Prompt linter

Check a prompt before any call is made. Each warning names the failure it tends to cause.

from counterfix import lint

for issue in lint("Return ONLY JSON with keys name and age for: Priya, 29."):
    print(issue.rule, issue.title, "->", issue.suggestion)

counterfix lint prompts/*.txt --fail-on-warning      # in CI
counterfix lint prompts/*.txt --judge                # also L2, L5, L6

Files: .txt / .md (one prompt), .json (a message list, or {"messages", "params"}), .jsonl (one such object per line).

Rule	Flags	Tends to cause	Needs a judge
L1	A constraint buried in the middle of a long prompt	F6	no
L2	Instructions that conflict ("be brief", "explain fully")	F6, F7	yes
L3	A factual question with no "I don't know" option	F4, F5	no
L4	JSON / YAML / CSV asked for in words only	F2	no
L5	System prompt and user message disagree	F6, F3	yes
L6	Request missing who, what or which	F9	yes
L7	Pasted outside text with no delimiters	prompt injection	no