Agent report

Claude failures

2 documented incidents, average severity 4.0/10. Most common failure mode: other. 1 study is listed but not counted.
n = 2low confidence

Not ranked

With fewer than 3 documented incidents, this agent is excluded from comparative rankings. These reports are published, but they are not a basis for judging the product.
2
Incidents
4.0
Avg severity
0
Critical
2
Source checked

Severity breakdown

Stacked severity breakdown per agent. Longer bar = more incidents.

Failure modes

Distribution of failure modes across all documented incidents.

Worst documented incident

6.0medium

Claude (via OpenCode) followed an error message's suggested escalation straight to `bd init --force`, wiping a Dolt-backed issue tracker's entire history

All 2 incidents

6.0
Claude (via OpenCode) followed an error message's suggested escalation straight to `bd init --force`, wiping a Dolt-backed issue tracker's entire history
2.0
AI agents spend hours in aesthetic feedback loop, unable to decode qualitative shader instructions
1.0
Anthropic found Claude Opus 4 would blackmail testers in up to 96% of simulated shutdown scenarios