Ingest Claude Code conversation history into the Obsidian wiki. Use this skill when the user wants to mine their past Claude conversations for knowledge, import their ~/.claude folder, extract insights from previous coding sessions, or says things like "process my Claude history", "add my conversations to the wiki", "what have I discussed with Claude before". Also triggers when the user mentions their .claude folder, Claude projects, session data, past conversation logs, local-agent-mode session
Detected risks:
The claude-history-ingest skill extracts valuable knowledge from your past Claude Code conversations and organizes it into your Obsidian wiki. When you work with Claude through the CLI or desktop app, all those coding sessions, problem-solving discussions, and accumulated insights are stored locally in JSONL files. This skill mines that conversation history to surface reusable knowledge, patterns, and solutions that would otherwise remain buried in session logs.
The skill intelligently processes two data sources: CLI sessions stored in ~/.claude/projects/ and desktop app agent sessions from ~/Library/Application Support/Claude/local-agent-mode-sessions/. It reads conversation transcripts, memory files (user profiles, project context, workflow feedback), and audit logs to extract structured knowledge. The system uses a manifest-based approach to track what's been ingested, supporting both incremental updates (append mode) and full rebuilds. This prevents duplicate processing while ensuring new sessions are captured efficiently.
Ideal for developers who want to build a personal knowledge base from their AI-assisted coding sessions. Whether you're documenting recurring patterns, preserving solution approaches, or building a searchable archive of technical discussions, this skill transforms ephemeral conversations into permanent, organized documentation. It's particularly useful for consultants working across multiple projects, teams maintaining institutional knowledge, or individual developers who want to learn from their past problem-solving approaches.
Append mode (default) only processes new or modified files since the last ingest by checking the manifest. Full mode reprocesses everything regardless of previous ingestion. Use append mode for regular updates and full mode after a wiki rebuild or when explicitly needed.
Claude Code stores data in two locations: CLI sessions in ~/.claude/projects/ with JSONL conversation files and memory folders, and desktop app agent sessions in ~/Library/Application Support/Claude/local-agent-mode-sessions/ with deeply nested conversation transcripts and audit logs. The skill scans both locations.
The skill requires OBSIDIAN_VAULT_PATH and CLAUDE_HISTORY_PATH (defaults to ~/.claude) configuration. It follows the Config Resolution Protocol by checking for .env files walking up from the current directory, then ~/.obsidian-wiki/config, or prompts for setup if not found.
The skill processes JSONL conversation transcripts, memory files (MEMORY.md, user profiles, feedback, project context), session metadata JSON files, and audit logs from local agent mode. It extracts knowledge from all these sources to build comprehensive documentation.
The skill maintains a .manifest.json file at the vault root that tracks which files have been ingested and their modification times. In append mode, it skips files already in the manifest unless their modification time is newer, ensuring only new or updated content is processed.
Quick Setup:
.claude/skills/Repository
ar9av/obsidian-wiki