home / skills / dhughes / claude-marketplace / conversation-search

conversation-search skill

safe

/plugins/conversation-history/skills/conversation-search

This skill searches indexed conversation history to locate past topics and project contexts, helping you answer when, where, or with whom discussions occurred.

npx playbooks add skill dhughes/claude-marketplace --skill conversation-search

Review the files below or copy the command above to add this skill to your agents.

Files (6)

SKILL.md

1.4 KB

---
name: conversation-search
description: Search indexed conversation history to find past conversations, locate when topics were discussed, or identify which project conversations occurred in. Use when user asks questions like "when did we discuss X?", "find conversations about Y", "in which project did we talk about Z?", or "show me conversations mentioning W". Works across all projects or within current project.
---

# Conversation History Search

Search through indexed past conversations with Claude to find relevant discussions.

## How to Search

Run the search command with the user's query:

```bash
bash ${CLAUDE_PLUGIN_ROOT}/skills/conversation-search/scripts/search.sh --query "search terms" --limit 10
```

Options:
- `--query`: Search terms (required). Supports prefix matching.
- `--limit`: Maximum results (default: 20)
- `--json`: Output results as JSON for programmatic processing

## Presenting Results

Show the user:
- Number of matching conversations found
- For each result: conversation ID, project path, date, and snippet
- Suggest using `/conversation-history:load-conversation <id>` to load full transcripts

## First-Time Setup

If search returns no results and the user seems surprised, they may need to run the initial indexing:

```
/conversation-history:index-history
```

This indexes all existing conversations from `~/.claude/projects/` into the database.

Overview

This skill searches indexed conversation history to find past discussions, identify when topics were discussed, and show which project a conversation belongs to. It works across all projects or can be scoped to the current project. Results include conversation IDs, project paths, dates, and short snippets to help you decide which transcripts to load.

How this skill works

The skill queries a prebuilt index of past conversations using prefix-matching search terms and returns the most relevant matches. Results include metadata (conversation ID, project path, and date) plus a snippet from the conversation. If no results appear, the index may need to be built or refreshed using the provided indexing command.

When to use it

When you need to know when a specific topic was discussed
To find all conversations that mention a person, feature, or bug
To locate discussions belonging to a particular project
When you want quick snippets before loading a full transcript
To audit or review past decisions across multiple projects

Best practices

Use concise, focused search terms for faster, more relevant results
Limit results to a manageable number if you only need the top matches
If searches return nothing, run the initial indexing command to populate the database
Combine names, keywords, and date ranges when possible to narrow results
Load the full conversation by ID when the snippet looks relevant

Example use cases

Ask 'when did we discuss the deployment rollback?' to find the date and context
Search for 'API key rotation' to list all conversations mentioning that phrase across projects
Query 'client X onboarding' to see which project folder held those discussions
Run a quick search for a colleague's name to locate decisions they participated in
Preview snippets to decide which full transcripts to open for meeting notes

FAQ

What if the search returns no results?

If you expect matches but see none, run the indexing command to index existing conversations, then retry the search.

Can searches be limited programmatically?

Yes — you can specify a limit on results so your application processes only the top matches.