Use with your coding agent
Let your agent install and walk you through selltraces, while you keep every decision.
Coding agents are good at running setup commands and explaining output. selltraces is built so an agent can do the typing while you make the choices.
The setup prompt
Paste this into Claude Code, Codex, Cursor, Gemini CLI or OpenCode:
Help me set up selltraces so I can sell my anonymized coding-agent sessions.
1. Read https://selltraces.ai/SKILL.md and follow it.
2. Install the CLI:
curl -fsSL https://selltraces.ai/install | bash
Windows: irm https://selltraces.ai/install.ps1 | iex
3. Run "selltraces login" and show me the code to approve in my browser.
4. Run "selltraces eligibility" and explain anything I still need to do.
5. Walk me through choosing sources and projects, one question at a time.
6. Run "selltraces upload --dry-run" and summarize what would be sent.
Rules: never upload, accept terms, or approve projects for me. Ask me first, every time.SKILL.md is a short guide written for agents. It lists the safe order of commands, which steps need you, and how to read --json output.
What the agent should and should not do
| The agent can | Only you can |
|---|---|
| Install and update the CLI | Approve the browser login |
Run status, eligibility, sources and dry runs | Accept the contributor terms in the dashboard |
| Explain redaction findings and exclusions | Decide which projects are approved or private |
| Suggest blocked terms, like client names | Confirm an upload plan |
The CLI enforces part of this. In a non-interactive shell, selltraces upload refuses to run unless both --confirm-policy and --confirm-upload are passed, and terms can only be accepted in the browser. Treat the rest as a rule for your agent, and tell it so.
When the command is not found
Agents often run commands in a fresh, non-interactive shell. The installer adds ~/.selltraces/bin to the startup files those shells read, but some agents cache their environment when they start. If selltraces is not found, use the full path:
~/.selltraces/bin/selltraces statusRestarting the agent session also picks up the new PATH.
Reporting problems from an agent
Agents can send structured feedback when a command behaves unexpectedly. The endpoint is described at /.well-known/agent-feedback.json. Feedback must not include session content or secrets.