Custom GPT tools
All 100 prompt-only tools are live inside ChatGPT. Browse the complete catalog here, inspect every public prompt, or use one of the seven detailed guides before sending a file or text.
These pages give web search engines useful public context. Inside the GPT Store, discovery depends on the GPT listing, category, platform signals, and the current Store index. A public page cannot guarantee a Store rank.
Authenticated readback on 29 July 2026 confirmed 100 live listings: 65 in the GPT Store and 35 public by direct link. Link-public tools are usable but are not claimed as Store-discoverable. Exact-title rank, web indexing, anonymous access, and guaranteed output quality remain unproven.
Inspect all public promptsThese evidence-backed pages explain the input, output, privacy boundary, limitations, and example prompts. The remaining tools stay in the single catalog below instead of becoming thin doorway pages.
A narrow Custom GPT for faithful transcription. It preserves visible structure, labels uncertainty, and returns the text before any explanation.
A focused Custom GPT for turning visible tables, dashboard rows, and structured lists into reusable data without completing cropped cells.
A one-job Custom GPT with a fixed 16-column schema for visible receipt and invoice data. Missing fields stay blank and obscured fields are marked.
A narrow formula assistant that diagnoses one formula at a time, preserves locale and references, and uses placeholders when workbook details are missing.
A focused Custom GPT that separates discussion, proposals, decisions, and explicit tasks while keeping unstated owners and due dates visible.
A focused Custom GPT for finding interface footguns in supplied API contracts, endpoint specs, schemas, and configuration shapes without claiming runtime access.
A focused Custom GPT that scores only the supplied page evidence, rewrites the core message without inventing proof, and ends with three prioritized fixes.
Search all 100 live GPTs by name, task, or category. Every result links to ChatGPT and its complete public prompt.
Showing 100 of 100 live GPTs.
Extract text from photos, screenshots, scans, and signs. Preserves layout and marks unreadable or uncertain text instead of guessing.
Turn screenshots of tables, dashboards, and lists into clean Markdown, CSV, or TSV without inventing cropped cells.
Transcribe handwritten notes into clean typed text while preserving line breaks and clearly flagging words that are hard to read.
Convert receipt and invoice photos into consistent CSV rows for line items or transaction summaries, including tax, tip, total, and payment details.
Write concise accessibility alt text for images, screenshots, charts, and social posts without guessing identity or intent.
Analyze customer interview transcripts or notes into pains, jobs, objections, evidence-backed quotes, follow-up questions, and one next experiment.
Upload a B2B landing-page screenshot or paste its copy. Get a focused review of clarity, proof, objections, CTA, rewrites, and the top three fixes.
Fix grammar, spelling, and punctuation while preserving your meaning, tone, formatting, names, numbers, links, and language.
Turn your resume facts and a job description into a concise, tailored cover letter without inventing experience, metrics, or qualifications.
Turn a meeting transcript or rough notes into a clear summary, decisions, owners, action items, due dates, and open questions without guessing.
Clean copied PDF or OCR text by removing repeated headers, page numbers, duplicate furniture, extra whitespace, cautious broken line wraps, and obvious OCR character noise.
Pressure-test a product idea for buyer, pain, workaround, proof path, distribution, rights, AI-agent fit, and a clear kill rule.
Rewrite stiff or AI-sounding text into natural prose while preserving your facts, intent, tone, and voice. Flags when the real problem is structure or evidence, not style.
Turn a pasted email thread and your intent into one concise, copy-ready reply in short, warm, firm, or executive tone without inventing dates, promises, or attachments.
Tailor your resume to a job description with a grounded summary, stronger bullets, and a keyword-to-evidence map without invented metrics, skills, or ATS guarantees.
Turn your notes into a concise study summary, grounded practice quiz, answer key, and optional flashcards without adding facts that are not in your notes.
Turn a visual idea or reference brief into one portable, structured English image prompt with concrete composition, palette, format, and constraints.
Fix or explain one Excel or Google Sheets formula from the formula, error, headers, and a small example - without inventing ranges or workbook data.
Turn rough bug notes, logs, or a screenshot into a clear evidence-grounded report with expected vs actual behavior, unknowns, and next verification.
Turn requirements or acceptance criteria into traceable happy-path, negative, edge, and regression test cases with explicit expected results.
Turn supplied commits, diffs, issue notes, and verification evidence into grounded user-facing release notes without inventing features or impact.
Check and repair supplied SRT or VTT subtitle text while preserving cue order, timestamps, and dialogue unless a specific change is requested.
Solve one math problem step by step from typed text or a clear image. Shows the method, states assumptions, checks the result, and never invents missing graph or diagram facts.
Explain a supplied research paper, abstract, or excerpt in plain language. Separates claims, method, evidence, limits, and unknowns without inventing results.
Paste an error, stack trace, failing command, or short code snippet. Get a plain-English cause, the smallest evidence-grounded fix, unknowns, and the next check.
Turn a topic, notes, or source text into a bounded slide-by-slide outline with titles, key bullets, speaker notes, and clearly marked factual gaps.
Turn notes, a draft, or an announcement into one concise professional social post with a clear hook, clean structure, and no invented facts, metrics, or outcomes.
Turn supplied book, article, website, journal, video, or DOI details into an APA 7 reference and optional in-text citations, with missing fields clearly flagged.
Translate supplied English text into clear Spanish while preserving meaning, names, numbers, links, and formatting, and flagging material ambiguity instead of guessing.
Explain one regex in plain English, show what it matches, flag grounded risks, and suggest the smallest safe rewrite without changing its intent.
Turn supplied product, policy, service, or onboarding text into a concise FAQ grounded only in the source, with unsupported gaps clearly listed.
Turn supplied hiring facts into one clear, realistic job description with responsibilities, requirements, nice-to-haves, and explicit placeholders for missing details.
Format, validate, and clean up pasted JSON into readable pretty JSON or compact minified JSON, while pinpointing syntax errors and never inventing missing data.
Explain what a pasted SQL query does in plain English, including filters, joins, grouping, ordering, and grounded risks, without inventing schema facts or running it.
Turn supplied product facts into one clear ecommerce listing description with concrete highlights and placeholders for missing details instead of invented claims.
Generate clear email subject lines from supplied message facts or a draft, while preserving meaning and avoiding invented urgency, offers, dates, or promises.
Turn a supplied role, level, and target skills into structured interview questions with evaluation focus and safe follow-ups, without candidate answers or hiring decisions.
Turn pasted rows, lists, or delimited text into a clean Markdown table while preserving visible values and leaving ambiguous structure unresolved instead of guessing.
Turn a supplied meeting goal, attendees, topics, and time limit into one clear agenda while flagging missing inputs and never inventing decisions or owners.
Turn a feature brief, requirement, bug note, or workflow into grounded user stories with acceptance criteria, assumptions, and explicit requirement gaps.
Turn a supplied podcast transcript, rough notes, or episode outline into grounded show notes, takeaways, chapters, link placeholders, and a concise summary.
Turn supplied identity and profile facts into one short social media bio without inventing achievements, employers, credentials, audience claims, or links.
Turn supplied repo facts, feature notes, setup steps, and usage details into a clean README draft without inventing commands, URLs, badges, license terms, or project facts.
Turn a supplied diff, commit notes, issue context, and verification evidence into a clear PR description without inventing tests, issue links, approvals, risk, deployment, or behavior changes.
Turn supplied process steps, roles, tools, and exceptions into one clear SOP without inventing policy, owners, timings, compliance claims, or missing steps.
Turn a supplied research goal, audience, and topic into neutral survey questions with suitable answer formats, without leading wording, diagnosis, or screening decisions.
Generate clear online video titles from supplied topic notes or a draft, while avoiding invented claims, results, dates, guests, trends, or fake urgency.
Turn a supplied diff, file list, or change notes into one clear version-control commit message without inventing scope, issue links, behavior changes, tests, or conventional prefixes.
Generate clear meta descriptions from supplied page facts, copy, or a draft while avoiding invented claims, offers, dates, rankings, or SEO promises.
Turn a topic, brief, notes, or source facts into a clear blog post outline without inventing claims, examples, statistics, quotes, links, or research.
Turn supplied employment facts into one clear resignation letter without inventing dates, reasons, notice terms, legal claims, handoff promises, or employer details.
Turn a supplied process, hierarchy, sequence, or relationship list into Mermaid code without inventing nodes, links, labels, directions, or system boundaries.
Convert supplied HTML into clean Markdown while preserving visible text, links, lists, headings, tables, and code blocks without inventing missing structure, URLs, or meaning.
Turn a supplied schedule into a cron expression and plain-English readback without inventing cron dialect, timezone behavior, platform compatibility, or unsupported timing rules.
Turn a supplied business report, draft, or notes into a concise executive summary without inventing recommendations, results, risks, metrics, or conclusions not supported by the source.
Write a clear thank you note from supplied context and facts without inventing gifts, favors, promises, shared history, or emotional claims not supported by the user.
Turn supplied options and decision criteria into one clear comparison matrix with visible assumptions, optional weights, and no invented scores, facts, or high-stakes final decisions.
Format supplied XML into clean, readable structure while preserving tags, attributes, text, namespaces, and order without inventing missing elements or unsafe fixes.
Add concise inline comments to supplied code without changing logic or inventing architecture, behavior, requirements, business rules, or undocumented intent.
Turn supplied project notes, milestones, blockers, and updates into one clear status report without inventing progress, owners, dates, risks, or next steps.
Generate one clear value proposition statement from supplied product or service facts without inventing customer pain, benefits, differentiation, proof, or results.
Format supplied YAML into clean, readable structure while preserving keys, values, comments, boundaries, and order except for clearly labeled redaction of obvious credentials.
Turn a supplied schema, field list, DDL, or table notes into a structured data dictionary without inventing field meanings, business rules, allowed values, or relationships.
Turn supplied workplace incident facts into one neutral report without inventing causes, blame, injuries, policy violations, legal conclusions, or missing timeline details.
Turn supplied organization, team, product, or founder facts into one clear mission statement without inventing customers, values, impact claims, or strategy.
Write one clear apology message from supplied facts without inventing blame, legal admissions, excuses, compensation, promises, forgiveness, or relationship details.
Write one compact PRD for one approved feature inside an existing product, using supplied users, behavior, constraints, and [TBD] gaps without generating backlog stories.
Review a pasted code diff or changed-files summary for bugs, regressions, security risks, and missing tests without pretending to inspect the repository or run the code.
Clarify a software or product request one focused question at a time, then produce a concise requirements brief with acceptance criteria, non-goals, and unresolved risks.
Turn supplied architecture decision notes into one clear ADR with context, options, decision, consequences, and revisit conditions without inventing rationale or approval.
Review a supplied API contract, endpoint spec, schema, or configuration for ambiguity, insecure defaults, misuse risks, and safer interface choices without claiming runtime access.
Review a supplied product launch or rollout plan for hidden assumptions, failure modes, missing gates, rollback gaps, and fragile sequencing without inventing project facts.
Design a seven-day behavior test for one already-approved product feature and hypothesis, with a schedule, signal, proposed threshold, privacy boundary, decision, and kill rule.
Review one supplied feature rollout decision through evidence, reversibility, user impact, delivery risk, and privacy, then return ship, hold, or limited-rollout.
Review an uploaded app, dashboard, form, modal, or settings screenshot for clarity, hierarchy, usability, accessibility, and visible friction with prioritized fixes.
Turn evidence from one completed engineering project into a section-by-section case study outline covering the problem, decisions, failures, verified result, and fact gaps.
Turn a supplied architecture description into a preliminary threat-model outline with assets, trust boundaries, abuse-path questions, controls, and explicit unknowns.
Review a supplied file tree, module map, or architecture note for coupling, unclear boundaries, and high-leverage refactoring opportunities.
Turn a feature or bug description into a practical red-green-refactor sequence built around observable behavior and stable test seams.
Turn project notes, chat context, decisions, and current status into a concise developer handoff that another person can resume immediately.
Classify a supplied bug or feature request, expose missing information, and produce a concise ready, needs-info, duplicate, or close verdict.
Turn one uncertain product or engineering question into a minimal throwaway prototype plan with a clear test, verdict, and stop condition.
Teach one technical concept at the learner's level through a plain-language model, worked example, quick check, and next lesson.
Explore a supplied product problem into distinct feature options, trade-offs, assumptions, and one recommended next experiment without rushing to a build.
Turn product rules, workflows, and edge cases into a clear domain model of entities, states, invariants, events, and unresolved decisions.
Review supplied frontend code for avoidable rendering, loading, bundle, and data-fetching costs, then rank the smallest measurable fixes.
Turn symptoms, logs, and recent changes into an evidence-first debugging loop with ranked hypotheses and one discriminating next test.
Turn a supplied feature specification and project context into ordered, testable implementation steps with scope, files, evidence, and rollback points.
Turn a supplied code smell or design problem into a behavior-preserving refactor sequence of small reversible steps and safety checks.
Explain pasted merge conflicts and produce a conservative resolved version that preserves both sides' supported intent without inventing surrounding code.
Check whether supplied sources actually support a claim, separating direct evidence, inference, contradiction, missing context, and unsupported wording.
Turn a feature, risk, or recent change into focused exploratory test charters with setup, tours, observations, and stop conditions.
Extract product terminology from supplied notes and create a consistent glossary that exposes synonyms, ambiguous words, and naming decisions.
Compare a completion claim with supplied requirements and fresh evidence, then return verified, partially verified, unverified, or contradicted.
Review supplied page code, copy, or screenshots for practical accessibility risks and return prioritized fixes without claiming formal compliance.
Audit supplied page titles, metadata, headings, copy, links, and crawl evidence, then produce a prioritized SEO action plan without pretending to crawl the site.
Turn a supplied product, audience, proof, and call to action into clear conversion-focused landing page copy without invented claims.
Turn supplied incident notes, timestamps, impact evidence, and follow-up gaps into a blameless postmortem without invented facts.
Plan one dependency upgrade from supplied versions, release notes, project usage, and test constraints without guessing compatibility.
Turn supplied work items, roles, and decision rights into a clear RACI matrix while exposing ownership gaps instead of guessing.
