# AINA §04 ONE-CALL TECHNICAL BRIEF — The Room: Why Git and Chat Are Not Coordination Planes Canonical: https://blayneaina.com/04/brief/ | Full Article: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/ | Raw Markdown: https://blayneaina.com/04.md | All §01–§09 Master Brief: https://blayneaina.com/ai-brief.txt Nav: PrevBrief=https://blayneaina.com/03/brief.txt | NextBrief=https://blayneaina.com/05/brief.txt | Site=https://blayneaina.com/ | Repo=https://github.com/SquadForge/blayne.aina | Contact=me@blayneaina.com Executive Summary: Problem: Git history and chat transcripts cannot establish who may act, receive, stop, or complete work now. | Proves: The Room revalidates territory, causal observation, sealed wakes, native replies, cancellation, and privileged acceptance. | Scope: Use the Multi-Agent Coordination & Acceptance Audit Challenge on one assignment with an uncertain outcome. Parts (5): Part 1: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/part-1/ (raw: https://blayneaina.com/04/part-1.txt) | Part 2: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/part-2/ (raw: https://blayneaina.com/04/part-2.txt) | Part 3: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/part-3/ (raw: https://blayneaina.com/04/part-3.txt) | Part 4: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/part-4/ (raw: https://blayneaina.com/04/part-4.txt) | Part 5: https://blayneaina.com/04-the-room-why-git-and-chat-are-not-coordination-planes/part-5/ (raw: https://blayneaina.com/04/part-5.txt) - Introduction & Thesis [Part 1/5]: # The Room: Why Git and Chat Are Not Coordination Planes Source history records what changed. Chat records what was said. The Room decides who may read, act, receive, stop, and close work now—and makes every transition prove it. By Blayne Aina · Principal Systems Architect… - The Whole Machine, Not a Wake Queue [Part 1/5]: The Room includes eleven cooperating mechanism families grouped into six architectural domains. Nine receive numbered subsystem sections below; worker staging and controller-side brokerage are developed within native execution and cancellation rather than numbered separately. - One Assignment Through the Machine [Part 1/5]: Consider a coordinator assigning renewal-evidence review to reviewer-beta. The path is not “write a prompt and hope the right agent answers”: If registration changes before dispatch, the old wake stops with wake_recipient_registration_changed. — Key Rules/Claims: Territory and principal. The coordinator's credential, native conversation, project object, and Room scope are re-derived. The recipient's registration is current. | Recognized work. An EISA submission supplies the constitutional act and its evidence handles. - 1. Git and Chat Are Necessary—and Insufficient [Part 1/5]: Git is an excellent source-history plane. Chat is an excellent human and model communication surface. The Room does not replace either. It prevents their artifacts from impersonating live authority. Clean merges do not prove semantic non-interference. — Spec/Table: Tool-call schema; Native value: The proposal has a valid shape; Additional question the Room answers: Is this exact act recognized, currently standing, and inside the caller's verb and tier authority? | Graph/workflow checkpoint; Native value: A prior plan can resume; Additional question the Room answers: Is that plan valid for the current target, conversation, task lineage, registration, generation, and… - 2. Subsystem 1: Territory and Principal Identity [Part 1/5]: Subtopics: Split-brain state-tree refusal; Project grants pin path and kernel object identity; CSPRNG enrollment without persistent plaintext secrets; Thread-local sessions and native conversation identity; Tier clamping includes the route, not just the tool name; Stdio and pre-model egress are separate boundaries — Authority starts by proving which state tree and which physical project object the act concerns. Before the engine opens its state tree, validate_state_dir requires the state-directory identity marker. — Key Rules/Claims: a unique project ID and bounded name; | the requested and canonical root paths; | root_object_id, a 192-bit combined identity: a 64-bit volume serial number plus the 128-bit NTFS/ReFS file ID; | opening actor and time; | an optional governing-source grant and its reading policy. - 3. Subsystem 2: The Room Graph Is a Visibility and Recovery Boundary [Part 2/5]: Subtopics: Withholding closes over descendants; Composition is intersection, not union, of authority; Recovery refuses reparse shortcuts and preserves causal order — Each Room owns a physically isolated graph shard derived from the workspace instance and Room name: binding.json permanently fixes Scope::Coordination or Scope::Project { id }. - 4. Subsystem 3: The Room Sync Engine and the