cube — Evaluation Index
Generated: 2026-08-06 CLI version: 1.7.16 Scope: Critical severity Failure modes evaluated: 22 of 74 (scope: Critical severity)
Score Summary
| Severity | Pass (3/3) | Partial (1–2) | Fail (0) | Total |
|---|---|---|---|---|
| Critical | 2 | 11 | 9 | 22 |
| High | 0 | 0 | 0 | 0 |
| Medium | 0 | 0 | 0 | 0 |
| All | 2 | 11 | 9 | 22 |
Average score: 0.8 / 3
Readiness Score
| Dimension | Score |
|---|---|
| Documentation Quality | 1/3 |
| Self-Description | 1/3 |
| Pre-built Integrations | 0/3 |
| Setup Reproducibility | 2/3 |
| Workflow Coverage | 3/3 |
| Total | 7/15 [C] |
Reports
| Report | Audience | File |
|---|---|---|
| Issues & Problems | AI agents and their builders | report-issues.md |
| Runtime Brief | AI agents at invocation time | report-runtime.md |
| Integration Guide | Agent developers | report-agent-dev.md |
| Fix List | CLI authors | report-dev.md |
Top Issues
- OAuth login launches a browser and remains in a polling loop under headless CI
- API responses are unbounded and fully buffered, with no truncation or heartbeat
- Destructive raw API requests have no dry-run, affected-scope preview, or confirmation contract
--jsonreturns raw JSON on success but prose-only stderr on failure- Mutations have no idempotency key or structured partial-failure recovery
Observed Bugs
17 bugs recorded during evaluation — see report-issues.md for details.