A — reliable local node for cyb

Deliver a real cyb → cyber → Cybergraph → BBG path with durable, authenticated operation receipts and explicit recovery. Existing local development profiles retain explicit primitive/version identities throughout this work.

A0: storage integration map

Owners: BBG and Cybergraph; product assembly in Cyber/Soft3.

• Trace RAM, fjall SSD, redb archival storage and application transactions from public APIs through disk reads, writes, commit and reopen.
• Exercise real disk backends and distinguish stored bytes, cache visibility, authenticated state and durable operation acceptance.
• Record evidence and the smallest required component changes in BBG audit.

Exit: one explicit integration map identifies which existing component owns each write and which guarantees are still absent. Source inspection and tests have separate evidence. Tests preserve the existing workspace and user data.

A1: complete the existing storage contract

Owners: BBG storage and Cybergraph. Prerequisite: A0.

Priority: P0 in BBG. The ordered component work and failure acceptance gates live in BBG storage reliability. A1 closes only after BBG D1–D5, including native graph publication and real-node recovery evidence, close. This dependency blocks the reliable local-node release.

• Carry I/O failures and ambiguous commit outcomes through the storage API. A failed flush cannot clear the only pending copy or return a durable receipt.

• Make disk reads and scans usable after reopening through the shared interface, with bounded reads and malformed-data errors.

• Preserve hot mutations across warm commits, retain the last copy during eviction, and define explicit archive population and commit boundaries.

  HOT/WARM mutation and last-copy eviction repairs are recorded in the BBG persistence audit. Archive population and error-returning commit boundaries remain in this package.

• Couple accepted native signal bytes, chain position, derived state/head and request receipt through a BBG-owned atomic boundary exposed by Cybergraph.

• Reuse application storage for local application records. Specify how native network publication joins that boundary; preserve the distinction between application heads and neuron SignalChain positions.

• Replace the Soft3 host's independent journal ownership with this path; provide explicit import/recovery for existing development homes.

Exit: new-process restart restores identical accepted operations and state; failed writes, lost replies, conflicting retries, truncated/corrupt input and concurrent writers have defined outcomes. Recovery preserves the last accepted history. Add process-crash tests, then an explicit disk-barrier/power-loss matrix. Root recomputation checks state correctness independently of storage receipts.

A2: authenticated cyb submission and stable retry

Owners: Cybergraph, Cyber, cyb; BBG supplies atomic persistence. Prerequisite: A1 and versioned network/domain identity.

• Carry the caller's signed native signal end to end, including every link in the signal, canonical encoding, origin and authorization context.
• Bind a stable request identity to the exact payload and network domain. Equal retries return the original receipt; changed payloads conflict.
• Validate identity, sequence, byte/count bounds and required proofs before mutation. Enforce reward policy at the receiving node.
• Migrate cyb's relay from individual JSON link retries to atomic signal submission, with network-scoped cursors and recovery from a lost reply.
• Give the unsigned development bridge an explicit capability boundary; reward-bearing acceptance requires its actual verification policy.

Exit: altered signatures, domains, sequences and reward evidence are rejected; a multi-link signal is applied atomically; disconnect/retry/restart gives one accepted operation and one authorized economic effect.

A3: node lifecycle and configuration

Owner: Cyber, consuming Cybergraph/BBG. Prerequisites: A1–A2.

• Version home/config migration and pin network, codec and primitive identity. Validate configuration atomically and keep credentials outside public descriptors.
• Distinguish starting, local_ready, syncing, network_ready, draining and failed. A configuration descriptor describes configuration; health describes liveness.
• Implement bounded SIGINT/SIGTERM drain, listener/lock release and consistent treatment of acknowledged operations and interrupted work.
• Implement doctor and structured errors using the versioned CLI contract.
• Consolidate legacy soft3/true-cyber launch, sync and link entry points around one state owner and explicit compatibility/deprecation behavior.

Exit: wrong-home, wrong-network, corrupt-store, busy-owner and offline cases produce meaningful errors; shutdown and restart preserve accepted state.

A4: actual cyb integration

Owners: cyb and Cyber. Prerequisites: A2–A3.

• Launch the real binary from cyb, consume its descriptor and readiness, and exercise status, graph/explorer, submission and supported balances.
• Test relay interruption, endpoint change within one network, switching networks, multi-link signals and retained personal history.
• Keep the personal cell's keys and private state in its own authority scope. Reuse shared graph/cell components as their contracts permit.

Exit: the actual GUI and node pass the same persistence/retry scenario in isolated homes. HTTP-shape tests remain useful component checks.

A5: local artifact acceptance

Owner: Cyber. Prerequisites: A1–A4 and the applicable distribution gates.

• Build the pinned source closure and run failure probes against dist/cyber.
• Publish accurate local capabilities, supported OS/architecture and recovery instructions with the graph and configuration entry point.

Exit: an independent clean setup can install and use the supported local cyb scenario. Continue with B.

discover all concepts