Who it is forA developer whose agent hits a bug or rough edge in an npm dependency and wants the maintainers to receive a structured report.
The problemFriction caused by a dependency stays in the local repository, so the project that could fix it never hears about it.
The outcomeA friction entry targeted at an opted-in project, scaffolded from that project's GitHub issue form.
What you need first
- A GitHub repository for the project that the agent works in
- npm, pnpm, bun, or yarn, or the standalone executable from the install script
- A choice between the Frog GitHub App and Action-only mode, made with the user before access is changed
- A PostgreSQL connection string, only if you want a shared database store
Steps
- Run
frog targetsto list the dependencies that accept friction reports. - Run
frog log --target viemwith the dependency name from the list, replacing viem with your target. - Fill in the prompts so the scaffolded entry answers the target project's issue form questions.
- Run
frog listto check that the entry shows its target. - Run
frog publishto report the pending entry to the target project.
Prompt for your agent
In the frog repository, run `frog targets` and then `frog log --target <package>` for the dependency that caused the friction. Expected output: a targeted entry scaffolded from that project's issue form. Verify by running `frog list` and checking that the entry shows the target package.
Caveat. A dependency only receives reports if it has opted in through its repository, so a target missing from `frog targets` cannot be reported to with this workflow.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for wevm/frog and "Report friction in a dependency to its maintainers"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer