@tiangong-lca/cli 0.0.7 → 0.0.8

package/README.md

@@ -17,8 +17,8 @@ checkPaths:
   - bin/**
   - src/cli.ts
   - src/main.ts
-lastReviewedAt: 2026-05-08
-lastReviewedCommit: 06dbec3287513cf94c5d64c73bd467ed5347f40a
+lastReviewedAt: 2026-06-01
+lastReviewedCommit: 24f96578290267865df3ee1244a96723129bc376
 ---
 
 # TianGong LCA CLI
@@ -117,6 +117,66 @@ tiangong-lca process get --id <process-id> --version <version> --json
 tiangong-lca process list --state-code 100 --limit 20 --json
 ```
 
+## Identity Preflight
+
+Use identity preflight before generating new process or flow rows. The command compares one target against local candidate rows and emits a machine-readable `IdentityDecision` so automation can reuse, update, block, or route uncertain cases before payload generation.
+
+```bash
+tiangong-lca process identity-preflight --input ./process-preflight.json --out-dir ./process-preflight --json
+tiangong-lca flow identity-preflight --input ./flow-preflight.json --out-dir ./flow-preflight --json
+tiangong-lca process identity-preflight --input ./process-preflight.json --candidate-input ./exports/processes.jsonl --candidate-input ./local-process-catalog --out-dir ./process-preflight --json
+tiangong-lca flow identity-preflight --input ./flow-preflight.json --remote-candidates --remote-query "electricity medium voltage" --remote-limit 20 --out-dir ./flow-preflight --json
+```
+
+Minimal input:
+
+```json
+{
+  "target": {
+    "name_en": "market for electricity, medium voltage",
+    "reference_flow_id": "flow-electricity",
+    "operation": "produce"
+  },
+  "candidates": [
+    {
+      "id": "existing-process",
+      "name_en": "market for electricity, medium voltage",
+      "reference_flow_id": "flow-electricity",
+      "operation": "produce"
+    }
+  ]
+}
+```
+
+Key outputs under `--out-dir`:
+
+- `outputs/identity-decision.json`
+- `outputs/identity-candidates.jsonl`
+- `outputs/identity-candidate-sources.json`
+
+`--candidate-input` is repeatable and accepts JSON, JSONL, or a directory scanned recursively for JSON/JSONL candidate rows. Embedded `candidates` from the request and local-scan candidates are evaluated together. Add `--remote-candidates` when the preflight should also call `process_hybrid_search` or `flow_hybrid_search`; `--remote-query` overrides the target-derived search text and `--remote-limit` caps returned candidate rows. Remote candidate search uses the normal Supabase session env: `TIANGONG_LCA_API_BASE_URL`, `TIANGONG_LCA_API_KEY`, `TIANGONG_LCA_SUPABASE_PUBLISHABLE_KEY`, and optional `TIANGONG_LCA_REGION`.
+
+Exact process exchange fingerprints with matching identity context block duplicate creation, while weaker inventory-only matches still route to manual review. Flow preflight also blocks alias-equivalent flows when type, reference property, unit, and category/CAS evidence match.
+
+## Build Plan Gate
+
+Use build-plan gates after identity preflight and before publish handoff. These commands validate the minimum authoring contract for a process or flow build plan, write a standard `GateReport` for Foundry/skill orchestration, and materialize deterministic canonical TIDAS payloads when no explicit payload is embedded in the plan.
+
+```bash
+tiangong-lca process build-plan validate --input ./process-build-plan.json --out-dir ./process-build-plan --json
+tiangong-lca process build-plan materialize --input ./process-build-plan.json --out-dir ./process-build-plan --json
+tiangong-lca flow build-plan validate --input ./flow-build-plan.json --out-dir ./flow-build-plan --json
+tiangong-lca flow build-plan materialize --input ./flow-build-plan.json --out-dir ./flow-build-plan --json
+```
+
+The minimum plan contract requires an automatic identity decision, EvidenceManifest sources and field bindings, name plan, and the relevant process reference-flow or flow-property fields. Process materialization carries name, quantitative reference, exchange, source evidence, modelling, administrative, and annual supply/production fields from the plan into `processDataSet`; when annual volume is not explicit, it falls back to reference-flow `meanAmount` and then `resultingAmount`. Flow materialization carries name, flow type, reference property, source evidence, administrative, and classification fields into `flowDataSet`. `--report-only` keeps exit code `0` while still reporting blockers.
+
+Key outputs under `--out-dir`:
+
+- `outputs/build-plan-gate-report.json`
+- `outputs/materialized-process.json`
+- `outputs/materialized-flow.json`
+
 ## Real DB Flow Review
 
 1. Search or otherwise collect exact flow refs.
@@ -193,11 +253,18 @@ Key `flow materialize-decisions` outputs:
 ## Other Common Commands
 
