Platforms

What each data source gives Syke.

Claude Code

Why it matters: Claude Code sessions are the richest signal. They capture what you’re building, how you think about problems, and what tools you reach for.

Method: Local JSONL parsing from ~/.claude/. Dual-store adapter reads both project-level and global transcript stores.

Data: Sessions, tool calls, projects, git branches, conversation flow.

syke --user you ingest claude-code --yes

ChatGPT

Why it matters: Research threads, brainstorming sessions, learning patterns. Often captures interests and curiosities that don’t appear in code.

Method: ZIP export parsing (Settings > Data Controls > Export).

Data: Conversations, topics, timestamps, message flow.

syke --user you ingest chatgpt --file ~/Downloads/chatgpt-export.zip --yes

GitHub

Why it matters: What you actually ship. Commit patterns, project scope, tech stack preferences.

Method: REST API (public data, or with GITHUB_TOKEN for private repos).

Data: Repos, commits, issues, PRs, stars, README content.

syke --user you ingest github --username your-username

Gmail

Why it matters: Communication patterns, professional context, who you interact with.

Method: OAuth API (via gog CLI or Python OAuth fallback).

Data: Subjects, snippets, labels, sent patterns.

syke --user you ingest gmail

MCP Push (Any Client)

Any MCP client can push events back to Syke. Your Claude Code session can log what you’re building. Your custom agent can log decisions.

{
  "source": "my-agent",
  "event_type": "decision",
  "title": "Chose PostgreSQL over MongoDB",
  "content": "Selected PostgreSQL for the new API..."
}