Rust Ai Native Terraform

Terraform a Rust codebase (brownfield adoption) {#root}

@fact:EXECUTING-THE-BROWNFIELD-PROTOCOL You are executing the BROWNFIELD protocol (spec://org.vibevm.ai-native/core-ai-native/mechanisms/BROWNFIELD-PROTOCOL-v0.1 — shipped copy under vibedeps/flow-core-ai-native/<version>/spec/mechanisms/; read it before the first phase, and skim 03-RAID-PLAYBOOK + 05-CAMPAIGN-FORM for the campaign machinery). @status:impl/done

@fact:founding-principles-lead The founding principles: @status:impl/done

• @fact:PRINCIPLE-INVENTORY-NOT-GATE inventory, not gate (the only precondition is "the workspace compiles"); @status:impl/done
• @fact:PRINCIPLE-ASPIRATION-MUST-BE-LABELED aspiration is legal only when labeled; @status:impl/done
• @fact:PRINCIPLE-CONTRADICTION-IS-DATA contradiction is data; @status:impl/done
• @fact:PRINCIPLE-CHARACTERIZATION-IS-THE-TRUTH-OF-RECORD characterization is the truth-of-record; @status:impl/done
• @fact:PRINCIPLE-MONOTONE-UTILITY monotone utility. @status:impl/done

@fact:DO-NOT-BULLDOZE-AN-INHABITED-WORLD Do not bulldoze an inhabited world. @status:impl/done

@fact:TOOLCHAIN-INSTALL-OR-RUN-IN-PLACE Toolchain: rust-ai-native (install once: cargo install --path vibedeps/<stack-slot>/crates/rust-ai-native-cli, or run via cargo run --manifest-path vibedeps/<stack-slot>/Cargo.toml -p rust-ai-native-cli --bin rust-ai-native -- <args>). @status:impl/done

Phase −1 — inventory (record reality; change nothing) {#phase-inventory}

1. @fact:INVENTORY-PRECONDITION Precondition: cargo build --workspace succeeds. (Red build → fix the build first; that is the one true gate.) @status:impl/done
2. @fact:INVENTORY-INIT rust-ai-native init — policies + empty registries. Every crate starts exempt-with-a-reason; nothing is gated yet, and that is correct. @status:impl/done
3. @fact:INVENTORY-TESTS-BASELINE Fill discipline/registry/tests-baseline.json with reality: run the suite once (cargo nextest run --workspace --no-fail-fast), record every failing test as failing-known with a since date and a debt id — do NOT fix them now (drive-by repairs destroy the accounting). @status:impl/done
4. @fact:INVENTORY-HARVEST-INTENT Harvest intent into discipline/registry/intent.json: WAL "Next" / TASKS / ROADMAP items, <!-- REVIEW --> markers, load-bearing TODO/FIXME. The carry-over guarantee: at exit every harvested intention is done | rescoped | rejected(reason) — zero unaccounted. @status:impl/done
5. @fact:INVENTORY-FILE-DEBT File debt into discipline/registry/debt.json: failing tests, known-unimplemented specs, contradictions found while reading — each with severity, evidence, disposition, and touch: tripwires on its watched paths. @status:impl/done
6. @fact:INVENTORY-CHARACTERIZE Characterize currently-passing observable behavior (golden transcripts under discipline/golden/, normalized for volatile fields): these pin "don't break it" independently of whether tests or docs are trustworthy. A pinned bug is visible debt; an unpinned bug is a landmine. @status:impl/done
7. @fact:INVENTORY-SPECMAP rust-ai-native specmap — mint the (initially small) index; commit the whole inventory as its own topic commits. @status:impl/done

Phase 0 — the first spec units {#phase-first-spec-units}

@fact:WRITE-THE-FIRST-SPEC-DOCUMENTS Write the project's first spec/ documents for the subsystems you will touch first: anchored headings ({#req-…}), kind lines (`req r1`). @status:impl/done

@fact:UNIMPLEMENTED-BY-PLAN-UNITS-ARE-MARKED-PLANNED Unimplemented-by-plan units are marked planned — zero coverage there is expected, not red. @status:impl/done

@fact:TAG-IMPLEMENTING-MODULES-AS-YOU-GO Tag the implementing modules (specmark::scope!("spec://<ns>/…")) as you go; rust-ai-native specmap after each batch keeps the index green. @status:impl/done

Phases 1…N — card raids, crate by crate {#card-raids}

@fact:raid-per-crate-in-dependency-order-lead Per the Raid Playbook skeleton (scope & freeze → card order → phases → batches → differential safety → exit criteria), and per crate in dependency order: @status:impl/done

1. @fact:RAID-DRAIN-THE-CRATE Drain the crate: unwraps out of domain code (restructure beats testify), one thiserror enum per layer with REQ-citing messages, a doctest per public seam, cells with oracles where variance exists. rust-ai-native conform check --scope <crate> is the per-crate lens; rust-ai-native fast-loop --cell <crate> keeps the loop under budget. @status:impl/done
2. @fact:RAID-FLIP Flip: add the crate to conform.toml's [rust] gated (remove its [[rust.exempt]] entry) — legal only at zero findings; a flip must never widen the baseline. @status:impl/done
3. @fact:RAID-BEHAVIOR-CHANGES-CARRY-AN-ORACLE Behavior changes carry a differential oracle (card scaffold-d); golden transcripts must fail loudly when stale, never auto-update. @status:impl/done
4. @fact:RAID-PHASE-CLOSES-WITH-THE-FLOOR-GREEN Each phase closes with the floor green (rust-ai-native floor) and its own topic commits, per the Campaign Form's phase-gate discipline. @status:impl/done

@fact:TRACK-THE-CAMPAIGN-PER-CAMPAIGN-FORM Track the campaign per 05-CAMPAIGN-FORM: a cold-executable PLAN, the BASELINE numbers you started from, PREDICTIONS, a LOG, and a closing REPORT. @status:impl/done

@fact:CAMPAIGN-RESUME-POINTER Resume pointer: with a WAL — the standing line at each phase boundary; without — the PLAN's status line + LOG tail (06-WAL-CONVENTION §4). @status:impl/done

Exit criteria {#exit-criteria}

@fact:EXIT-CRITERIA rust-ai-native floor green with every crate either gated or exempt-with-a-living-reason; the test-gate baseline shrunk truthfully (promotions, not silence); the carry-over guarantee met (every intent done | rescoped | rejected); the REPORT written. @status:impl/done

@fact:HELD-BY-THE-RECURRING-SWEEP From here the tree is held by the recurring sweep: /rust-ai-native-sweep. @status:impl/done

The generation-time assistant (before you edit, not instead of the floor) {#generation-time-assistant}

@fact:STACK-SHIPS-AN-AGENTIC-TYPE-ORACLE The stack ships an agentic type oracle. @status:impl/done

@fact:check-the-hypothetical-content-lead Before writing a nontrivial .rs edit, check the HYPOTHETICAL content instead of paying a red floor iteration: @status:impl/done

shCopy code
vibe bin exec rust-ai-native-tcg -- validate src/cells/<cell>.rs \
    --content-from - --root .   # the edit on stdin; exit 1 = would fail

@fact:MCP-ALTERNATIVE or, when the vibevm MCP server is mounted, call tcg_validate with language: "rust" and the content argument (plus tcg_scope / tcg_complete / tcg_type for in-scope symbols, type-valid completions, and quick info). @status:impl/done

@fact:RESPONSES-CARRY-THE-SAME-CONFORM-FINDINGS Responses carry the SAME conform findings as the gate, flagged baselined or new, with guide-citing advice — a new finding in the answer means the floor WILL go red if you write that edit. @status:impl/done

@fact:ORACLE-PREREQUISITE Prerequisite: rust-analyzer on the machine (rustup component add rust-analyzer — a stack obligation). @status:impl/done

@fact:ORACLE-HONESTY Honesty: the oracle is rust-analyzer, not rustc; a clean answer shortens the distance to green, and the floor stays the truth (TCG-ORACLE-RUST §5). @status:impl/done