Pre-Action Work Gate [Part 2/5]: Subtopics: A durable per-Room sync index; Handle-pinned reconciliation; Checkpoints bind actor, conversation, credentials, Room, and view; The Work Gate requires frontier and causal readiness — A coordination plane cannot rely on an agent remembering what it has read. It must measure the observation. Each Room has sync_index.db under squad.room_sync_index.v1. — Key Rules/Claims: journal_mode = WAL; | synchronous = FULL (2); | foreign_keys = ON; | busy_timeout = 5000 ms. | The conversation checkpoint is ready and confirmed_offset covers the live coordination frontier. - 5. Subsystem 4: Contribution Verbs Have Different Authority Effects [Part 2/5]: The Room is not freeform chat with locks added later. Its verbs deliberately separate status, answer, assignment, transfer, and completion. An untargeted say reply cannot resolve a mixed set of pending obligations. — Spec/Table: note; Wake and task rule: Never wakes; visible @mentions refused; Constitutional effect: Status or identity telemetry without mentions, obligations, or dispatch | reply; Wake and task rule: Never wakes; explicit reply_to required; Constitutional effect: Resolves an addressed mention or obligation after validating that it belongs to the caller | say; Wake and task rule: Coordination-only wake… - 6. Subsystem 5: A Wake Is a Sealed Authority Transition [Part 3/5]: Subtopics: The 18-field room-wake-job-v3 binding; Five live dispatch checks; Single flight, capacity, and split-lock claim; Evidence is delivered to the recipient, not inherited from the sender; Exhibit 1 — A sealed wake binding excerpt; Exhibit 2 — Registration changed after sealing; Exhibit 3 — Deadline consumed before dispatch — A queued prompt is not execution authority. It is a proposal to deliver work later, under conditions that may have changed. binding_sha256 covers these exact queue fields: The explicit task: null matters: adding or stripping a task binding changes the digest. — Key Rules/Claims: Sealed integrity and replay eligibility. Schema, binding digest, canonical queue object, and retry_allowed must hold. Tampering records wake_job_binding_changed as an integrity signal. - 7. Subsystem 6: Native Session Delivery and Reply Are Different Proofs [Part 3/5]: Subtopics: Session Factory v1; The five delivery/reply proof states — A tool transport can deliver a request to a process. It cannot prove that the intended native conversation received it or that the resulting obligation was answered. — Spec/Table: dispatched; What is proven: A delivery attempt exists, not body delivery; Obligation status: delivery_unproven; unresolved | delivered_to_body / ide_delivered; What is proven: The prompt reached the native session body; Obligation status: pending_reply; unresolved | body_visible_pending_reply; What is proven: The target thread UI or store shows the prompt; Obligation status: pending_reply;… - 8. Subsystem 7: Active Cancellation and Worker Authority [Part 3/5]: Subtopics: Authenticated control and held in-flight operations; The 1 Hz owned-Job observer and durable stop outbox; Worker staging minimizes data before confinement; The controller-side MCP broker revalidates the worker and the call — Room control is authenticated and physically enforced. It does not rely on a worker noticing a cancellation message eventually. Lifecycle actions require control_actor standing through an authenticated operator or local owner console. - 9. Subsystem 8: Task Lineage and Privileged Acceptance [Part 4/5]: Subtopics: Delegation cannot outlive or widen its ancestry; Seven acceptance kinds, evaluated by the engine; Completion revalidates authority and live evidence; Exhibit 4 — Successful artifact acceptance; Exhibit 5 — file_absent cannot turn an observation error into absence — A Room reply says the worker answered. Acceptance asks whether the actual artifacts satisfy the task. Task scopes exclude controller state, recovery storage, shipped binaries, skill mirrors, and authoritative Room ledgers. — Spec/Table: sha256: Reads the in-scope file and compares its live SHA-256 with the expected digest | file_exists: Uses Path::try_exists(); an OS or permission error is unconfirmed, not success | file_absent: Uses the same observation law: an error fails closed instead of becoming absence | command: Executes an absolute, owner-allowlisted binary with no shell or PATH lookup, optional exact argument… — Key Rules/Claims: Basic requires at least min task submissions with one terminal outcome and no pending, recording-pending, or worker-unread results. | StrictReturn requires one applied submission with the exact worker, target, expected fields, and version delta +1, then proves that the original worker received the result… - 10. Subsystem 9: Five Runtime Invariants Constrain Event Law [Part 