Critique blog post drafts for technical errors
Posted on Dec 20, 2025 • Edited on Jan 24. Task: The author built a custom 'critique' command in Gemini CLI using the gemini-3-pro-preview model to review Markdown blog drafts for clarity, structure, and technical accuracy. The workflow involved generating a critique.toml configuration, manually refining its argument handling logic, and saving it to the .gemini/commands directory. The author then executed the command via terminal pipes or slash commands to analyze drafts, identifying issues like app initialization errors and inefficient debugging practices.; tools: Gemini. Human role: The original post does not say how a person stayed involved.; limitations: The workflow requires manual refinement of the generated critique.toml to ensure correct argument parsing and logic handling. It depends on the availability of the Gemini 3 Pro Preview model and specific CLI configurations. The source does not detail costs, latency, or how the tool handles non-technical content or complex architectural critiques beyond the demonstrated examples..
Tools
Published on Agent Work Atlas:
Overview
Posted on Dec 20, 2025 • Edited on Jan 24
Workflow
The author built a custom 'critique' command in Gemini CLI using the gemini-3-pro-preview model to review Markdown blog drafts for clarity, structure, and technical accuracy. The workflow involved generating a critique.toml configuration, manually refining its argument handling logic, and saving it to the .gemini/commands directory. The author then executed the command via terminal pipes or slash commands to analyze drafts, identifying issues like app initialization errors and inefficient debugging practices.
Reported result
The tool identified specific technical flaws, including a 'white screen of death' caused by unhandled Firebase initialization errors and inefficient environment variable validation that halted on the first missing key. The author used these insights to refactor code, adding error handling for app initialization, adjusting Remote Config fetch intervals for development mode, and modifying validation to report all missing environment variables at once. No quantitative metrics on time saved or error reduction rates were reported.
Human role
The original post does not say how a person stayed involved.
Limitations
The workflow requires manual refinement of the generated critique.toml to ensure correct argument parsing and logic handling. It depends on the availability of the Gemini 3 Pro Preview model and specific CLI configurations. The source does not detail costs, latency, or how the tool handles non-technical content or complex architectural critiques beyond the demonstrated examples.
Try it
Act as a technical editor using Gemini CLI. I have a draft blog post at [replaceable_input_file_path] about [replaceable_topic]. First, generate a custom command configuration named critique.toml that reviews Markdown files for clarity, structure, technical accuracy, and Dev.to readiness. Manually verify the TOML logic handles file paths correctly. Save it to .gemini/commands. Then, run the command against my draft using the gemini-3-pro-preview model. Output the critique in Markdown format with sections for Typos, Clarity, Structure, Impact, and Technical Code Review. Do not publish or modify the original file yet. Present the critique results to me for review. If the critique identifies code errors, suggest specific refactoring steps but wait for my confirmation before applying any changes to the source code. Ensure the command handles both piped input and direct file arguments.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.