 ```bash
+tiangong-lca process identity-preflight --input ./process-preflight.json --candidate-input /abs/path/to/process-candidates.jsonl --out-dir /abs/path/to/process-preflight --json
+tiangong-lca flow identity-preflight --input ./flow-preflight.json --candidate-input /abs/path/to/flow-catalog --out-dir /abs/path/to/flow-preflight --json
+tiangong-lca process identity-preflight --input ./process-preflight.json --remote-candidates --remote-limit 20 --out-dir /abs/path/to/process-preflight --json
+tiangong-lca process build-plan validate --input ./process-build-plan.json --out-dir /abs/path/to/process-build-plan --json
+tiangong-lca flow build-plan validate --input ./flow-build-plan.json --out-dir /abs/path/to/flow-build-plan --json
 tiangong-lca process auto-build --input ./examples/process-auto-build.request.json --out-dir /abs/path/to/process-run --json
 tiangong-lca process resume-build --run-dir /abs/path/to/process-run --json
 tiangong-lca process publish-build --run-dir /abs/path/to/process-run --json
 tiangong-lca process batch-build --input ./examples/process-batch-build.request.json --out-dir /abs/path/to/process-batch --json
 tiangong-lca dataset validate --input ./rows.jsonl --type auto --out-dir /abs/path/to/dataset-validate --json
+tiangong-lca dataset evidence-search plan --query "中国2026年电力结构数据" --out-dir /abs/path/to/evidence-search --json
+tiangong-lca dataset evidence-search run --input ./evidence-search.request.json --results ./search-results.json --out-dir /abs/path/to/evidence-search --json
 tiangong-lca dataset references rewrite --input ./rows.jsonl --from flow:<old-id>@<old-version> --to flow:<new-id>@<new-version> --out-dir /abs/path/to/dataset-rewrite --json
 tiangong-lca lifecyclemodel auto-build --input ./examples/lifecyclemodel-auto-build.request.json --out-dir /abs/path/to/lifecyclemodel-run --json
 tiangong-lca lifecyclemodel validate-build --run-dir /abs/path/to/lifecyclemodel-run --json
@@ -209,6 +276,8 @@ tiangong-lca review process --rows-file ./process-list-report.json --out-dir ./r
 tiangong-lca review process --run-root /abs/path/to/process-run --run-id <run_id> --out-dir ./review
 tiangong-lca process save-draft --input ./patched-processes.jsonl --out-dir /abs/path/to/process-save-draft --dry-run --json
 tiangong-lca process save-draft --input ./patched-processes.jsonl --out-dir /abs/path/to/process-save-draft --commit --json
+tiangong-lca flow publish-version --input-file ./ready-flows.jsonl --out-dir /abs/path/to/flow-publish --dry-run --json
+tiangong-lca flow publish-reviewed-data --flow-rows-file ./reviewed-flows.jsonl --out-dir /abs/path/to/reviewed-publish --dry-run --json
 tiangong-lca publish run --input ./publish-request.json --dry-run
 tiangong-lca doctor --json
 ```
@@ -217,10 +286,22 @@ For `publish run`, relative `out_dir` values from either the request body or `--
 
 For `review process`, `--rows-file` accepts either raw process rows as JSON/JSONL or the full JSON report emitted by `tiangong-lca process list --json`, as long as it contains a `rows` array.
 
+For `process identity-preflight` and `flow identity-preflight`, canonical TIDAS wrappers are schema-checked when present. Loose target objects are accepted for early planning and produce `schema_validation.status: "not_applicable"` until materialization. Candidate rows can be embedded in the request, loaded from repeatable `--candidate-input` local files/directories, or fetched through explicit `--remote-candidates` hybrid search; `identity-candidate-sources.json` records scanned files, remote endpoints, queries, filters, and row counts.
+
+For `process build-plan` and `flow build-plan`, canonical payloads embedded in the plan are schema-checked during `materialize`. Plan-only materialization now creates deterministic canonical `processDataSet` / `flowDataSet` wrappers from the build plan and validates them with the TIDAS SDK before reporting `passed`.
+
 For `process save-draft`, canonical process payloads are validated locally with `ProcessSchema` before any `--commit` write. Schema-invalid rows remain in `outputs/save-draft-rpc/failures.jsonl` instead of being persisted.
 
+For `flow publish-version`, canonical flow payloads are validated locally with `FlowSchema` before remote visibility planning or writes. The command always writes `flow-publish-version-gate-report.json`; blocked rows are written to the remote-failure JSONL without calling the remote service.
+
+For `process publish-build`, canonical process payloads are validated locally with `ProcessSchema` before publish handoff artifacts are written. The gate report is `reports/process-publish-schema-gate.json`.
+
+For `publish run`, `verification-report.json` is written next to `publish-report.json` and summarizes the publish ruleset status, blockers, failed entries, deferred entries, and executed entries.
+
 For `lifecyclemodel save-draft`, canonical lifecyclemodel payloads are validated locally with `LifeCycleModelSchema` before any `--commit` write. Schema-invalid rows remain in `outputs/save-draft-bundle/failures.jsonl` instead of being persisted.
 
+For `dataset evidence-search`, `plan` creates the field-level query matrix and search budget. `run` accepts normalized external search results from browser/web-search tools or a generic JSON provider endpoint, then writes `outputs/evidence-search-plan.json`, `outputs/evidence-search-results.jsonl`, `outputs/evidence-search-report.json`, and `outputs/evidence-search-declaration.json` when evidence is absent or only partial. The CLI records scope and normalization; Codex/skills still own semantic judgement and source selection.
+
 For `dataset references rewrite`, `--commit` executes the state-aware save-draft path for patched process and lifecyclemodel rows; without `--commit`, the command only writes local rewrite artifacts.
 
 ## More Docs