A better starting point
for your next idea.

Use the commit messages and diff I attach. Draft Added, Changed, Fixed and Known limitations sections, omitting empty sections. For each claim, cite the file or commit supporting it. Distinguish code changes from behavior verified by tests. Do not invent test results, deployment status, speed improvements or bug reports. Exclude secrets and internal identifiers from reader-facing copy. Finish with questions that require developer confirmation. If only a commit title is available, label its behavior unverified.

Read the invoice lines I paste or attach. Extract description, quantity, unit price and printed line total without guessing missing fields. Recalculate quantity × unit price for each fully specified line. Show a table of printed total, calculated total and difference. Sum the calculated line totals separately from tax, shipping and discounts. Apply those adjustments only when the source specifies their amount or exact rule. Flag ambiguous currency and rounding. Do not assert the invoice is valid or paid. Ask me to check the arithmetic in a calculator before acting.

Plan exactly three short shots of {{subject}} in {{setting}}. Define a continuity lock with colors, markings, props, direction of motion and lighting that must stay fixed. Create a table with shot number, visible action, camera distance, camera move, continuity checks and a standalone generation prompt. Keep actions physically simple. Do not promise that a generator will preserve identity or create exact timings. End with a low-cost review order: generate one shot, inspect the fixed details, then decide whether to continue.

Translate the attached or pasted SRT captions into {{target_language}}. Keep every cue number, start time, end time and blank-line boundary exactly as supplied. Translate dialogue only; preserve proper names unless I supply an approved glossary. Do not merge or split cues, add timings, or silently repair malformed input. Report malformed cues separately and ask for clarification. Return the SRT in a code block, followed by a cue-count check and lines that may be too long. Translation can change reading length; it needs a timed playback review before publication.

Inspect the image I attach. Draft one concise alt text sentence that describes the main subject, relevant action and spatial relationship. Do not infer a real person's identity, exact location, emotion or text you cannot read. If an image is a graph, list its visible labels and values separately rather than inventing data. Then provide a short uncertainty note for details that are unclear. Ask for the image if missing. Do not add "image of" unless it helps the context.

Use only the transcript I paste or attach. Turn it into a short brief with Goal, Deliverables, Constraints, Open questions and Next action. First extract names, dates, amounts and negations verbatim into a protected-facts list. Keep tentative statements tentative. Remove fillers but do not add owners, deadlines or promises. If two statements conflict, retain both under Open questions. End with a table showing each important fact and its supporting source phrase. If no transcript is present, ask for it.

Help me review a small content experiment without mistaking correlation for causation.
Goal: {{goal}}
Posts and available metrics: {{metrics}}
What changed between posts: {{changes}}
Audience or timing context: {{context}}
Summarize observations, plausible explanations and limits. Pick one variable to test next week and define a success measure tied to the goal. Include a simple experiment log with hypothesis, change, measurement and stop rule. Do not invent benchmark numbers, claim statistical significance from a tiny sample or optimize views when the goal is useful engagement. If data is missing, propose measurement first.

Review a small AI-assisted app before I share it.
App purpose: {{purpose}}
Main user flows: {{flows}}
Stack and repository notes: {{stack}}
Known issues: {{known issues}}
Make a prioritized manual test plan covering empty states, invalid inputs, failed requests, keyboard access, narrow screens and saved-data persistence. For each test, state setup, action, expected result and severity if broken. Separate confirmed defects from untested risks. Do not claim to have inspected code or executed tests unless you actually did. Finish with a short release checklist and the smallest fixes for any confirmed blockers.