Prompt Modifiers
Output Format Variants (Modifier)
A modifier to paste in with any audit prompt to change its scope, depth, evidence rules, or output format.
- Best for
- Appending to any audit to receive its findings in the shape a workflow needs — pull request comments ready to paste or post one finding at a time, tickets with title, labels, priority mapped from severity, description, acceptance criteria, and technical notes, a chat summary of a few lines for a channel, an executive report with risk, impact, and recommendations and no code, a machine-readable list for tooling, or a checklist for a release gate — with the same findings and severities under every format and the rule that the format never drops evidence
- Use when
- Findings must become tickets, review comments, a channel update, or a leadership summary; an agent or script will consume the report; a release checklist needs the findings as checks; the same audit feeds several audiences; or a report arrived in a shape the team could not act on
Append this to any other prompt and name the format you need.
You are operating with Output Format Variants. The base prompt's audit runs unchanged and its findings, severities, and evidence are fixed. This modifier changes only the shape of the deliverable. If no format is named, produce the base prompt's own format.
Available Formats
Pull request comments. One finding per comment, ready to paste into a review or to post with the repository tool when asked: a severity tag, a one-line title, the file and line, two or three sentences on what is wrong and why it matters, and a suggested fix as a code block where a code change is the fix. Group by file in the order of the diff.
Tickets. Each finding as a standalone ticket: title prefixed with the severity, labels (security, performance, accessibility, ux, data, reliability, bug), priority mapped from severity (Critical to the highest priority, then High, Medium, Low in order), a description of what is wrong and why it matters, acceptance criteria as checkboxes including the test or check that proves the fix, technical notes with files, lines, and the evidence, and an estimate band (small, medium, large). Related findings reference each other by title.
Chat summary. Three to six lines for a channel: the subject audited, the count by severity, the single most important finding with its location, the first action, and a link or path to the full report. No code, no tables.
Executive report. For decision makers: one paragraph on overall risk, a short list of the top findings with business impact and a plain-language recommendation each, the investment needed in bands, and what is not at risk. No code, no file paths, severities expressed as risk.
Machine-readable list. One record per finding with fields: id, severity, category, title, file, line, summary, evidence, fix, confidence. Stable field names, one record per line, no prose outside the records.
Release checklist. Each finding as a check with a box, the condition that must be true, the command or test that verifies it, and the owner; Critical and High checks marked as blocking.
Rules Under Every Format
The findings, severities, and evidence are identical across formats; a format may omit detail for its audience but never changes a severity or drops a Critical or High finding. Every format links or points to the full report. Where the audience should not see internals (executive report), the evidence is summarised, not removed from the underlying report. Several formats may be requested in one run; produce each in full.
What This Mode Does Not Change
Scope, methodology, severity, and evidence are the base prompt's. Combined with other modifiers, their rules on scope and evidence apply to the audit; this modifier governs only the deliverable's shape.
Output Format
Produce the named format exactly as specified above. If several formats are named, produce each under its own heading, in the order requested, each complete on its own.
Want this applied to a live stack?
See the project work behind these tools, or start a conversation if you want help using one in context.