← E2T System V1 portal

03 — MAP-AI

Validate physical proposals and nominate semantic relationships using indexed evidence only.

ACTIVE Built and verified

12 tool files copied and hash-verified against the original.

ML proposal-only Page model 2 archived step 03 of 26

Where this output sits

Source evidenceCandidate evidenceHuman decisionAudit acceptanceReleaseProduction

Source evidence, candidate evidence, human decision, audit acceptance, release and production are six separate things. This step produces the highlighted one and nothing further.

Scope

A planner call routes page ranges; a detail call returns a compact typed schema of groups, titles, furniture, table flags, sequence and continuation. Every reference must point at an existing region or token id. Responses are validated against the schema, and anything unparseable or unreferenced is rejected rather than repaired by guesswork.

This role is allowed to

How it works

Inputs
Original page imageIndexed MAP-GEO regionsIndexed CAP lines/tokens
ML proposal-only Constrained structured vision AI with deterministic schema validation.
  1. Planner call routes the issue into page ranges
  2. Detail call returns compact schema v65 — groups, titles, furniture, table flags, sequence, continuation
  3. Deterministic schema validation: every reference must point at an existing region or token id
  4. Accept, reject or HOLD — unparseable or unreferenced output is rejected, never repaired by guesswork
Must pass
  • All references exist
  • No new text/geometry
  • Every decision has evidence and confidence
passes →
Outputs
Accept/reject/HOLD decisions on regionsMerge/split/refine requests by region IDFurniture and relationship proposals
hands off to MAP-REFINE · ROLE/TITLE · ORDER · TAG
fails →
HOLD
AI_SCHEMA_INVALIDUNKNOWN_REFERENCEFREE_TEXT_VIOLATIONMAP_AI_AMBIGUOUS

Nothing continues on a failed gate. Uncertainty becomes an explicit HOLD, and no later step may read an unanswered item as an accepted one.

Non-scope — what this role does NOT own

Explicitly forbidden

Dependencies and position

Starts
After MAP-GEO and CAP aliases are frozen.
Previous step
02 MAP-GEO / MAP-M
Next step
04 MAP-REFINE
Hands off to
MAP-REFINE, ROLE/TITLE, ORDER, TAG

Exact inputs

Exact outputs

Performer and AI/ML boundary

Performer
Constrained structured vision AI with deterministic schema validation.
Class
ML proposal-only
AI boundary
Yes, proposal-only.

Training information

Not trained by this project. Gemini 3.8 Flash, pinned, prompted under response schema v65. It is constrained to proposal-only: it may reference existing ids and propose merges, splits, roles and relations, but may not return free OCR text or invent coordinates.

Deterministic validation and acceptance gates

HOLD and failure behaviour

Failure codes this step may emit:

Uncertainty becomes an explicit HOLD. Omission never converts uncertainty into acceptance, and no downstream step may treat an unanswered item as an accepted one.

Downstream handoff

MAP-REFINE, ROLE/TITLE, ORDER, TAG

Active tool

Status
Active tool
Run / inspect
python tools/run_autonomous_parallel.py --issue <id>
Input
schemas/planner-response.schema.json + indexed CAP/MAP references
Output
schemas/detail-response.schema.json (compact schema v65)

Dependencies

Why this is the active version

V114 is the highest-numbered package in the Flash MAP pipeline chain (V47 to V50-V100 to V108-V114) that contains executable code, and each README explicitly supersedes the prior: V111 added resume, V112 fixed the token budget, V113 added tier amendment, V114 fixed candidate/feedback pairing. V114 is also the package captured as 02_SYSTEM_SNAPSHOT inside the sealed SUCCESS_001 issue-4547 trial — independent confirmation that this is the version which produced the live output.

Copied files — source receipt

FileOriginal SHA-256BytesCopy
marsoom_map.py 76b6eec65d1726e9… 44,332 verified
run_autonomous_parallel.py b6c01cb0dd4bc2fe… 46,107 verified
continue_partial_run.py d9e8bb91ff3f1bdf… 10,787 verified
resume_paused_continuation.py 8075fe6fedb6e7ab… 10,330 verified
prompts/planner.txt e184e4e1a60d3442… 1,113 verified
prompts/planner_attachment.txt 88b1c04f2cd027f9… 1,217 verified
prompts/detail.txt bf87efab3d5ab7aa… 2,083 verified
prompts/detail_attachment.txt 71221ce54a6b0921… 1,994 verified
schemas/planner-response.schema.json c9f46ce6d83cdd5d… 1,065 verified
schemas/detail-response.schema.json 4b73e8c39581ca16… 2,721 verified
config/ontology.json 15b0da5dcb1df276… 1,419 verified
config/confidence_rules.json 534a0fd8779dbe66… 1,278 verified

12 file(s), all hash-verified against the original. Full detail in tool/SOURCE_RECEIPT.json. Copy-only: the historical source is never modified.

Archived versions

Archived versionWhy supersededRetained valueBytes
MARSOOM_REV6_PHASE1_AUTONOMOUS_PARALLEL_FLASH_MAP_PIPELINE_APPEND_ONLY_V50_2026-09-03 Superseded by the V108-V114 resumable-repair line First autonomous parallel generation; earliest form of the current contract referenced
MARSOOM_REV6_PHASE1_UNIQUE_OWNERSHIP_FLASH_MAP_PIPELINE_APPEND_ONLY_V100_2026-09-03 Superseded by the V114 candidate/feedback pairing fix Introduced unique-anchor ownership, the property proven in the SUCCESS_001 trial referenced

2 archived version(s). Historical packages are never deleted or mutated; large ones are referenced with verified paths rather than copied, because the source archive is 11 GB.

Known limitations

Security and privacy

Page images and indexed references are sent to an external provider. Credentials live outside this repository. Model output is candidate evidence, never truth.

Cost behaviour

Metered per call. The sealed issue-4547 trial cost USD 0.878210625 across 1,487,266 tokens, 36 logical calls and 108 attempts, mixing Flex and Standard tiers.

Copying a tool into this repository does not authorise running it, retraining it, calling a model, or processing a new issue. No paid call may be made without the owner's explicit authorisation and a hard cost cap.

Provenance

Registry
registry/roles.json (schema marsoom.e2t.roles.v1)
Derivation
28 legacy roles - VER - HUMAN + REV = 27 active steps
Legacy role id
MAP-AI
Legacy source SHA-256
6f5d5d7ed8869e45307424c2193f95ffab84abc44a6a188f5c7f98c2a48ec64d
Generated
2026-09-04T22:13:12.433Z

Feedback and decisions

Feedback is recorded per source and never merged into an invented consensus. Where sources disagree, both positions stand and the owner decides.

SourceEvents
Naser / owner
Final authority. Overrides every other source.
none recorded
Codex / orchestrator
Architecture and sequencing.
none recorded
Builder / designer
Implementation reality and constraints.
none recorded
Independent reviewer
Adversarial review of claims.
none recorded
Auditor
Evidence verification against artifacts.
none recorded
Human REV reviewer
Page-level truth from the review site.
none recorded

Recorded events

No feedback events recorded yet. The ledger exists and is append-only: feedback/FEEDBACK_LEDGER.jsonl.

Editing feedback is not possible: a change is a new event whose supersedes names the one it replaces, and the original stays exactly as written.


Generated from registry/roles.json by tools/build-reports.mjs. Do not hand-edit — edit the registry and rebuild.
Original page pixels are the visual authority. CAP owns the exact captured text.