Getting Started

1. Connect GitHub

Sign in with GitHub, then install the Trace GitHub App and pick the repositories it may read. The app holds two read-only permissions (Contents, Metadata) and nothing else; Trace reads commit metadata and your README through it and can never write to your repositories.

2. Add a repository

Go to Repositories and click “Connect repo.” Select from your GitHub repos. Trace syncs the last 90 days of commits on connect.

3. Run your first analysis

Go to Project Analysis and select your repo. Click Generate to produce your first content piece. Trace analyzes your README, commit history, and project arc to find the story worth telling.

4. Find content in commits

Go to your repo dashboard. Click any commit to evaluate its story value. Commits that score high enough unlock a Create Content button.

5. Your content lives in the Content tab

Every generated piece, from both Project Analysis and individual commits, appears in the Content tab. Filter by pattern or format, expand any piece to read and copy it.

Project Analysis

Project Analysis looks at your repository as a whole, not commit by commit. It reads your README, project description, and commit history to understand what kind of project this is, who it's for, and what story it's been telling.

Commit Analysis

Commit Analysis evaluates individual commits for their story value. Not every commit is worth a post. Commit Analysis tells you which ones are.

Evaluation is free. Evaluating a commit does not count against your monthly generation limit. Only clicking Create Content counts as a generation.

The Six Patterns

Every piece of content Trace generates is classified into one of six story patterns. The pattern determines how the content is framed and what emotional angle it leads with.

Content Formats

Trace generates content in five formats. Each has a distinct structure.

Content Tab

The Content tab is your unified content library. Every piece generated by Trace, from any repo, from any analysis, appears here.

Publishing

Mark as posted

Every draft has a “Mark as posted” link beside its copy buttons. Pick the platform, paste the link if you have it, and Trace records it. The Content tab then shows the draft under Published, and “Ready to post” means saved but not posted anywhere yet. Undo removes only Trace's record, never the live post.

Publish to LinkedIn

Connect LinkedIn once, from Settings or from the “Connect LinkedIn” link on any draft. After that, drafts with a LinkedIn version show a Publish button. Publishing always asks “Post this to your LinkedIn feed now?” first; nothing is ever posted without that click. Trace stores the post link, records it under Published, and never posts the same draft twice.

LinkedIn grants access for 60 days; when it expires, Settings shows a Reconnect link. Trace asks LinkedIn for exactly one posting permission and cannot read your feed, your connections or your messages. X is manual for now: copy the draft, post it, mark it as posted.

AI tools (MCP)

Trace is also an MCP server, so Claude, Cursor and any other MCP client can ask it questions and act on your behalf within limits you set. Create a token in Settings → AI tools. Read tokens can list and read your stories, activity, usage and publications. Write tokens can additionally save a draft into Trace and mark a story as posted. No token can publish: that stays a click in the app.

Claude Code

claude mcp add --transport http trace https://www.usetrace.net/api/mcp --header "Authorization: Bearer trace_mcp_..."

Cursor and other clients

Add a remote server with the URL and an Authorization header:

{
  "mcpServers": {
    "trace": {
      "url": "https://www.usetrace.net/api/mcp",
      "headers": { "Authorization": "Bearer trace_mcp_..." }
    }
  }
}

Claude Desktop

Claude Desktop connects to remote servers through Settings → Connectors. Where a connector cannot send a custom header, use the mcp-remote bridge in its config:

{
  "mcpServers": {
    "trace": {
      "command": "npx",
      "args": ["-y", "mcp-remote", "https://www.usetrace.net/api/mcp", "--header", "Authorization: Bearer trace_mcp_..."]
    }
  }
}

Tools

trace_summarize_recent_activity, trace_list_stories, trace_get_story, trace_list_publications and trace_get_usage read your data. trace_save_draft and trace_mark_published write, and only work with a write token. Revoke a token at any time in Settings; every write is recorded in your audit log.

Your data

Trace stores commit metadata (messages, timestamps, lines changed), your README, the content it generates, and product analytics about how you use it. It never fetches or stores source code or diffs. Repository access goes through a GitHub App with read-only Contents and Metadata permissions that you install per repository.

Settings has two buttons at the bottom. Download export gives you one JSON file with everything Trace holds about you. Delete account removes it all immediately, revokes the GitHub installation, cancels any subscription and deletes your login. Analytics rows are kept for 400 days and job records for 30, then purged automatically.

Free vs Pro

Commit evaluations never count against your generation limit on any tier. Cached results are free forever; the free tier has a fair-use cap of 300 new evaluations a day, Pro has none.