← Index

io.github.antonbabenko/deliberation

io.github.antonbabenko/deliberation·v0.1.0·AI & LLMs

Second opinions in Claude Code and any MCP host from GPT, Gemini, Grok, and 400+ OpenRouter models.

Trust verdict · v1 advisory · method
NOT YET SCREENEDno verdict on file

Verdict not yet evaluated for this tool. The hybrid eval runs adversarial cases first; coverage rolls out as the corpus expands (15 of 150 labels to graduation). Until a verdict is recorded, an agent should treat this tool as not-yet-cleared and fall back to its own checks. Method: hybrid eval, four-state verdict, honest limits.

Own this server? Screen its description →

MCP quality score · maturity, not trust · methodology
freshness
25
completeness
5
installability
25
documentation
10
stability
5
Alternatives in AI & LLMs