice impact, and meaningful future participation.
audience-payoff-editorFlags generic dialogue, repetitive exposition, cliché reversals, and stalled mystery.
freshness-and-specificity-editorTests runtime, cast, location, asset, performance, continuation, and staging feasibility.
production-editorConverts aggregate audience behavior into compact story signals.
audience-analystProposes consequential choices that remain compatible with canon and protected arcs.
crossroads-designerTurns an approved story into beats, coverage, performance direction, and audio intent.
directorReviews deterministic shot routes for quality, continuity, and expected accepted cost.
model-routerCoordinates dialogue, voice, shot prompts, media requests, subtitles, and assembly metadata.
productionChecks continuity, identity, performance, audio, artifacts, story, and policy before review.
qaChooses one dominant reviewed treatment and records compatible borrowing, rejected ideas, and founder taste decisions.
treatment-synthesizerProviders
8 entriesLocal mock text route with treatment, continuity.
ready · Local mockLocal mock video route with face, stillness.
ready · Local mockLocal mock video route with motion, wide.
ready · Local mockLocal mock voice route with restrained, interview.
ready · Local mockLocal mock sound route with room-tone, machinery.
ready · Local mockLocal mock music route with tension, minimal.
ready · Local mockLocal mock edit route with dialogue, timing.
ready · Local mockLocal mock quality route with continuity, performance.
ready · Local mockPublic
4 entriesWorkflows
4 entriesRuns the reviewed story package through human gates and production.
MastraKeeps audience proposals behind a separate human decision gate.
MastraRuns controlled candidates through metrics, review, and a verdict.
MastraStreams Mastra state, interrupts, and reviewed activity to the browser.
AG UINative AG UI capabilities
5 entriesUNSPOOL Studio
mastraStreaming and HTTP binary responses.
NativeSnapshots, deltas, and persistent state.
NativeSubagent orchestration with delegation.
SupportedApprovals and interrupts are supported.
SupportedUNSPOOL custom capabilities
3 entriesunspool.studio.command, unspool.studio.approval
2 typesconfirmation
1 reasonrecord_writers_room_feedback, submit_writing_revision, approve_treatment, approve_script, promote_editorial_precedent, run_generation, approve_quality, submit_note, approve_regeneration, approve_final, approve_crossroads, revise_crossroads, reject_crossroads, approve_budget_overage, reset_demo, propose_production_decision, approve_production_decision, reject_production_decision
18 commandsAPI
13 entriesRead the current Studio world and production state.
ReadGET /api/studio/agentDiscover the Studio agent's native and UNSPOOL capabilities.
ReadStream Studio state, activity, and human interrupts through AG UI.
StreamReset the local Studio fixture without a stream.
WriteRead Bench suites, runs, providers, and insights.
ReadStart a controlled Bench run.
WriteApply a Bench review or workflow command.
WriteRestore the local Bench fixture.
WriteRead one Bench run and its progress.
ReadRead the completed benchmark result.
ReadRead the blinded review package.
ReadRead one blinded review asset.
ReadRead an immutable suite version.
Read