4/5]: Subtopics: Exhibit 6 — An in-doubt TTL record — Pre-action gates decide whether an act may proceed. The runtime invariant engine continuously checks the event stream for structural violations and records dedicated invariant_violation nodes through the journal and Room graph. — Spec/Table: exactly_once; State carrier and rule: A bounded 10,000-entry FIFO IdempotencyRegistry rejects a repeated confirmed idempotency key; Severity and boundary: Reject; deduplicates registered events, not every external effect | well_formed; State carrier and rule: A lifecycle tracker per (agent_id, parent_id) enforces valid Begin → Effect → Resolve transitions; Severity and boundary: Reject; an… - 11. The 2026 Coordination Evidence Landscape [Part 4/5]: The 2026 literature and protocol specifications reinforce the same architectural distinction: state management, native session identity, revalidation, and measured completion are separate control surfaces. Each source has a bounded claim. — Spec/Table: Liu et al., Multi-agent Collaboration with State Management (STORM) (https://arxiv.org/abs/2605.20563), May 2026 preprint; What it actually establishes: Worktree isolation defers conflicts to post-hoc merge; STORM mediates shared-workspace state at write time and outperforms the baseline by 18.7 points on Commit0-Lite and 1.4 on PaperBench, with comparable or better cost efficiency.; How the Room composes… - Bring One Assignment With an Uncertain Outcome [Part 5/5]: The Multi-Agent Coordination & Acceptance Audit Challenge evaluates one assignment whose delivery, ownership, or completion is currently hard to prove. The demonstration maps causal readiness, replays sealed-wake acceptance and refusal, and measures the resulting artifact… - The Twelve Coordination Claims [Part 5/5]: Spec/Table: 04-C01; Bounded claim: Split-brain state trees are refused, and project grants revalidate canonical path plus 192-bit combined kernel object identity; Evidence and ceiling: Implemented and author-verified on local NTFS/ReFS territory; not a remote-share or host-admin guarantee | 04-C02; Bounded claim: Enrollment, thread-local sessions, native conversation binding, tier clamping, and multiplexer verb… - The Article 04 Claim Card [Part 5/5]: Key Rules/Claims: Mechanism: | The AINA Room Coordination Plane | (state identity, project grants, connection/session authority, tool policy, | graph shards, visibility, sync, contribution verbs, wake permits, | Session Factory, worker confinement, cancellation, acceptance, invariants) | Failure it governs: | Source history and chat impersonating live authority; | split-brain state and redirected territories; | stale context, causal-parent blindness, and ambiguous obligations; | stale queued wakes and wrong-session delivery; | process exits and agent prose impersonating completion; | unobserved cancellation and unverified task closure. - The Room Is Where Authority Becomes Current [Part 5/5]: A recognized act can be mathematically sound and still be operationally stale. The evidence may be valid, the partition correct, the destination plausible—and the worker may be reading an old frontier, holding a replaced registration, answering a superseded task, or exiting… - Frequently Asked Questions [Part 5/5]: Subtopics: Why aren't Git worktrees enough for multi-agent coordination?; Why isn't a workflow checkpoint enough?; What makes a Room wake different from a prompt queue?; Is delivery to an IDE session proof of completion?; Why does file_absent use try_exists()?; Do the five invariants prove exactly-once external effects? — Worktrees separate source histories and defer conflict resolution to a later merge. They do not establish current resource ownership, causal observation, wake validity, native delivery, or artifact acceptance. - Continue the Series [Part 5/5]: Key Rules/Claims: Series index: From Proposal to Consequence: The AINA Transition-Governance Series (https://blayneaina.com/) | Previous: Article 03 — Semantic Switch and EISA: Protected Computation and Entity Resolution (https://blayneaina.com/03-eisa-four-gate-verdict-machine-and-entity-resolution/) | Next: [Article 05… - Primary 2026 Sources [Part 5/5]: Key Rules/Claims: Liu et al. — Multi-agent Collaboration with State Management (STORM), arXiv:2605.20563 (https://arxiv.org/abs/2605.20563) | Sarkar — Before the Pull Request: Mining Multi-Agent Coordination, arXiv:2606.19616 (https://arxiv.org/abs/2606.19616) | [Nechepurenko and Shuvalov — Coordination as an Architectural… - Publication Metadata [Part 5/5]: Title tag: The Room: Why Git and Chat Are Not Coordination Planes Meta description: Why Git worktrees and chat transcripts cannot govern live agents: inside The Room's territory pins, causal sync gate, sealed wake permits, native-session proof, cancellation, and…