resend — Evaluation Index
Generated: 2026-07-06 CLI version: resend-cli v2.8.1 Scope: Critical Failure modes evaluated: 22 of 74 (scope: Critical)
§75 (Safe-Default Execution Mode Absent), a Critical mode added in the same cycle, was not evaluated in this run.
Score Summary
| Severity | Pass (3/3) | Partial (1–2) | Fail (0) | Total |
|---|---|---|---|---|
| Critical | 4 | 12 | 5 | 22 |
| High | 0 | 0 | 0 | 0 |
| Medium | 0 | 0 | 0 | 0 |
| All | 4 | 12 | 5 | 22 |
Average score: 1.2 / 3
Readiness Score
| Dimension | Score |
|---|---|
| Documentation Quality | 3/3 |
| Self-Description | 2/3 |
| Pre-built Integrations | 3/3 |
| Setup Reproducibility | 3/3 |
| Workflow Coverage | 3/3 |
| Total | 14/15 [A] |
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
whoamireports a synthetic--api-keyas authenticated, which can make agents trust an invalid credential.openanddocslaunch the OS browser even under quiet/JSON-oriented usage, with no structured URL fallback.- Large dry-run content is emitted in full without
meta.truncated,meta.total_bytes, or an output-size cap. - User-provided HTML is returned as ordinary JSON data with no
trusted:falseor external-data boundary. - All observed error cases collapse to exit code 1, forcing agents to parse error strings for retry policy.
Observed Bugs
6 bugs recorded during evaluation — see report-issues.md for details.