memex

The memex skill is designed for efficient transcript searching, catering to both humans and AI agents. It supports Claude Code, Codex CLI, and OpenCode, making it a versatile tool for developers and AI practitioners. With its intermediate complexity, memex allows users to quickly locate specific information within transcripts, streamlining the process of data retrieval. This skill is particularly beneficial for teams that rely on extensive documentation or communication logs, enabling them to save time and enhance productivity. One of the key benefits of memex is its ability to significantly reduce the time spent searching for relevant information. By leveraging advanced search capabilities, users can quickly pinpoint crucial data points, thus accelerating decision-making processes. Although the exact time savings are not quantified, the efficiency gained from rapid transcript searches can lead to substantial improvements in workflow automation. This skill is ideal for developers, product managers, and AI practitioners who require quick access to information for project development and management. Memex is particularly useful in scenarios where teams need to analyze meeting notes, customer interactions, or any form of recorded dialogue. For example, a product manager can use memex to swiftly find feedback from past meetings, enabling them to make informed decisions without sifting through lengthy transcripts. Similarly, AI agents can utilize this skill to enhance their responses by accessing relevant historical data, thus improving user interactions. Implementing memex is straightforward, with an estimated time of 30 minutes to get started. Its intermediate difficulty level means that users should have some familiarity with automation tools and coding practices. By integrating memex into AI-first workflows, organizations can enhance their operational efficiency, ensuring that both human and AI agents can work seamlessly together to achieve optimal results.

1. **Install and Configure Memex**: Install the Memex tool (e.g., via `pip install memex` or `npm install -g memex`). Configure it to index your session transcripts by running `memex init` and pointing it to your transcript directory. 2. **Run a Search Query**: Use the prompt template above in your AI assistant (e.g., Claude Code or Codex CLI) by replacing the placeholders. For example: ``` Search the memex transcript archive for all sessions where 'API rate limiting' was discussed. Return the top 5 most relevant sessions with timestamps, participant names, and a 1-2 sentence summary of each. Focus on sessions where a decision was reached. Format the results as a markdown table. ``` 3. **Refine Results with Filters**: If the initial results are too broad, add filters to narrow down the scope. For example, search only sessions from the last 30 days or filter by specific participants: ``` Search the memex transcript archive for sessions from the last 30 days where 'auth service' was discussed. Include only sessions with Alex or Morgan as participants. ``` 4. **Export or Integrate**: Copy the results into your project documentation or use the Memex API to programmatically fetch sessions for further analysis. For example, use `memex export --format json --output sessions.json` to save results for later review. 5. **Automate with Agents**: Set up an AI agent (e.g., using OpenCode) to periodically search Memex for specific topics (e.g., 'bug fixes' or 'deployment decisions') and summarize findings in a Slack channel or email digest.