text stringlengths 3 8.33k | repo stringclasses 52
values | path stringlengths 6 141 | language stringclasses 35
values | sha stringlengths 64 64 | chunk_index int32 0 273 | n_tokens int32 1 896 |
|---|---|---|---|---|---|---|
fixtures and evidence handling contract** - `84a1b3f` (feat)
## Files Created/Modified
- `evidence/templates/gate-report.schema.json` - Closed sanitized gate report with state-specific approval and waiver constraints.
- `evidence/templates/operator-checklist.schema.json` - Human-only signed checklist bound to a revie... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-02-SUMMARY.md | Markdown | 5cdb6bafea05c8323377b51ae32470ae04375357552b626100ddbbc67057016d | 1 | 672 |
---
phase: 01-contract-and-device-proof
plan: "03"
type: execute
wave: 2
depends_on: ["01-01", "01-02"]
files_modified:
- tools/verify/compare_results.py
- tools/verify/tests/test_compare_results.py
- scripts/verify-phase-01-contracts
autonomous: true
requirements: [NFR-CONTRACT-001, NFR-COORD-001, OPS-DEVICE-001... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-03-PLAN.md | Markdown | a46c28998a8eff5ba95aa6f43cbfade06d8a184c835b4ab190322cdd119d0a37 | 0 | 896 |
`full`, and `gate` modes. Quick runs available deterministic checks. `full` is the complete automated preflight: it runs all three runtimes, mutation gates, package/schema checks, native unit/UI/build checks, release-surface inspection, and secret/diff checks, then atomically emits a valid sanitized automated-preflight... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-03-PLAN.md | Markdown | 0692e45f9bb1733664ffe01e96c4685833fcc54d54dbac4820d8ca11a2e82658 | 1 | 513 |
---
phase: 01-contract-and-device-proof
plan: "03"
subsystem: verification
tags: [fixtures, comparator, oracle-integrity, preflight, physical-gates]
requires:
- phase: 01-contract-and-device-proof
provides: Immutable fixture manifests and evidence schemas from Plans 01-01 and 01-02
- phase: 01-contract-and-dev... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-03-SUMMARY.md | Markdown | 6cb4ddf2a9ed29fc5deb3760ec647112ee125c06187b2a5d65c71344b9a238f8 | 0 | 896 |
for changed oracle bytes, missing/extra/duplicate/reordered cases, every artifact field, manifest/result digest drift, duplicate runtimes, absent runtimes, and cross-runner disagreement.
- Added eight stable modes. `quick` runs all available deterministic checks; `full` requires the complete automated surface and atomi... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-03-SUMMARY.md | Markdown | a7a2dc1e0bce4513e32bdfa670aa359e369b79331a194b54fb29753affba6e35 | 1 | 782 |
---
phase: 01-contract-and-device-proof
plan: "04"
type: execute
wave: 1
depends_on: []
files_modified:
- tools/javascript/package.json
- tools/javascript/package-lock.json
- tools/python/requirements.in
- tools/python/requirements.lock
- ios/Packages/ReRoomContracts/Package.swift
- ios/Packages/ReRoomContr... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-04-PLAN.md | Markdown | 53ee8d68da21d91ca01e1a7a37737606870d6ea01f9460ebb2e8aaef46af9651 | 0 | 896 |
=='pending' for n in e); assert 'SUS' not in e" && test ! -f tools/javascript/package-lock.json && test ! -f tools/python/requirements.lock && test ! -f ios/Packages/ReRoomContracts/Package.resolved && test ! -d node_modules && test ! -d .venv && test ! -d tools/javascrip... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-04-PLAN.md | Markdown | 75760005b7bfb912dea4fbdb61bd8862c82837c8736e83d0fe125c8b51813b71 | 1 | 896 |
.lock --swift-package ios/Packages/ReRoomContracts/Package.swift --swift-resolved ios/Packages/ReRoomContracts/Package.resolved</automated></verify>
<done>Dependency use is explicitly authorized or replaced by a verified fallback.</done>
</task>
</tasks>
<threat_model>
| Threat ID | Category | Component | Severity | ... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-04-PLAN.md | Markdown | 643eaf50238140de72107fe90ae17ce70384b87852905162fb2d4e75fc668377 | 2 | 236 |
---
phase: 01-contract-and-device-proof
plan: "04"
subsystem: tooling
tags: [supply-chain, npm, python, swiftpm, lockfiles, offline-verification]
requires:
- phase: 01-contract-and-device-proof
provides: Human-approved six-candidate dependency audit from Task 1
provides:
- Six exact human-approved dependency d... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-04-SUMMARY.md | Markdown | 26cec8cd0cf2e2fa85b0c4e3cde1b88cdc69cf8c14e00414a695e7d67bc014dc | 0 | 896 |
swift-json-schema` without changing their audited versions, revision, or artifact digests.
- Generated npm, hash-pinned Python, and SwiftPM locks containing three, two, and one approved direct roots respectively, plus ten exact audited transitives reachable from those roots.
- Added a bounded standard-library verifier ... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-04-SUMMARY.md | Markdown | ba34f47d99f1c4e3a4ba4135a3f68b07fc66b5362c62b1b381969975d0993387 | 1 | 805 |
---
phase: 01-contract-and-device-proof
plan: "05"
type: execute
wave: 3
depends_on: ["01-01", "01-03", "01-04"]
files_modified:
- tools/javascript/src/loader.mjs
- tools/javascript/src/schema-validator.mjs
- tools/javascript/src/canonical-json.mjs
- tools/javascript/src/wire-frame.mjs
- tools/javascript/src/... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-05-PLAN.md | Markdown | 68a5f252b5b495c70cabaa6e67882a3820775098e6da8505b1dd4892f43b7053 | 0 | 896 |
<files>tools/javascript/src/coordinate.mjs, tools/javascript/src/runner.mjs, tools/javascript/test/runner.test.mjs</files>
<read_first>fixtures/policies/RR-COORD-1/rev-001/manifest.json, tools/javascript/src/schema-validator.mjs, tools/javascript/src/canonical-json.mjs, fixtures/runner-result.schema.json</read_first>... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-05-PLAN.md | Markdown | c2e2ef3312b5f75c6636bfb7a59f7aefcaf4a95ff13ff0de8335bc89621fa4d9 | 1 | 426 |
---
phase: 01-contract-and-device-proof
plan: "05"
subsystem: contract-tooling
tags: [javascript, json-schema-2020-12, rfc-8785, sha256, rrfp, coordinates, fixtures]
requires:
- phase: 01-contract-and-device-proof
provides: Frozen contract/JCS/coordinate fixture corpus and RunnerResultV1 from Plans 01-01 and 01-... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-05-SUMMARY.md | Markdown | 6ce0eb5619f457797145e3aa9ea65f7bc3859f27c21e9d7ab57300b6a489def0 | 0 | 896 |
16T12:11:32Z
- **Completed:** 2026-07-16T12:27:23Z
- **Tasks:** 2
- **Files modified:** 7
## Accomplishments
- Added bounded immutable fixture loading with strict UTF-8/JSON parsing, duplicate-name and lone-surrogate rejection, safe resolved paths, exact byte/digest checks, schema registry enforcement, and Draft 2020... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-05-SUMMARY.md | Markdown | 2b31c25e261484cb1fff0aaa3e2408c14cb1f03cf3e0e2401143cc749414af62 | 1 | 896 |
Task 2 GREEN review
- **Issue:** `assert.doesNotThrow` cannot observe a rejected promise returned by the async result validator.
