TVIJO
RUCart 0

COMP · 2026-10-15 · v1

COMP: inspect every star.

A reproducible editorial rubric for a buyer decision.

TVIJO-authored editorial coverage scores; not independent testing, security certification or a performance benchmark. Competitors were not installed here.

Scope before score

A personal-agent runtime, a coding agent, an SDK and a scoped engineering service solve different problems. The table compares six dimensions, not interchangeable subscriptions.

Five checks, ten points

Each criterion lists five atomic checks.2 means documented support within the named scope;1 means a scoped implementation or a building block requiring custom work;0 requires explicit evidence of absence. Missing evidence is null and prevents a criterion total. These are editorial judgments, not measured capability guarantees.

Evidence and independence

TVIJO is the publisher and seller. Its scores are self-assessed and deliberately distinguish scoped project components from ready operator products. Competitor scores use primary documentation observed15 October 2026; those products were not installed or benchmarked here. SDK ecosystem services are named separately.

COMP in the publishing system

Apply COMP with RULEZ/ALKNDA and RULEZ2: buyer question → comparable scope → versioned sources → atomic checks → uncertainty → practical example → relevant service. Keep a prior edition and record the reason for a changed score. A new source must not silently overwrite an old conclusion.

How to run a fair pilot

Use the same public or synthetic input, task limit and acceptance rubric on each candidate. Record completion, unsupported claims, recovery after interruption, operator time and actual provider charges. Set permissions before execution. Publish the denominator and failures before making a speed, cost or quality claim.

Catalogue, not a new daemon

The versioned SQLite catalogue stores products, criteria, assessments, edition and many-to-many sources. Static pages and downloads are exported from that catalogue. There is no paid scoring loop. CSV exports only the selected products; JSON contains the full evidence edition.