session-search

The session-search skill enables CLI agents without subagent support (e.g., Codex CLI) to search through previous code agent sessions, helping users find specific work, decisions, or code patterns. This skill solves the problem of quickly locating relevant information from past sessions to inform ongoing development without manual sifting through numerous files. By automating the search process, users can quickly identify key insights from previous discussions and decisions, enhancing their workflow efficiency and decision-making processes.

The skill operates by querying stored session data using the `aichat search` command, which can be customized with filters like project or date. Once the query is run, the results are parsed and summarized into a concise report, including key findings, relevant sessions, and specific content such as code snippets or decisions made. The output is designed to help users easily review past work and decisions related to specific topics, such as code patterns or architectural designs.

This skill is primarily targeted at developers and software engineers working on projects where multiple sessions are involved, such as large codebases or ongoing software design efforts. It is particularly useful for those needing to recall or analyze past discussions about code architecture, decision-making, or coding practices, without the need for deep, manual investigation of past session files.