link-cli — Evaluation Index
Generated: 2026-06-08 CLI version: 0.7.1 Scope: Critical Failure modes evaluated: 22 of 71 (scope: Critical)
Score Summary
| Severity | Pass (3/3) | Partial (1-2) | Fail (0) | Total |
|---|---|---|---|---|
| Critical | 3 | 11 | 8 | 22 |
| High | 0 | 0 | 0 | 0 |
| Medium | 0 | 0 | 0 | 0 |
| All | 3 | 11 | 8 | 22 |
Average score: 1.2 / 3
Readiness Score
| Dimension | Score |
|---|---|
| Documentation Quality | 1/3 |
| Self-Description | 1/3 |
| Pre-built Integrations | 3/3 |
| Setup Reproducibility | 2/3 |
| Workflow Coverage | 3/3 |
| Total | 10/15 [B] |
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
- Semantic failures all exit 1 and JSON errors omit an
exit_code, so agents cannot branch reliably on process status. - Default
--format jsonis not consistently enveloped; callers must know to add--full-output. - Auth/API timeout and polling behaviors return generic or success-shaped output instead of machine-actionable timeout/partial states.
- Mutating commands lack idempotency, dry-run, danger, and effect contracts.
- Command metadata does not declare required credential scopes.
Observed Bugs
6 bugs recorded during evaluation — see report-issues.md for details.