Nothing to hand? Load the
— a 600-word instructions file that opens with a persona, an
applyTo: **/*.tsx written without quotes so the frontmatter never parses, a
testing file scoped to "**", a prompt that asks for a pull request from
ask mode, a chat mode with tools: [] and no description, a rule that
contradicts one in another file, and an mcp.json with a literal token — or
the , where the
correct verdict is ready-to-commit and the useful output is what to add next.
Paste the kit — the prescan is free
No upload, no AI: the prescan splits your paste on # file: markers, reads each
file's YAML frontmatter key by key, parses mcp.json for real, splits the
Markdown into atomic rules and extracts every ${variable}. Then it lists what
it found — every file, rule, glob, variable, tool and server — and the flags:
frontmatter that is missing or never closed, an applyTo glob written without
quotes (a YAML scalar cannot start with *, so the file silently loses its
scope), a scoped file applying to **, two files claiming the same scope, globs
that cannot match, a prompt file with no mode that asks for actions,
tools: [], missing descriptions, frontmatter keys Copilot does not read,
${variables} the editor never substitutes, rules stated twice, rules that
contradict each other, unenforceable phrasing, persona padding, files big enough to tax
every request, paths outside the repository, an AGENTS.md with no commands,
and MCP servers that are unpinned, plaintext or carrying a literal credential. Each group
explains why it matters. This part costs nothing and happens while you type.
The AI reviews it — this is the metered part
A senior developer-experience engineer's pass over the whole kit: a posture with the single most important change named, the inventory with each item's role, and prioritized findings across scope, frontmatter, effectiveness, conflict, safety and hygiene — each with the problem, the concrete impact, the fix and a corrected fragment. Every prescan flag is confirmed or explicitly set aside, and findings may only cite files, globs and rules that actually appear in your paste. Pricing is honest: a worst-case amount is reserved before the run and only what the run actually uses is charged — the meter next to the button shows both.
Take the rewritten file, then re-check for free
The headline output is a whole .github/copilot-instructions.md, sectioned,
with every unenforceable line either made falsifiable or dropped — and it is measured
in your browser rather than trusted: word count against the file you pasted, section count,
and whether any unenforceable phrasing survived. Alongside it: every corrected fragment in
one paste-ready block, the findings as PR review comments with ```suggestion
fences, a tickable checklist for the ticket, the findings as CSV, and Markdown or JSON
export of the whole review. Then start fixing: the prescan re-runs as you edit and the strip
above the run button counts the flags you have cleared, the ones still open and any you have
just introduced — in the browser, for free, before you pay for a second review.
Reviews are saved to your SkillSafe account when you are signed in, so they follow you to
another machine; restore puts the kit back in the form too.
Derived from the @github/awesome-copilot skill (MIT). Not affiliated with GitHub.