1. Claim label
Check: name the public claim being inspected before stronger wording appears.
Stop condition: if the label is vague or combines several claims, split it before continuing.
Guided proof-path tour
Start with one public claim, then stop where evidence stops.
This tour shows how a teammate or reviewer can read an existing public-safe evidence surface: name the claim, follow the proof path, check the rule, tier, coverage, source context, limitation, and owner, then decide whether the public wording must stop.
Public claim level: concept. No public conclusion without evidence. This is a guided explanation, not a proof engine, runtime trace, AI analysis, release approval, or operational approval flow.
Positioning
The tour reads one claim; nearby routes keep their own jobs.
/proof-paths/Canonical proof-path overview and route families; this tour is the guided reading flow for one claim.
/proof-source-catalog/Source-family catalog; the tour uses it as a reference, not a replacement catalog.
/demo/evidence-trail/Demo evidence trail; the tour stays concept-level and claim-to-proof-path focused.
/review-room/Review agenda; the tour is not a workspace, queue, approval, or decision room.
/packets/ and /packets/assembly/Packet guidance; the tour does not publish or assemble delivery packets.
/validation/ and /limitations/Validation and broad boundaries stay separate; this page links to them without presenting itself as a validation result.
/demo/runbook/Demo runbook; the tour avoids operational runbook and release-readiness language.
/review-claim-checklist/, /review-claim-checklist/drill/, and /glossary/The checklist decides repeat, downgrade, hold, internal, or do-not-repeat; the drill offers practice rows; the glossary defines terms. This tour applies those ideas to one reading path.
Tour steps
Each field either carries evidence or keeps the conclusion bounded.
2. Public claim level
Check: confirm whether the claim is concept, demo, main, future, hidden, or another visible level from the glossary.
Stop condition: concept-level copy cannot be promoted into demo or main wording by this tour.
3. Proof path
Check: follow the public route, checked-in doc, or public-safe reference trail that supports the claim.
Stop condition: a local-only or private trail is not a public proof path.
4. Supporting public route/artifact
Check: link to a route such as the source catalog, validation, limitations, or a reviewed public summary.
Stop condition: if the route is missing or private, the public claim must stop or move to owner follow-up.
5. Rule ID/family
Check: record the deterministic rule ID, extractor family, or documented judgment that produced the evidence.
Stop condition: no evidence-bearing conclusion without a rule ID or rule family.
6. Evidence tier
Check: identify Tier1Semantic, Tier2Structural, Tier3SyntaxOrTextual, or Tier4Unknown.
Stop condition: unknown or syntax-only evidence must remain labeled.
7. Coverage label
Check: keep full, partial, reduced, unknown, unavailable, or gap-labeled coverage visible.
Stop condition: missing coverage blocks complete or absence-style wording.
8. Commit SHA/source context
Check: record the commit SHA or public-safe source revision context that bounds the scan or summary.
Stop condition: if the source context is absent, do not repeat the evidence as current.
9. Extractor version
Check: record the extractor, schema, or rule-family version behind the evidence.
Stop condition: if the extractor version is missing, the deterministic support is incomplete.
10. Limitation
Check: state the limitation beside the evidence field it bounds, not as a generic footer.
Stop condition: a claim without its limitation is not ready to repeat.
11. Non-claim
Check: say what the inspected evidence does not prove before the reader carries the wording onward.
Stop condition: if the non-claim is missing, send the wording to the claim checklist.
12. Next owner
Check: name the reviewer, implementation owner, validation owner, or human decision maker for the next action.
Stop condition: if no owner is named, the tour ends with owner follow-up, not a stronger conclusion.
Illustrative worked example
This placeholder example is not a real product claim.
It demonstrates the reading shape only. The claim label, commit SHA, extractor version, and conclusion are authored placeholders.
Claim labelPlaceholder claim: public demo summary row is inspectable through a proof path.
Public claim levelconcept.
Proof pathProof path index to source catalog to validation expectations.
Rule ID/family
public.demo.summary.v1 as an illustrative rule family.Evidence tier
Tier2Structural for the authored example; Tier4Unknown would stop stronger wording.Coverage label
PartialAnalysis, so the reading outcome must stay bounded.Commit SHA/source context
abc1234, illustrative only and not a TraceMap finding.Extractor version
example-extractor-v0, illustrative only.Supporting public route/artifactDemo evidence trail and checked-in public-safe route metadata, not raw local artifacts.
LimitationThe row shape can orient a reviewer, but partial coverage blocks broad or absence-style wording.
Non-claimThe example does not prove product behavior, runtime behavior, release approval, or complete coverage.
Next ownerReviewer checks the public route; validation owner supplies missing evidence if the claim needs stronger wording.
Bounded non-claim conclusionThe placeholder claim may be read as concept-level orientation only; stop before repeating it as a real TraceMap finding.
Where to stop
Stop when an evidence-bearing field is absent or incomplete.
- Stop when rule ID/family, evidence tier, coverage label, commit/source context, extractor version, limitation, or supporting public route/artifact is absent or incomplete.
- Claim label, public claim level, proof path, non-claim, and next owner are framing or handoff fields; if one is missing, the tour is incomplete even when it is not the evidence-presence gate.
- Stop when a supporting path points only to raw material, private review notes, hidden validation details, or generated local output.
- Stop when wording would outgrow the visible public claim level, especially when the page is concept-level guidance.
Non-claims
The tour is not stronger than the evidence it reads.
- Do not claim runtime behavior, production traffic, endpoint performance, outage cause, release safety, operational safety, release approval, or complete coverage from this tour.
- Do not present TraceMap as AI/LLM impact analysis, embeddings, vector databases, prompt classification, autonomous approval, or a replacement for tests, code review, source review, runtime observability, or human judgment.
- Do not publish raw facts, raw SQLite content, analyzer logs, raw source snippets, raw SQL, config values, secrets, local paths, raw remotes, generated scan directories, private sample names, hidden validation details, raw command output, or credential-like values.
Continue