- **Fix:** Awaited `assert.doesNotReject` so RunnerResultV1 validation failures fail the test.
- **Files modified:** `tools/javascript/test/runner.test.mjs`
- **Verification:** All three Node... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-05-SUMMARY.md | Markdown | bb2b158be4be47d2c02cb364c6c212a69a40d203c1dbffded474e69228ab6c18 | 2 | 552 |
---
phase: 01-contract-and-device-proof
plan: "06"
type: execute
wave: 3
depends_on: ["01-01", "01-03", "01-04"]
files_modified:
- tools/python/reroom_verify/__init__.py
- tools/python/reroom_verify/loader.py
- tools/python/reroom_verify/schema_validator.py
- tools/python/reroom_verify/canonical_json.py
- too... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-06-PLAN.md | Markdown | add61c27b80a3c3de7afd6b7d58c25726238f98d082047795e46beb614b283d8 | 0 | 896 |
tools/python/reroom_verify/coordinate.py, tools/python/reroom_verify/runner.py, tools/python/tests/test_runner.py</files>
<read_first>fixtures/policies/RR-COORD-1/rev-001/manifest.json, tools/python/reroom_verify/schema_validator.py, tools/python/reroom_verify/canonical_json.py, fixtures/runner-result.schema.json</re... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-06-PLAN.md | Markdown | f0267b4fe7c8b02e05a5567b123465583e4d63e68ecea09db31ed7f1dfe5171e | 1 | 407 |
---
phase: 01-contract-and-device-proof
plan: "07"
type: execute
wave: 4
depends_on: ["01-05", "01-06"]
files_modified:
- scripts/run-reference-parity
- tools/verify/tests/test_reference_parity.py
- tools/javascript/test/parity-mutations.test.mjs
- tools/python/tests/test_parity_mutations.py
autonomous: true
re... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-07-PLAN.md | Markdown | a9a4ce96bda9956166b9cf688f9b43dbc0893e26b0f780519061f8a9ebe24e7d | 0 | 896 |
acceptance_criteria>
<verify><automated>node --test tools/javascript/test/parity-mutations.test.mjs && .venv/bin/python -m unittest tools.python.tests.test_parity_mutations tools.verify.tests.test_reference_parity -v</automated></verify>
<done>Green parity cannot be explained by shared omissions or an ineff... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-07-PLAN.md | Markdown | 1f7eea89660659dd077b7657171bbaef99358ef53c0c2a3661e1f7ddab447215 | 1 | 257 |
---
phase: 01-contract-and-device-proof
plan: "08"
type: execute
wave: 5
depends_on: ["01-01", "01-04", "01-07"]
files_modified:
- ios/Packages/ReRoomContracts/Package.swift
- ios/Packages/ReRoomContracts/Package.resolved
- ios/Packages/ReRoomContracts/Sources/ReRoomContracts/ContractValidation.swift
- ios/Pack... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-08-PLAN.md | Markdown | c0156be6f8412845e52531ed71cf7324acdf71561b1bbe0aa9092e61f6a39c97 | 0 | 896 |
/fallback); compatibility evidence records fixture, revision, environment, raw-evidence digest, metric, evaluator, and exact five-schema agreement.</acceptance_criteria>
<verify><automated>swift test --package-path ios/Packages/ReRoomContracts --filter ContractValidationTests</automated></verify>
<done>The Swift sc... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-08-PLAN.md | Markdown | c74b1896445764971788894f9221cf8a9d0bdbb6cb2fa268105b87b27fff5646 | 1 | 518 |
---
phase: 01-contract-and-device-proof
plan: "09"
type: execute
wave: 6
depends_on: ["01-08"]
files_modified:
- ios/Packages/ReRoomContracts/Sources/ReRoomContracts/CanonicalJSON.swift
- ios/Packages/ReRoomContracts/Sources/ReRoomContracts/WireFrame.swift
- ios/Packages/ReRoomContracts/Sources/ReRoomContracts/Ar... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-09-PLAN.md | Markdown | 80859e8d0521f246525d0f7c98c4d98b4c7ee5c24745f8e0501bbbf52707f77e | 0 | 896 |
payload tamper, truncation, and one-byte trailing rejection.</acceptance_criteria>
<verify><automated>swift test --package-path ios/Packages/ReRoomContracts --filter 'CanonicalJSONTests|WireFrameTests'</automated></verify>
<done>Swift matches language-neutral serialization and wire policies.</done>
</task>
<task ty... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-09-PLAN.md | Markdown | 865932fcc8a9b7bb304048be42ef2599bd00f5b60846513a011a4994366eaa61 | 1 | 630 |
---
phase: 01-contract-and-device-proof
plan: "10"
type: execute
wave: 7
depends_on: ["01-07", "01-08", "01-09"]
files_modified:
- ios/Packages/ReRoomContracts/Package.swift
- ios/Packages/ReRoomContracts/Sources/ReRoomContractRunner/main.swift
- ios/Packages/ReRoomContracts/Tests/ReRoomContractsTests/RunnerTests... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-10-PLAN.md | Markdown | f0eb3c1746b5047b3ca7012ffc6ccacb634e863257693223c36f9a94343640fe | 0 | 896 |
compatibility/coordinate-agreement.json</files>
<read_first>scripts/run-reference-parity, tools/verify/compare_results.py, docs/canonical/TEST_AND_EVALUATION_PLAN.md, fixtures/runner-result.schema.json</read_first>
<action>Run fresh isolated Swift/JavaScript/Python outputs, validate and compare all results, then wr... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-10-PLAN.md | Markdown | 127575ae63c0d224b625c5877930afb5e269553d8a0da477dbd8965425f7a370 | 1 | 386 |
---
phase: 01-contract-and-device-proof
plan: "11"
type: execute
wave: 8
depends_on: ["01-10"]
