Submit code blocks or diffs to an external LLM for structured critique.
npx skills add https://github.com/cropsgg/skills --skill second-opinion-cropsgg
Please help me install this Agent Skill. Skill: second-opinion Source: https://github.com/cropsgg/skills/tree/main/skills/engineering/second-opinion Command: npx skills add https://github.com/cropsgg/skills --skill second-opinion-cropsgg
Surface blind spots in code by using an external LLM to critique proposals.
Submit the code block or full diff to the external LLM review workflow and read its critique.
High-intent search queries and answers about installing and using this skill.
An external LLM code review works by submitting a code block or full diff to a deterministic workflow that enforces a consistent prompt structure. The external model then critiques the proposal to surface risks and blind spots independently.
To get a second opinion on a non-trivial implementation, submit the code block or full diff to the external LLM review workflow. The process applies a structured submission format to generate an independent critique of your design decisions.
Yes, you can use this for PR reviews in high-stakes domains. The workflow applies a multi-model review with consistent prompt structure to provide external verification where independent validation is valuable for risk mitigation.
An external review uses an independent LLM to provide an unbiased critique of your code, complementing the internal session model. This independent verification surfaces blind spots that the original model might miss due to context bias.
No specific dependencies or components are required to run the diff review workflow. The process relies on a defined submission format and external prompts to enforce the deterministic review structure.
skyoxuGuide human-in-the-loop review of code diffs, design proposals, or documentation changes.
Lord1EgyptSimulates diverse expert archetypes to analyze complex problems and decision-making dilemmas.
dsidloAutomate code reviews, security audits, and optimization analysis on software artifacts.
neuron-oneRoute code changes to a different AI model for independent review.
