stage-2-validate-templateslisted

# Stage 2 artifact templates All files carry standard frontmatter (artifact-schema). Key structures: ## evidence-ledger.md ```markdown --- artifact: evidence-ledger stage: 2 gate: V1 status: draft evidence_grade: B ... --- # Evidence ledger (single source of truth — grade D never enters; real contacts as P-ids only) ## Sampling frame (V1) — POINTER ONLY; the frame itself lives in `sampling-frame-v1.md` - Frame file: `sampling-frame-v1.md` · manifest: `private/manifest-v1-frame.json` · sha256: ___ - Journaled as `sampling-frame-snapshot` on: YYYY-MM-DD (BEFORE collection began) > The frame is **not** duplicated here. Two copies means the protected one and the narrated one can > diverge, and only the file is hashed — so the metric could be argued against an edited copy. Read the > file for the queries, sources, window, inclusion/exclusion, dedupe rule, stopping rule and claim scope. | id | date | source | root_source_id | type | url_or_ref | retrieved | via | verbatim_or_observation | assumption | grade | bearing | scope_limits | relationship | supersedes | |---|---|---|---|---|---|---|---|---|---|---|---|---|---|---| > Rows from targeted behavior-searches are tagged `exploratory` in `via` and excluded from the V1 denominator. > `root_source_id` = the ORIGINAL source, not the page you read it on (`RS-<slug>`): rows sharing a root > are **one** source in every denominator, however many reposts carried the same complaint — independence > is computed from distinct roots, never d