files_modified:
- ios/ReRoomDeviceProof/ReRoomDeviceProof.xcodeproj/project.pbxproj
- ios/ReRoomDeviceProof/ReRoomDeviceProof/App.swift
- ios/ReRoomDeviceProof/ReRoomDeviceProof/DeviceProofModel.swift
- ios/ReRoomDevi... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-11-PLAN.md | Markdown | 4116035e588fb7c5e2d6576841e94df68e1f4ba58ef5e1143171ac8f7290a653 | 0 | 896 |
, recorder, audio file, or audio payload. Keep iOS 26.0 labeled ASSUMED as only the declared Phase 1 proof baseline pending Plan 01-14 device evidence; do not encode a broader product minimum-OS claim or call this candidate promoted under D-05.</action>
<acceptance_criteria>Project lists only portrait orientations, b... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-11-PLAN.md | Markdown | 2e831261ba92cc658d5e716727099e81565a4483c267c61cfb51df6cb1912875 | 1 | 853 |
---
phase: 01-contract-and-device-proof
plan: "12"
type: execute
wave: 9
depends_on: ["01-11"]
files_modified:
- ios/ReRoomDeviceProof/ReRoomDeviceProof/OrientationGate.swift
- ios/ReRoomDeviceProof/ReRoomDeviceProof/WorldEpochController.swift
- ios/ReRoomDeviceProof/ReRoomDeviceProof/CaptureAttemptMachine.swift
... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-12-PLAN.md | Markdown | 8cbcb6908b754b370d0ef6797f822c0a9fdcf1dcd53be52ab5bab0acad19fc22 | 0 | 896 |
rejection, session preservation, and retry coaching.</acceptance_criteria>
<verify><automated>xcodebuild test -project ios/ReRoomDeviceProof/ReRoomDeviceProof.xcodeproj -scheme ReRoomDeviceProof -destination 'platform=iOS Simulator,name=iPhone 17' -only-testing:ReRoomDeviceProofTests/CaptureAttemptTests -only-testing... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-12-PLAN.md | Markdown | b5d666a997777b4408b6a0426f7e78e4c6cc8f716fdca0ec4bdb6a834ca40acd | 1 | 896 |
>
| Threat ID | Category | Component | Severity | Disposition | Mitigation Plan |
|---|---|---|---|---|---|
| T-01-15 | Tampering | world epoch/correction | high | mitigate | Single owner, monotonic version, directed validation, fail-closed quarantine. |
| T-01-16 | Tampering/Repudiation | capture durability | high | m... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-12-PLAN.md | Markdown | 6b46dc48256f53563ad608a1821e5d34694fb10597febad3715b4ca5ff6b99db | 2 | 220 |
---
phase: 01-contract-and-device-proof
plan: "13"
type: execute
wave: 10
depends_on: ["01-02", "01-11", "01-12"]
files_modified:
- ios/ReRoomDeviceProof/ReRoomDeviceProof.xcodeproj/project.pbxproj
- ios/ReRoomDeviceProof/ReRoomDeviceProof.xcodeproj/xcshareddata/xcschemes/ReRoomDeviceProof.xcscheme
- ios/ReRoomDe... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-13-PLAN.md | Markdown | b79ddb0c5c6fa093d6b57d4fbb71a47138b5b91c6e2b24b4858384ee576e7ebb | 0 | 896 |
iPhone 17' -only-testing:ReRoomDeviceProofTests/EvidenceExporterTests CODE_SIGNING_ALLOWED=NO && python3 tools/verify/verify_evidence.py --fixtures evidence/fixtures</automated></verify>
<done>Diagnostics generate only valid sanitized non-self-approved evidence.</done>
</task>
<task type="auto" tdd="true">
... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-13-PLAN.md | Markdown | 0a9f7cc15169ae1da64682d4d2290cab802ff1ba3f692cc485979f162a65ba09 | 1 | 845 |
---
phase: 01-contract-and-device-proof
plan: "14"
type: execute
wave: 11
depends_on: ["01-13"]
files_modified:
- evidence/device/phase-01/automated-preflight.json
- evidence/device/phase-01/gate-013-report.json
- evidence/device/phase-01/gate-013-operator-checklist.json
- evidence/device/phase-01/gate-002-repo... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-14-PLAN.md | Markdown | 5186eabccc27a1dc32e2a93e3228e17c660b23e40763c2583f5918390b5d7885 | 0 | 896 |
/microphone authorization state tests and rename-before-journal crash test), release-surface, secret, diff, archive, signing-entitlement, and evidence-template checks. On success, `full` atomically emits the sanitized preflight with candidate build digest and environment; validate that output immediately. Leave both ph... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-14-PLAN.md | Markdown | 29ec546009a8e6782fdf812b7bbd000152e71fbed382191041c4c0762eb2974a | 1 | 896 |
automated-preflight.json, evidence/device/phase-01/gate-013-report.json, docs/canonical/RISK_AND_KILL_GATES.md, docs/canonical/TEST_AND_EVALUATION_PLAN.md, fixtures/policies/RR-COORD-1/rev-001/manifest.json</read_first>
<action>Per D-04, D-13, D-14, D-15, and D-16, the human operator runs the versioned checkerboard/g... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-14-PLAN.md | Markdown | 8d6ad090edea397d0e5e4df3d093d34a9f06108afd2296f9e4b0f15b7ecc141e | 2 | 712 |
# Phase 1: Contract and Device Proof - Context
**Gathered:** 2026-07-16
**Status:** Ready for planning
<domain>
## Phase Boundary
Establish one executable contract and coordinate vocabulary across Swift, JavaScript, and Python, and prove the signed portrait-only base-iPhone path without rear-LiDAR semantics. This ph... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-CONTEXT.md | Markdown | 5f707076108b79589c0123ab53eb5c61384ed6295f59b61497b0dfa15bbd8643 | 0 | 896 |
-DEVICE-001` acceptance evidence.
- `docs/canonical/PRD.md` — normative requirement statements and fallbacks.
- `docs/canonical/DEVELOPMENT_STRATEGY.md` — S0 sequencing, evidence, and recovery boundary.
### Coordinates and physical-device proof
- `docs/adr/ADR-002-native-iphone-and-web-split.md` — native Mode A bounda... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-CONTEXT.md | Markdown | 0447999e64331fac7a17e0ec149c86daa75064316de7e88210dd275c74bac686 | 1 | 743 |
# Phase 1: Contract and Device Proof - Pattern Map
**Mapped:** 2026-07-16
**Proposed file surfaces classified:** 15
**Executable analogs found:** 0 / 15
**Repository state:** Greenfield product implementation; canonical schemas and planning artifacts are authority, not source-code analogs.
## Mapping Result
Th... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-PATTERNS.md | Markdown | 75ce1263841ff5b9e1f5227ca7b6898c6f8418084d6c591f348c1fc1ebcb83af | 0 | 896 |
|
## Pattern Assignments
### Language-neutral fixture corpus
**Applies to:**
- `fixtures/contracts/1.0.0/rev-001/**`
- `fixtures/policies/RR-COORD-1/rev-001/**`
- `fixtures/policies/RR-JCS-SHA256-1/rev-001/**`
**Executable analog:** None.
**Use this researched pattern:** `01-RESEARCH.md:123-141` defines one immut... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-PATTERNS.md | Markdown | f13a103c78b97143f3a11cdd17eda0074f6f432bb89fbd815dc610151ef9bce4 | 1 | 896 |
docs/contracts/README.md:3-11` identifies the five frozen schema authorities; `docs/contracts/README.md:15` requires closed exact-version behavior; `docs/contracts/README.md:33` fixes the dialect.
**Planner constraint:** Pin exact candidate versions and artifact integrity only after the research-mandated package verif... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-PATTERNS.md | Markdown | 409adda90af332d21685ad9797cc4d3e2ff7da9dc8b48ea1e76f84de3eaf9ae9 | 2 | 896 |
logs, signing details, device identifiers, user paths, and room data remain outside Git behind opaque IDs and SHA-256 digests.
**Authority anchors, not copy sources:** `01-RESEARCH.md:309-326` maps requirements to automated and physical evidence and states that device/human evidence cannot be delegated or fabricated. ... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-PATTERNS.md | Markdown | 8365e9c55c14243efd50300510f0c6017ce10610510b1222dd7df4297c452510 | 3 | 896 |
Guardrails
1. Treat every Phase 1 implementation file as greenfield and cite `01-RESEARCH.md` plus canonical authority in each plan action; do not claim a repository code pattern that does not exist.
2. Create only Phase 1-owned directories from `01-RESEARCH.md:168-184`; avoid speculative workspace, web, gateway, prov... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-PATTERNS.md | Markdown | 9bbb6b57fa60e1b928dcf291232a38aa95ede06769b6f243249430826a011061 | 4 | 289 |
# Phase 1: Contract and Device Proof — Research
**Researched:** 2026-07-16
**Status:** Ready for planning
**Confidence:** MEDIUM — canonical requirements are verified locally; current dependency/API findings are source-backed, but the Swift JSON Schema candidate and all physical-device results still require Phase ... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | 773e544229f47d4791a6d8f1f883a6112a1bf68e30a31110e6651ab81c23b193 | 0 | 896 |
and RFC 8785 digest cases. JavaScript and Python have mature current packages; no authoritative, production-proven native Swift RFC 8785 implementation was found. Plan a small Swift JCS kernel, restricted to the exact canonical profile, and accept it only against checked-in RFC/cross-language vectors. This is the one j... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | d1173205418e4686195350e303ef83f4e0982ae27390a743065f95c561cd87eb | 1 | 896 |
decision when authorities conflict. [VERIFIED: `AGENTS.md`]
- Preserve the two byte-exact archived source files; audit records do not override current authority. [VERIFIED: `AGENTS.md`]
- Use the global GSD 1.7 installation and committed `.planning/` surface. Do not create a repository-local GSD install or generated ag... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | 34655a18f322042d6d3d9251ad2aecce18575c5c5e8b2ceb5c53e66d20941fba | 2 | 896 |
, local references, formats, and invalid fixtures pass. [CITED: https://github.com/ajevans99/swift-json-schema/tree/0.13.1] |
| JavaScript validation | `ajv` 8.20.0 + `ajv-formats` 3.0.1 | Current official packages; select the Draft 2020-12 entry point explicitly and compile frozen schemas once. [CITED: https://ajv.js.... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | 6eade9fab5cbee01d2166100d85f1450de2f7f1e3bdc34e11a3e0a1db5523ce5 | 3 | 896 |
runner must emit the same small result envelope, for example `case_id`, `verdict`, `rejection_class`, and `output_digest`. A generator may propose the next revision but cannot overwrite expected results. Revisions are append-only. [VERIFIED: D-09–D-12]
### Pattern 2: Pure kernel beside an imperative device shell
Keep... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | 88bb164bd0475df328923391d14b1ac564aa3ea8ce0477f4b275c5fe79030314 | 4 | 896 |
, gate status, or evidence lifecycle. [VERIFIED: canonical contracts/glossary]
- Do not regenerate expected bytes/digests during ordinary tests. [VERIFIED: D-10]
- Do not use an image re-encode to obtain digest bytes when the contract names the captured payload bytes. [VERIFIED: CON-001]
The exception is a narrow Swif... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | febc1163677692afa32eeacb085ff269e2a3793cae27c64fd638565fabcfa45e | 5 | 896 |
``python
from jsonschema import Draft202012Validator
import rfc8785
Draft202012Validator.check_schema(schema)
errors = sorted(Draft202012Validator(schema).iter_errors(instance),
key=lambda error: list(error.absolute_path))
canonical_bytes = rfc8785.dumps(instance)
```
[CITED: https://python-jsonschema... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | eddd465fd3e7ba4301592c971ff11d41f6b0073e3557a3ac0704fb87ff7c6cec | 6 | 896 |
from the runtime path and activates the frozen ReRoom-profile validator; schemas are never weakened. The adopted result and license/revision evidence are recorded in the research ledger.
3. **GATE-013 diagnostic durability boundary:** Plan 01-12 implements one local diagnostic capture through CON-001's exposed `network... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | cfa1cff13885cfca1005caa51dd7efb108fa03583dab3046994912cde1838a76 | 7 | 896 |
human checklist signature |
Suggested commands are planning targets, not current evidence:
```text
xcodebuild test -project native/ReRoomDeviceProof.xcodeproj -scheme ReRoomDeviceProof -destination 'platform=iOS Simulator,name=iPhone 17'
node --test tools/javascript/test/*.test.mjs
python3 -m unittest discover -s too... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | f614e906c3ee5c3047f0cecb0b61615d20131cde54ee15f03ce4b7738e0b2e17 | 8 | 896 |
-testing/)
- [Apple CryptoKit SHA256](https://developer.apple.com/documentation/cryptokit/sha256)
- [Ajv JSON Schema versions](https://ajv.js.org/json-schema.html)
- [python-jsonschema validation API](https://python-jsonschema.readthedocs.io/en/stable/validate/)
- [Swift JSON Schema 0.13.1](https://github.com/ajevans99... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-RESEARCH.md | Markdown | 72f43144ca7dc12f2a32812589632b8db0c6e08bb2515e792f7ac2608ff7e5eb | 9 | 258 |
---
phase: 01
slug: contract-and-device-proof
status: draft
nyquist_compliant: false
wave_0_complete: false
created: 2026-07-16
---
# Phase 01 — Validation Strategy
> Final-plan validation contract. Commands remain planning targets until their owning task creates them.
## Test Infrastructure
| Property | Value |
|-... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-VALIDATION.md | Markdown | 04636772d946f1fea31d3778b522f5a0ebcb4d1d2124461c7030e43717c8fe05 | 0 | 896 |
','ajv-formats','canonicalize','jsonschema','rfc8785','swift-json-schema'}" && test ! -f tools/javascript/package-lock.json && test ! -f tools/python/requirements.lock && test ! -f ios/Packages/ReRoomContracts/Package.resolved && test ! -d node_modules && test ! -d .venv && test ! -d tools/javascript/node_modules && te... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-VALIDATION.md | Markdown | 2672fb54d67edf541a6c4c9a76bff6c818522d71944957cc2794efacde20f2a2 | 1 | 896 |
m unittest tools.python.tests.test_parity_mutations tools.verify.tests.test_reference_parity -v` | ❌ W0 | ⬜ pending |
| 01-08-01 | 01-08 | 5 | NFR-CONTRACT-001 | T-01-11 | Swift validator supports used keywords and agrees | compatibility | `swift test --package-path ios/Packages/ReRoomContracts --filter ContractValidat... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-VALIDATION.md | Markdown | 8471c19d8668b6759211216040281523b9bcbf50404021926adc2fbe9e1c0471 | 2 | 896 |
/ NFR-CONTRACT-001 | T-01-17/18 | Export is sanitized and cannot self-approve | security/unit | `xcodebuild test -project ios/ReRoomDeviceProof/ReRoomDeviceProof.xcodeproj -scheme ReRoomDeviceProof -destination 'platform=iOS Simulator,name=iPhone 17' -only-testing:ReRoomDeviceProofTests/EvidenceExporterTests CODE_SIGNI... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-VALIDATION.md | Markdown | 0f44b21f22a77e8e18d48061b3c15cafc34431f62f9adda6e97426ab0aa29fed | 3 | 896 |
evidence and sign bound checklist. RED validates as failure evidence but does not complete the gate. Promote only on GREEN; RED/pending remains unpromoted and blocks dependent mobile work. |
| Portrait geometry, landscape rejection/coaching, and explicit physical world reset with epoch advance plus correction or quaran... | openai-build-week | .planning/phases/01-contract-and-device-proof/01-VALIDATION.md | Markdown | 1364e35810abed2777bc2f8ca88241fd3715b04fb4bcded621ee0f2777558663 | 4 | 287 |
# ADR-001: Product Modes, Exact P0 Scope, and B1 Isolation
Status: Accepted
Date: 2026-07-14
## Context
The archived inputs mix an exact four-operation promise with a kill rule that could silently demote empty removal. They also describe enough Mode B1 detail that a planning agent could put photoreal refinement on... | openai-build-week | docs/adr/ADR-001-product-modes-and-p0-scope.md | Markdown | 4541d2e90366bc7ce89b4229e235d5e046e34c676766dfeda5d518aaf961477a | 0 | 579 |
# ADR-002: Native iPhone Hero and Separate Web Client
Status: Accepted
Date: 2026-07-14
## Context
Mode A needs direct ARKit frames, transforms, recording, native rendering, and predictable thermal behavior. Mode B0 needs upload, replay, debugging, sharing, and cross-device inspection.
## Project constraints
- T... | openai-build-week | docs/adr/ADR-002-native-iphone-and-web-split.md | Markdown | 7a04c0500633db14e9f0de7bbc3f152e6d258944aded7e4cdee997d6eaf4e34d | 0 | 548 |
# ADR-003: ARKit World Authority and RR-COORD-1
Status: Accepted
Date: 2026-07-14
## Context
Incorrect orientation, crop, intrinsics, pose direction, matrix layout, timestamp precision, or world-reset handling can make every spatial subsystem appear plausible while being wrong.
## Project constraints
- The base ... | openai-build-week | docs/adr/ADR-003-arkit-authority-and-coordinates.md | Markdown | 758c24a298d48c1515f51cadf2b6bb69bc23698eb6edf7d2493486dcd3461e4f | 0 | 687 |
# ADR-004: Record-First Capture, Transport, and Replay
Status: Accepted
Date: 2026-07-14
## Context
The archived plan says frames are enqueued before upload, while the PRD permits recording “before or while” upload. Neither wording defines a crash-safe boundary or which accepted sequence replay must reproduce.
