Glossary
Report terminology
Every label on an INGEST report maps to a measurement-contract meaning. Use this glossary when a glyph, status, or score needs a precise reading — not a marketing gloss.
Topics
Coverage & scores
How INGEST counts what was measured, how Composite is computed, and why a high score with gaps is still honest.
9 terms
Access & robots
Robots policy, HTTP delivery, and content usability are three different axes. The crawler matrix keeps them separate.
6 terms
Rendering & JS dependency
Raw HTML and the post-JavaScript DOM are two representations. Dependency is measured; it is not automatically a defect.
5 terms
Facts & parity
Atomic propositions and how visible HTML agrees or conflicts with machine representations.
4 terms
Structured data
Parse survival is not schema truth. INGEST tests whether intended values survive ordinary parsing.
5 terms
Indexability
Discrete states describing whether the URL looks indexable to ordinary crawlers — not a ranking forecast.
1 terms
Evidence & handoff
Every actionable claim should point at inspectable evidence. Tickets are remediation-shaped views of findings.
4 terms
Byte depth
Where critical facts sit in the initial HTML byte stream — deeper often means higher ingestion risk for truncated consumers.
1 terms
A–Z
- AI readinessDerived blend of crawler access, structured data, fact parity, and JS dependency — only when all four are measured.
- Assertion mode (explicit / inferred)Whether the proposition was stated outright or inferred under fail-closed rules.
- Atomic factA subject · predicate · object proposition the source supports, with provenance and modality preserved.
- Blocked (composite)Convenience composite: robots Disallow OR content not usable.
- Challenge · CAPTCHA · WAF hintHeuristic friction signals observed on the response — not courtroom proof of intent.
- CompositeMean of measured leaf scores only. Null leaves are excluded from the average.
- CoverageHow many leaf score components have a real (non-null) value for this run.
- Crawler identitiesPublic crawler User-Agent classes used for the access matrix (search, AI, archive, other).
- Critical (UI) vs fail vs CRITICALThree different “critical” meanings that must not be collapsed.
- Critical fact byte depthLog-scale offsets of title, canonical, primary JSON-LD, H1, and first price-like fact inside the initial HTML.
- Developer handoff ticketsRemediation tickets derived from non-info findings (fail → P0, warn → P2).
- Diff class (UNCHANGED / JS_ADDED / JS_REMOVED / JS_CHANGED / UNMEASURED)How a field changed between raw HTML and rendered DOM.
- Differential deliveryMaterial field divergence across crawler identities for the same URL.
- EvidenceProvenance payload: URL, timestamp, source, selector, snippet, headers, JSON, agent/rule fields when relevant.
- Fact classes (page-level)Page-level counters for how fact-bearing text shifted after JS — not per-fact provenance labels.
- Fact recovery costHow expensive it is for a machine to recover important facts from the page representation.
- Fetch healthWhether the default fetch produced usable content, with the fetch verdict as the bar label.
- FindingA check result with severity, title, why/interpretation, confidence, and evidence payloads.
- Graph identityDuplicate or unresolved @id values inside JSON-LD blocks.
- Indexability statesINDEXABLE · NOINDEX · UNAVAILABLE · ROBOTS BLOCKED · POSSIBLE_SOFT_404 (and related labels on the bar).
- Intent survivesWhether the intended values remain recoverable without depending on non-standard repair paths.
- JS dependency (LOW / MEDIUM / HIGH / UNMEASURED)Magnitude of reliance on JavaScript to surface content machines might need — not an automatic fail grade.
- Leaf componentsThe atomic measured scores that can feed Composite. Null stays null.
- Measurement contract versionSchema/version tag for the report (currently 1.0.0). Bump when report meaning changes.
- One unescapeWhether the block parses after a single HTML-entity unescape repair.
- Parity statusAgreement class between visible HTML facts and machine-layer facts for the same subject·predicate.
- ProvenanceWhether the fact’s origin is known (PROVENANCE_KNOWN) or unknown (often inferred).
- Raw versus renderedField-level comparison of initial HTML vs browser DOM after scripts run.
- Robots policy (A / D / ?)Whether robots.txt allows, disallows, or leaves UNKNOWN the agent’s request for this URL.
- Schema injectedWhether more JSON-LD blocks appear after render than in initial HTML.
- Search interpretationDeferred leaf. Always unmeasured in contract 1.0.0.
- Standard parseWhether JSON.parse succeeds on the raw JSON-LD script text.
- Unknown countHow many gaps are still open: unmeasured leaves, skipped leaves, and render unavailability when applicable.
- UsableWhether the HTTP response is classified as usable content — independent of robots.txt.
- Values agreeWhen both standard and unescape parses succeed, whether their values are equal.
Read with the contract
Methodology is the narrative model. The glossary is the term-level dictionary for operators reading a live report.