CLAIMS & BOUNDARIES
What we claim, and what we do not claim
Askovo's value lies in knowing how far the evidence reaches. This page puts the product's own claims under the same discipline: what can be claimed is stated plainly, and what must not be passed off as more is said outright.
Six conclusion types must not be conflatedSIX CONCLUSION TYPES
| Type | Answers | Must not substitute for |
|---|---|---|
| need | Who has what unmet job in what scene | Feature requests, single cases, prevalence |
| acceptance | Whether users understand or accept the solution | Liking ≠ will use / will buy |
| usability | Whether the task can be completed; learnable and low-error | Usable ≠ demand established |
| performance | Whether it meets the bar under stated conditions | Benchmark score ≠ experience |
| commercial | Purchase, retention, repurchase, returns | Stated intent ≠ transaction |
| safety | Usage risk and compliance gaps | User trials ≠ certification |
E0–E5 grade demand and commercial evidence strength only — they are not a product/commercial reasoning score; usability, engineering performance, safety, and feasibility each have their own verification requirements.
What the tests and the sample proveWHAT TESTS PROVE
| Evidence | Can claim | Must not pose as |
|---|---|---|
| Unit-test matrix (Python 3.9–3.12; current size in CI) | Command behavior, structure validation, and gate logic match the spec | That real demand holds, product quality, market size |
| Offline synthetic sample (askovo demo) | The chain, evidence references, and audit logic are self-consistent | Any real user or commercial fact |
| CI passing (tests.yml) | Those behaviors are reproducible on supported versions | Release feasibility, compliance, or certification |
| Collection via the official GitHub/Reddit APIs | Raw records from authorized sources, with bias recorded | A sample representing a population; record count = user count |
Inferences we never makeNEVER
- No purchase intent treated as a real purchase, no adoption rate treated as a product outcome, no one-time liking treated as repurchase
- No convenience sample extrapolated to prevalence or market size; a market estimate must state population, boundary, and source in full, and null means unknown, not zero
- No
decided(a record of the current choice) passed off as release approval, commercial proof, or causal effect — delivery, outcome, and causality are three independent states - No unauthorized scraping; blocked connectors, login walls, and platform challenges are never bypassed
- All automated deliverables are drafts by default; a passing CLI run never replaces human semantic review, product sign-off, or risk acceptance
The full boundaries are in the repository's README boundaries section and core concepts.
If this check was useful to youSTAR, NEVER REQUIRED
If Askovo helped you, a star on the repository helps more people find it — entirely voluntary, with no conditions attached.