GLOSSARY

Glossary

Askovo's core concepts on one page. The definitions match the repository docs and each one can be traced back; this page is itself written to be citable — when someone needs to explain evidence grading, they can quote it from here.

The seven-node reasoning chainSCENE → BUSINESS

The main chain from user reasoning to product and business, and the skeleton of every deliverable:

Scene→ User→ Pain→ Need→ Feature→ Product→ Business

scene → user → pain → need → feature → product → business. Every node must record: the current statement, the rationale, the assumptions it rests on, the supporting evidence (evidence_ids), and a status. MRD, PRD, BRD, and the decision package are all projections of this chain.

E0–E5 evidence gradingEVIDENCE LEVELS

An evidence grading internal to the engine, used for gates and audits; it does not enter the body of business deliverables (the narrative uses business language, and the grading stays in the method archive at the end). Level by level:

LevelEvidence formQuestion answered
E0Activity, role, or scene contextThe scene occurs
E1Unmet job, goal, or difficultyThe problem is stated
E2Current practice, workaround, failure, or switching costSubstitutes and their friction
E3Explicit acceptance of or preference for the research conceptThe concept is accepted under stated conditions
E4+Price anchor, purchase or payment intentDirect commercial intent
E4−Refusal, cancellation, return, abandonmentDirect negative commercial behavior
E5Paid possession, deployment, continued use, repurchaseRealized behavioral evidence

Two hard rules: a need node reaches validated only with at least one piece of E2 (observed behavior) — E1 verbal statements alone stop at needs-validation; saying it does not count, doing it does. And business nodes accept only E4+/E5: counter-evidence and substitute behavior will not lift a business node to "observed".

E0–E5 describe demand and commercial evidence strength only — they are not a product/commercial reasoning score; usability, engineering performance, safety, and feasibility each carry their own verification requirements.

●◐○ evidence-strength markersREADING AIDS

A reading aid inside the body of the Business Plan, taken per node: ● observed · ◐ clues · ○ assumption. Scene/user/pain/need accept behavioral records at E2 and above; feature/product accept only E3 (concept acceptance) and above; business nodes accept only E4+/E5.

SURE evidence engineSURE ENGINE

SURE = Structured User Research with Evidence. Askovo is the product (CLI / Agent Skill / MCP / package); SURE is the research and evidence engine underneath it: it owns method selection, evidence forms and grading, claim boundaries, anti-survivorship bias, safety gates, and judgment constraints. In one line: Askovo handles "reasoning from users to product and business", and SURE makes sure "every step knows how far its evidence actually reaches".

Six conclusion typesSIX CONCLUSION TYPES

need / acceptance / usability / performance / commercial / safety are never mixed: need does not pose as acceptance, and stated intent (E1) is never paying behavior (E4+/E5). The boundary for each type is on the claims page.

OverclaimOVERCLAIM

A conclusion that exceeds what its evidence can support — for example, writing E1 stated intent up as "users are willing to pay". askovo audit checks every conclusion for overclaims and names them, exiting with code 1 so CI can gate on it.

Machine truthMACHINE TRUTH

product-reasoning.json is the only semantic source of the seven-node chain; the MRD, PRD, BRD, and decision package are reader views projected from it, not parallel sources of fact. To change a judgment, change the truth — the projections follow.