Claude Code Plugins

Community-maintained marketplace

Feedback

load-conversation

@dhughes/claude-marketplace
0
0

Load the full content of a previous Claude Code conversation into current context. Use when user asks to "load conversation <uuid>" or "show me conversation <uuid>" or references loading/viewing a past conversation by its ID.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name load-conversation
description Load the full content of a previous Claude Code conversation into current context. Use when user asks to "load conversation <uuid>" or "show me conversation <uuid>" or references loading/viewing a past conversation by its ID.
allowed-tools Bash, Read

Load Conversation Skill

Load the complete transcript of a previous Claude Code conversation, providing full context of the user-assistant dialogue and files involved.

How to Load a Conversation

1. Parse the Conversation ID

Extract the conversation UUID from the user's request. It should be a UUID format like:

  • e06a3702-af08-41d7-a425-403622c2f266
  • f792926c-99b9-44ca-8733-e9f9db276da2

2. Execute the Loader Script

The loader script is located in this skill's scripts/ directory. Execute it:

cd <skill-base-directory>/scripts && ./load-conversation.sh "<conversation-uuid>"

Note: The <skill-base-directory> is automatically provided by Claude Code as the base directory for this skill.

3. Present the Conversation

The script will output:

  1. Header: Conversation metadata (ID, project, date, message count)
  2. Transcript: Full conversation flow with User and Assistant messages
  3. Files Reference: List of unique files that were read or edited during the conversation

4. Output Format

Present the entire output from the script to load the conversation context. The transcript format is:

=== CONVERSATION ===
ID: <uuid>
Project: <project-path>
Date: <timestamp>
Messages: <count>

=== TRANSCRIPT ===

[User]:
<message content>

[Assistant]:
<message content>

[User]:
<message content>

...

=== FILES ACCESSED ===

Read:
- /path/to/file1.rb
- /path/to/file2.rb

Edited:
- /path/to/file3.rb

Important Notes

  • The conversation file must exist at ~/.claude/projects/<encoded-path>/<uuid>.jsonl
  • If the file is not found, inform the user that the conversation doesn't exist
  • The transcript includes only user and assistant messages, not tool execution details
  • File references help understand which code was involved without cluttering the transcript

Examples

User: "load conversation e06a3702-af08-41d7-a425-403622c2f266" → Execute loader script with the UUID, display full transcript

User: "show me what we discussed in f792926c-99b9-44ca-8733-e9f9db276da2" → Execute loader script with the UUID, display full transcript

User: "can you load the conversation from earlier where we worked on retry_before_failing?" → User should first use conversation-search to find the UUID, then load it