##... | openai-build-week | docs/adr/ADR-004-atomic-capture-and-record-first-replay.md | Markdown | b9d176ea2a01a66400aed756c5f52a4405518bf3685cdea444de7023efb0b89b | 0 | 666 |
# ADR-005: RealityKit-First Camera-Feed Compositor
Status: Provisional
Date: 2026-07-14
## Context
Mode A must cover a real target with spatial reveal geometry and virtual assets while the camera remains the photoreal background. RealityKit shortens the asset/anchor path, but documentation alone cannot establish a... | openai-build-week | docs/adr/ADR-005-realitykit-first-compositor.md | Markdown | 6ea295ddd3d3c811380f81409bb78a2715641d5d3b0b646e93c20719c70f28b8 | 0 | 680 |
# ADR-006: Separate Fast Interaction and Dense Geometry Tracks
Status: Accepted
Date: 2026-07-14
## Context
Making the first edit wait for learned depth, metric alignment, fusion, semantic lifting, and mesh extraction creates the longest and least debuggable failure chain. ARKit planes plus calibrated target masks... | openai-build-week | docs/adr/ADR-006-fast-and-dense-geometry-tracks.md | Markdown | 8bc86d21777403dd66f74368cfa692480c2771508cbd18461afeaf1ba25fd5e4 | 0 | 558 |
# ADR-007: Versioned Segmentation and Depth Providers
Status: Provisional
Date: 2026-07-14
## Context
The archived plan freezes SAM 3.1 and initially favors DA3Metric-Large. Both are current models, but model recency does not establish one-week suitability. SAM 3.1 is large, access-gated, and under a custom licens... | openai-build-week | docs/adr/ADR-007-segmentation-and-depth-providers.md | Markdown | d6446853123f564b414406cb1d0f386b80436ea8dd4b30862f01597679bd64dc | 0 | 896 |
-update p95 at most 450 ms, no two-minute queue growth, and no OOM. Timebox: four hours before any dense integration; failure selects no-dense live and may retain the provider for offline research only. LingBot receives no P0 time unless `GATE-008` is already green.
## Requirements and contracts affected
`FR-TARGET-0... | openai-build-week | docs/adr/ADR-007-segmentation-and-depth-providers.md | Markdown | 6aa8df5a1aa3e1b070da20c8def28023e7e4987a95923722b2c2868cb9935dfb | 1 | 132 |
# ADR-008: Canonical Scene Identity and Capability Readiness
Status: Accepted
Date: 2026-07-14
## Context
Renderer indices, provider buffer positions, one global `editable` flag, and server-only artifact readiness would create identity drift and premature user promises.
## Project constraints
- Mode A and B0 mus... | openai-build-week | docs/adr/ADR-008-scene-identity-and-readiness.md | Markdown | 9a07cf190ac7eb8fa84b52c9cfdca22b979454da56fba3dc9a93518d195422ba | 0 | 459 |
# ADR-009: Multi-Surface Reveal and Supported View Envelope
Status: Provisional
Date: 2026-07-14
## Context
Empty removal must cover camera pixels belonging to a real object using background evidence that may span floor, wall, and trim. A single plane or unconstrained inpaint is unlikely to remain credible while t... | openai-build-week | docs/adr/ADR-009-multi-surface-reveal.md | Markdown | 9f971edc34c18f323e9f384ca53da989d0ad894de107979162167b0af51c110b | 0 | 716 |
# ADR-010: Curated Asset Contract and Prevalidated Derivatives
Status: Accepted
Date: 2026-07-14
## Context
Placement and replacement depend on geometry, units, origin, collision, mobile delivery, visual cover, and licensing. Runtime conversion or an unbounded catalog would create hidden critical-path work.
## Pr... | openai-build-week | docs/adr/ADR-010-asset-contract.md | Markdown | 08c16c5fc6a65c93ede794590a276501d47acf6cf7ff6b253d36728e81c91e6a | 0 | 506 |
# ADR-011: Agent Intent and Deterministic System Boundary
Status: Accepted
Date: 2026-07-14
## Context
Typed/tap semantic proposals are part of the deterministic P0 path. Voice and GPT may make the product more legible and design-aware without becoming authorities for geometry, scene state, credentials, or mutatio... | openai-build-week | docs/adr/ADR-011-agent-and-deterministic-boundary.md | Markdown | f9e7372abd74fc32f3f77c3cc39397ce9add886b7ee8fedaf565e05eaf757ba2 | 0 | 730 |
# ADR-012: Revisioned Transactions and Offline Restore
Status: Accepted
Date: 2026-07-14
## Context
Preview, retry, reconnect, and local restoration are unsafe without a precise distinction between canonical transaction state and client synchronization state. Mutating a committed transaction into “undone” would de... | openai-build-week | docs/adr/ADR-012-transaction-and-offline-restore.md | Markdown | 40706301389b84be8e3c0b92b5f81b178b4e137fe1c004f38c4369447ec345fb | 0 | 813 |
# ADR-013: Guaranteed Mode B0 Minimum
Status: Accepted
Date: 2026-07-14
## Context
The archived inputs make B0 guaranteed but also couple it to learned ordinary-video reconstruction, TSDF output, and a warm GPU. That makes the promised fallback depend on the least mature providers.
## Project constraints
- Recor... | openai-build-week | docs/adr/ADR-013-mode-b0-guarantee.md | Markdown | 08b310b7e28516bf366382792634ca1ab1f64a44ae12c1b9113b74b025df293a | 0 | 566 |
# ADR-014: Minimal Service Topology and Hardware Tiers
Status: Accepted
Date: 2026-07-14
## Context
The archived plan assigns five owners, several always-on GPU services, multiple incompatible model environments, and a specific warm-cloud posture. Two developers cannot debug all of those simultaneously, and a spec... | openai-build-week | docs/adr/ADR-014-service-topology-and-hardware-tiers.md | Markdown | ec065727bfc65b48efa105ffd695c8d3df366756cd52c1d60edce7876b869b26 | 0 | 726 |
# ReRoom Source Archive
These files are immutable historical inputs. They are excluded from the GSD ingest manifest and are not current implementation authority.
Archive date: 2026-07-13
| Original path | Archived path | Original title | Original status claim | SHA-256 before | SHA-256 after |
|---|---|---|---|---|-... | openai-build-week | docs/archive/README.md | Markdown | 68b2ad0a094fa453f77ad4e4cf645dd563821985e1d119db4d4a74309f2868ff | 0 | 298 |
# ReRoom Master Technical Plan v3.2
**Status:** Architecture freeze for OpenAI Build Week
**Date:** 13 July 2026
**Owners:** Five-person JKU Linz team
**Supersedes:** ReRoom v3 Complete Consolidated Specification and all earlier architecture notes
**Living-document rule:** Changes to a locked decision or exter... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 0933cb5361eabbf8ed584a3b6588958f6d9de510ac535c04933c866b29a75ee5 | 0 | 896 |
.** Mode B1 is forbidden from consuming critical-path time.
