What problem does it solve? Self-review by a single AI assistant cannot catch its own blind spots, and asking the same vendor's model again produces the same biases. This Skill gets a genuine second opinion by routing your document, plan, or decision to AI assistants from other vendors installed on the machine, then synthesizing where they agree, where they disagree, and which position is better supported. ## Core Features & Use Cases - Cross-Vendor Peer Review: Discovers installed assistant CLIs (Antigravity, Claude Code, Codex), lets the user pick which peers to convene, and relays their verbatim responses. - Structured Synthesis: Reports convergence, disagreements, errors caught by peers, and unresolved issues, with the host adjudicating splits by reasoned argument rather than majority vote. - CLI Trap Avoidance: Documents per-CLI invocation pitfalls (argument ordering, stdin vs argument prompts, permission flags, session resume) so reviews do not silently answer the wrong question. - Use Case: Before committing to an architecture decision, ask the assistant to run it past Codex and Claude; receive both critiques verbatim plus a synthesis of what the host's own analysis missed. ## Quick Start Ask the assistant to run this plan past the other AI assistants installed on this machine and report where they disagree.
