13 · Build Plan
The build plan turns architecture into evidence-gated tasks. Status describes repository proof—not aspiration—and no task becomes Verified without tests and completion evidence.
Specification → foundation → exchange/risk → modes → operations → release
Acceptance, evidence and risk
Audit existing specification and deployment
Dependencies
None
Priority / risk
Critical / Low
Blocking condition
None
Last update
2026-07-16
Acceptance / evidence: docs/architecture-audit.md; production/local SHA-256 match
Next action: Maintain as the compatibility baseline
Create multi-page information architecture
Dependencies
SPEC-001
Priority / risk
Critical / Low
Blocking condition
None
Last update
2026-07-16
Acceptance / evidence: 17 generated routes, shared navigation and breadcrumbs
Next action: Keep cross-links valid as content changes
Verify vendor-sensitive claims
Dependencies
SPEC-001
Priority / risk
High / High
Blocking condition
Some product decisions need implementation proofs
Last update
2026-07-16
Acceptance / evidence: Claim ledger and dated source register
Next action: Recheck before each affected implementation prompt
Adopt t3code upstream and establish additive packages
Dependencies
SPEC-003
Priority / risk
Critical / High
Blocking condition
Product repository not yet scaffolded
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 01, then Prompt 02
Privy authentication and master-wallet boundary
Dependencies
FOUND-001
Priority / risk
Critical / High
Blocking condition
Wallet ownership and recovery runbook need implementation validation
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 05 after Prompts 01–04
Typed Hyperliquid adapter and discovery
Dependencies
FOUND-001, FOUND-002
Priority / risk
Critical / High
Blocking condition
No production code
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 06
Deterministic risk kernel
Dependencies
EXCH-001
Priority / risk
Critical / High
Blocking condition
Canonical numeric limits are open decisions
Last update
2026-07-16
Acceptance / evidence: None
Next action: Approve testnet limits and run Prompt 07
Narrow typed tools in t3code runtime
Dependencies
RISK-001
Priority / risk
Critical / High
Blocking condition
Upstream extension point must be proven
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 08
Self lifecycle and recovery
Dependencies
AGENT-001
Priority / risk
High / Medium
Blocking condition
No execution loop
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 09
Authoritative match state machine
Dependencies
SELF-001
Priority / risk
Critical / High
Blocking condition
Custody/legal and settlement decisions require review
Last update
2026-07-16
Acceptance / evidence: None
Next action: Resolve ADR-007 then run Prompt 10
Approved wager accounting and idempotent settlement
Dependencies
VS-001
Priority / risk
Critical / High
Blocking condition
Legal/operational approval and exact custody implementation are not evidenced
Last update
2026-07-16
Acceptance / evidence: Specification baseline only
Next action: Obtain specialist review; then run Prompt 11
Append-only replay and agent-mind interface
Dependencies
VS-001
Priority / risk
High / Medium
Blocking condition
Canonical events not implemented
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 12
Recovery, observability, and security hardening
Dependencies
SELF-001, VS-001, REPLAY-001
Priority / risk
Critical / High
Blocking condition
Runtime does not exist
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompts 13 and 14
Testnet end-to-end evidence pack
Dependencies
OPS-001, SETTLE-001
Priority / risk
Critical / High
Blocking condition
Previous milestones incomplete
Last update
2026-07-16
Acceptance / evidence: None
Next action: Run Prompt 15
Production product deployment
Dependencies
VERIFY-001
Priority / risk
Critical / High
Blocking condition
End-to-end verification required
Last update
2026-07-16
Acceptance / evidence: Specification site deployment is separate; product is not built
Next action: Run Prompt 16 only after testnet evidence passes
Evidence is required
Passing tests, relevant source files, screenshots, deployment output, API responses, logged state transitions, recorded test transactions and verified database/artifact records are acceptable evidence. “UI renders,” “code written” or an agent assertion alone is not completion.