Urgent.News

What's breaking now, across thousands of outlets.

AI

How to Search Your Claude Code Conversation History

You talked through an approach with your agent a few days ago. The agent explained the tradeoffs, you picked a direction, and it wrote the code. Now you need that reasoning again — maybe for a code review, maybe because you're hitting an edge case the original conversation anticipated. You know it's somewhere. Claude Code saves every conversation as JSONL under ~/.claude/projects/ . Codex groups…

Searching your Claude Code conversation history can be challenging since the code saves every conversation as JSONL under the ~/.claude/projects/ directory. While you can retrieve the full record of prompts, reasoning, and tool calls, finding specific conversations among the stored data requires specialized tools or methods. The introduction of PonyMux 0.10 brings full-text search capabilities directly into the interface you already use for working with agents.

By pressing Cmd+K within the interface, you can search for keywords in both terminal names and conversation content. This search spans every prompt, response, and tool usage recorded during your interactions. The search results highlight matched snippets, show the session title, agent type, and last activity time, making it easy to identify the relevant conversation.

The search index is built separately from your metadata and Terminal data, stored in a SQLite database named search.db, and updates automatically as new conversations are generated. A notable change includes the display of the last conversation within a sleeping Terminal, providing a more informative snapshot of the agent's previous activities.

This enhancement allows you to resume conversations from previous sessions rather than just the most recent one, significantly improving the workflow when investigating specific discussions or debugging. Although the search functionality is limited to locally stored conversations and has a 30-day retention window, it significantly streamlines the retrieval process for archived interactions without relying on external tools or services.

Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at dev.to →

More in AI

Stop Making AI Agents Grind Through Huge Codebases — I Built a Deterministic Wiki Build System for Them

Joining an unfamiliar codebase, what you want isn't more code — it's a wiki that explains how the thing actually works: which modules exist, where the boundaries are, how a request travels from entry…

  • Repowiki tool creates structured wiki for any repository
  • Designed to handle large codebases with deterministic CLI commands
  • Ensures reliability through validation and wiki-as-code output

Why 'monitoring' isn't enough for AI agents — and how I made delegation cryptographically verifiable

The problem nobody talks about with AI agents We're rushing to give AI agents autonomy. An orchestrator agent calls a research agent, which calls a writer agent, which calls a tool.

  • Monitoring AI agents alone insufficient, provides only claims, not proof
  • Delegations between agents must be cryptographically signed, not just logged
  • Verified delegation chain enables proving accountability, answering "prove it"

More from Tuesday 22 September →