Auto-file Svelte mistakes as GitHub issues
They added an opt-in skill that turns concrete Svelte mistakes into draft or filed issues in the Svelte AI tools repo.
- What they did
- They installed the svelte-mcp-feedback skill on its own, since no bundled plugin included it. When the agent made a Svelte or SvelteKit mistake, they corrected the code, asked it to build a minimal reproduction, and, if Svelte MCP was available, had it run the autofixer on that reproduction. The skill then checked for existing reports and either filed a GitHub issue in sveltejs/ai-tools with the GitHub CLI or opened a prefilled issue page.
- How it went
- The skill is set up to skip vague complaints, unrelated bugs, and style preferences, and to report only clear mistakes that could help others. Automatic filing depends on the GitHub CLI being installed and logged in; otherwise the user still has to submit the prepared form themselves.
- Tips
- Use it only for specific, repeatable errors, not for preference changes. If you want automatic filing, confirm the GitHub CLI is available and authenticated first.
Try it yourself with Claude Code
Set up an opt-in feedback skill for [framework or library] that watches for concrete mistakes your agent makes while helping with it. When a mistake happens, fix the code, create a minimal reproduction, compare it against existing issues in [repo name or URL], and prepare a GitHub issue with: what I asked for, what went wrong, the correct behavior, and the smallest example that shows the bug. Pause and show me the draft before any submission, and do not continue until I approve. If [tool required for filing] is installed and authenticated, use it to file the issue; otherwise open a prefilled issue form for me to review and submit myself.
Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.
Get 3 agent uses like this every morning
With the day's top AI news. Free, and one click to unsubscribe.
Related agent uses
Build a staged agent pipeline for PRs
Create a multi-step system that assigns issues, tests changes, and leaves the final merge to a person.
Break one goal into a sequenced coding plan
Ordewell turns a single objective into a step-by-step task list for coding agents, with dependencies, review, and parallel runs.
Shrink an AWS cost-triage MCP server
They rebuilt a read-only AWS cost checker with far fewer tools and cut the calls needed to answer a waste question from seven to two.
Build a terminal calling tool with human backup
They wired a CLI agent to place calls and loop in a person only when it needed an answer.
Test firmware in simulation before using hardware
Simantic lets agents run firmware against virtual devboards, with a reviewer checking results before any real board is needed.
Use parallel agents to build a database engine
They split a SQLite implementation across several coding agents working in separate branches, then merged the pieces after repeated test-and-fix loops.