### 0.3 Provenance markers
- `[TEAM]` — explicit team ruling, including answers 1A/2A and confirmed native readiness/hero scene.
- `[V3]` — retained from the Fable v3 consolidated specification.
- `[SOL]` — introduced or materially corrected in the GPT-5.6 S... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 7971491bf8e9fec863a26bdb13af288dcccf05d313fe046b8190cf15a2526043 | 1 | 896 |
--> ATLAS --> FILL --> KIT
PLANES --> TX
LIFT --> TX
KIT --> TX
TX -->|revisioned EditKit WS| CACHE
VOI <-->|WebRTC| RT
RT --> AG
AG <--> TX
CAT <--> AG
TSDF --> TWIN
SES <--> DASH
```
### 2.1 Local loop
The local loop is independent of the network and targets 60 FPS:
1. ARKit... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | c0ad8c5ad84db9832e5f981ea6d5d90a664b72377e6a5c995ba192837bb392fa | 2 | 896 |
, LingBot owns the camera trajectory and geometry. Mode B0 shares scene, transaction, catalog, semantics, and reveal services with Mode A.
### 3.3 Mode B1 — Photoreal polish
B1 is disabled until the D7 P0 acceptance gate passes.
Pipeline:
```text
immutable session snapshot
→ select 80–160 sharp, diverse keyframes
→... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 4a1da2086d895e232eeb0375a0e33de42976cfa64f8cf5eef87259278b98f3ba | 3 | 896 |
3.1 point/box-seeded video tracking | Persistent target identity; no whole-room discovery required | `[SOL]` |
| D11 | Diminished reality | Multi-surface reveal proxies; camera background remains untouched | Simpler and more physically coherent than arbitrary pixel replacement | `[V3]`+`[SOL]` |
| D12 | Renderer | Real... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 7ea27063c7d3451042052069516ed4042e196fd70e3091b2be65afb9e4a4b2cc | 4 | 896 |
--:|---|
| 0 | 4 | ASCII magic `RRFP` |
| 4 | 2 | `protocol_version` (`uint16`) |
| 6 | 2 | flags (`uint16`) |
| 8 | 4 | JSON metadata length (`uint32`) |
| 12 | 4 | image payload length (`uint32`) |
| 16 | 8 | `frame_id` (`uint64`) |
Header is followed by UTF-8 JSON metadata and then the image payload. P0 payload is ... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 7702520665cbe8d586fbc99be60510502b598662b778c0ecb77d194aa4e01fd1 | 5 | 896 |
revisions are monotonic per observed anchor.
---
## 9. `.rrcap` contract
While recording, `.rrcap` is a directory. On clean finish it may be zipped without changing its internal paths.
```text
session.rrcap/
manifest.json
frames/
000000000001.jpg
000000000002.jpg
keyframes/
000000000842.jpg
fram... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 8a003f8d5e322a9696de13449a821a17d4e8b86f1e9eed12688de4f83364d9ac | 6 | 896 |
# 11.1 Object artifacts
```json
{
"object_id": "chair_04",
"artifact_revision": 3,
"mask_volume": {
"type": "voxel_rle",
"world_from_volume": ["16 float32"],
"voxel_size_m": 0.025,
"dimensions": [48, 39, 41],
"payload_url": "signed://objects/chair_04/mask-volume-v3.bin",
"conservative_dil... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 85ec3c44f900bba8fc3dcf1452413bfb40818319b084bdb3278103c02243f1cd | 7 | 896 |
```json
{
"asset_id": "chair_red_02",
"category": "armchair",
"style": ["modern", "warm"],
"color": "red",
"dimensions_m": [0.79, 0.91, 0.82],
"origin": "floor_contact_center",
"forward_axis": "-Z",
"delivery": "bundled_p0",
"local_bundle_key": "chair_red_02.usdz",
"usdz_url": "signed://assets/chair... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | a171e4748f18d29ce799e796dfb69045eee933243a65c6016bd10fe2a473f665 | 8 | 896 |
v3",
"surface_mesh_ref": "objects/chair_04/surface-v2",
"obb": {
"center_world": [1.42, 0.48, -2.18],
"extent_m": [0.82, 0.96, 0.78],
"yaw_rad": 1.43
},
"support_surface_ids": ["floor_01"],
"reveal_bundle_ref": "reveal_chair_04_v5",
"state": {
"visibil... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | c04e1f338f5d7ffa022985fb93936ce21746c0b25203d5b92a67f75347749886 | 9 | 896 |
D2 exit:** deterministic `.rrcap`, coordinate projection test, stable floor plane, frame upload, and replay on the physical iPhone 17.
### 14.1 Technology
- SwiftUI application shell.
- `ARSession` with `ARWorldTrackingConfiguration`.
- RealityKit view for the primary compositor; Metal fallback owned by P3.
- Native ... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 17e1717fcd6ddeb698bb29e90c6a46571a1a99a3ae9706bf181dea0d2d57b01f | 10 | 896 |
may render a projected mask volume into stencil and composite reveal textures explicitly. Do not spend more than the gate window trying to force an unsuitable RealityKit path.
### 15.4 Revision behavior
- Artifact revisions are downloaded before activation.
- Crossfade reveal texture revisions over 150–250 ms.
- At 3... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | cd127f93fe11fa542542cd291a22354a02fb8e8536f7c0646d7c126063d2ddd8 | 11 | 896 |
;
- floor-plane RMSE;
- three taped-distance errors;
- wall thickness in TSDF;
- edge quality around the hero object;
- temporal depth flicker;
- loop-return drift;
- rejected-frame rate;
- qualitative side-by-side mesh.
Choose one ARKit depth provider in a one-page ADR. Do not build a general benchmark platform.
###... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | e569ca2747b9e50f729224666890c0e6d8526bdba08369ad08a1a05787c102a9 | 12 | 896 |
**Owner:** P4
**D5 exit:** hero reveal bundle passes coverage and qualitative seam gate from a half-circle walk.
### 20.1 Plane atlas
Each stable plane defines:
- world origin;
- orthonormal U/V basis;
- metric atlas bounds;
- color atlas;
- observed-weight map;
- source-frame/provenance map;
- synthesized mask.
... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 1ba89892b1b696acebe40493fe65e2c3edc21a59cd8b17e77c56c12b741db477 | 13 | 896 |
Preview local reveal + asset.
- Commit as one transaction.
- Undo restores target visibility, hides reveal and replacement, and needs no server call.
---
## 22. Gateway, GPT-5.6 Sol, and voice
**Owner:** P5
**D6 exit:** push-to-talk hero command → Sol decision → deterministic validation → phone commit; typed path ... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 3634e380dfa2c69c13fd67f3bb27eec11d8714969f2cdb54025a5778c5ff4a3d | 14 | 896 |
v1.
---
## 24. Mode B1 polish worker
**Owner:** unassigned until P0 is green; then P2/P5
**Start gate:** all D7 release criteria pass.
- Immutable input snapshot.
- Select 80–160 keyframes by sharpness, coverage gain, baseline, loop value, and semantic relevance.
- Use only the MapAnything Apache-compatible model... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | b32976275bad257b217e074a97e886435348c96c71a58d8e2adb64a1863652f6 | 15 | 896 |
, and either explicit session affinity or externalized recoverable state. Cold scale-to-zero is suitable for B0/B1 batch jobs, not for the hero Mode A session.
---
## 26. Transport
### 26.1 Data planes
- Control REST: iOS/web ↔ gateway.
- Frame ingest binary WebSocket: iOS → gateway ingest/router; the gateway forwa... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | c84981a544d9332756f914ab04a0df7f1565ed2c4601f77271404bf8cc7d0ace | 16 | 896 |
target-to-capability latency;
- mask IoU on approximately 20 hand-annotated frames;
- `cover_score` and reveal-view coverage;
- synthesized-texel fraction versus observed viewpoints;
- floor RMSE and taped-distance errors;
- anchor drift over a three-minute fixed-marker test;
- provider bake-off results;
- golden-path ... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 1dd27788a5362b50be81c43abb3378d6d794185192162cc5d982e83a23aac5b3 | 17 | 896 |
.
- Projection/intrinsics tests.
- Stable floor plane and taped metric check.
- Target seed → SAM track → conservative volume/OBB.
- Canned multi-surface reveal + USDZ replacement on phone.
- Written gate decision.
### 15 July — D3: dense geometry and placement
- Provider ADR.
- Scale-aligned TSDF and metric threshol... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | d260407dbab29c50dd1915abe0a357dea617a951fa44485e8bfccc31ecb710ec | 18 | 896 |
3. place one small asset and undo it locally;
4. when the D5 gate passed, remove the hero object and undo so the original real object returns;
5. replace through a Sol-mediated command with deterministic fit evidence;
6. walk a half-circle with stable anchoring and coherent reveal/occlusion;
7. demonstrate network-loss... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 4efda062c04a03f2f184d937cf7d2ec133f52f9983a1e40c8b897bb460643764 | 19 | 896 |
definition of done;
- prohibited scope.
### 35.3 Parallel work rules
- Contracts merge before implementation branches depend on them.
- One owner per contract; reviewers from every consuming language.
- Agents may generate adapters from JSON Schema, but generated output is committed and tested.
- No agent silently ch... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 6d6499989b4482562c8c7f55d41f143ff68aa39fad8d439648c7219458920ff8 | 20 | 896 |
/specs/`
4. Depth Anything 3 official repository: `https://github.com/ByteDance-Seed/Depth-Anything-3`
5. LingBot-Map official repository: `https://github.com/robbyant/lingbot-map`
6. SAM 3 official repository: `https://github.com/facebookresearch/sam3`
7. SAM 3.1 release notes: `https://github.com/facebookresearch/sam... | openai-build-week | docs/archive/source/ReRoom_Master_Technical_Plan_v3.2.md | Markdown | 8725a74c539faa85454a1b07e7e9227cb6091b66ec11ab0a5dff3d7d7417532f | 21 | 750 |
# ReRoom Product Requirements Document v1.0
**Status:** P0 product freeze for OpenAI Build Week
**Date:** 13 July 2026
**Working title:** ReRoom
**Technical authority:** `ReRoom_Master_Technical_Plan_v3.2.md`
**Release target:** Build Week submission by 21 July 2026
---
## 1. Product summary
ReRoom lets a u... | openai-build-week | docs/archive/source/ReRoom_PRD_v1.0.md | Markdown | 62a3c506098df0417b1c7e30a5b84e5c0743c98f2de9a76848d81b91da26289a | 0 | 896 |
## Beat 1 — Start and look around
**User action:** opens ReRoom and points the phone naturally around the hero object for 10–30 seconds.
**What the user sees:**
- live camera at 60 FPS;
- a small tracking/coverage indicator;
- concise coaching such as “show me a little more floor” or “move to the left side of the ch... | openai-build-week | docs/archive/source/ReRoom_PRD_v1.0.md | Markdown | 2e3828d0ba9ad6b0a9efc5667dc49bae7fd296f2e3770814b551a0e92a929688 | 1 | 896 |
— Photoreal twin
Optional after release readiness. B1 may create a polished splat scene for the web but must not delay or destabilize A or B0.
---
## 8. Functional requirements
### FR-1 — Session and consent
- User must explicitly grant camera/microphone access and consent to cloud upload.
- UI must display record... | openai-build-week | docs/archive/source/ReRoom_PRD_v1.0.md | Markdown | 0864ea9c3a54117729cb9f0ad97cb8b9de7fd1c459d820c0273dae9765305242 | 2 | 896 |
.
- Use the same sparse high-resolution keyframes and canonical scene IDs as Mode A.
- Send typed commands through the same tools.
- Inspect scene/artifact revisions in debug mode.
### FR-10 — Resilience
- The live backend runs on a warm, session-sticky worker; cold queue-style serverless execution is not allowed for... | openai-build-week | docs/archive/source/ReRoom_PRD_v1.0.md | Markdown | fb6445d813189a590c7dc2d78b9506f71aa9a942fa860e2434879874e2f0db1f | 3 | 896 |
hashes are recorded;
10. primary and over-the-shoulder demo footage are captured before final submission day.
---
## 12. Risks in product language
The main risk is not raw model speed; it is visual credibility at the boundary between the real camera and the reconstructed background. ReRoom protects the demo by stagi... | openai-build-week | docs/archive/source/ReRoom_PRD_v1.0.md | Markdown | 618e68441ecc7cf25099fe0786eabaf1f543ddb44b117835ddcc2300b4aa7c7e | 4 | 817 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.