Run Codex CLI web research and save findings as timestamped Markdown, JSON, and text files.
npx skills add https://github.com/shoei05/codex-search --skill codex-search-shoei05
Please help me install this Agent Skill. Skill: codex-search Source: https://github.com/shoei05/codex-search/tree/main Command: npx skills add https://github.com/shoei05/codex-search --skill codex-search-shoei05
This Skill removes the overhead of running web searches repeatedly by automatically collecting research results and saving them in multiple reusable formats with timestamps.
codex exec with a structured research prompt to produce Markdown-formatted findings..md, .json, .txt) under ~/.claude/data/codex-search/ with a UTC timestamp.Ask the skill to search for the topic and store the results by saying: use /codex-search "Latest updates on TypeScript 5.8".
High-intent search queries and answers about installing and using this skill.
You can automate web research by running Codex CLI with a structured prompt to generate Markdown findings, then automatically save the output as timestamped .md, .json, and .txt files to a configurable directory.
For academic literature queries, the research prompt instructs Codex to extract DOIs when available and explicitly mark when a DOI is not found, storing these results in structured Markdown and JSON files.
Yes, executing automated web research requires both the codex CLI and python3 installed in your environment to process queries and handle the script-based saving mechanism.
Yes, you can use the dry-run mode to execute the research prompt through Codex CLI without writing the generated .md, .json, or .txt artifacts to the output directory.
Saving each run into .md, .json, and .txt formats provides reusable structured data for engineering, product, and literature searches, allowing downstream parsing or direct reading of the timestamped findings.
The primary limitation is that DOI extraction depends on Codex finding the identifier within the source, meaning it will explicitly mark when a DOI is not found rather than inferring it from partial data.
KYRIE66nbInstall and configure oh-my-codex across project and user directories.
thucpruGenerate structured research plans and Markdown reports from input requirements.
otrebuBatch-convert JavaScript-rendered web pages to a timestamped Markdown file.
Tnemo65Search academic databases and return structured paper metadata and abstracts.