--- product: ChatGPT Work audience: agent framework_version: 1.0.0 currentness: 2026-08-10 source_site: https://work-guide.openai.chatgpt.site default_mode: read_only loading_policy: progressive_disclosure --- # ChatGPT Work: focused six-stage agent reference Start with real work. Give ChatGPT the right context. Work through an example and correct what matters. Draft the Skill while that context is still fresh. Test it on new examples before you trust it. Automate carefully and stay in control. ## How to use this guide 1. Identify the real job, human owner, and trusted sources. 2. Read only the stage that fits the current work. 3. Prepare or explain the next useful action. 4. Verify the current source before making a factual claim. 5. Ask before sending, publishing, deploying, deleting, or changing access. ## Start with the right reference - [Complete one-time onboarding manual](https://work-guide.openai.chatgpt.site/START-HERE.md): the full public-safe operating system, guided interview, and approval boundaries. - [Complete manual in Agent View](https://work-guide.openai.chatgpt.site/agent/agent-guide): the same complete manual in its current agent-readable surface. - [Machine-readable discovery index](https://work-guide.openai.chatgpt.site/llms.txt): a map of approved public learning resources. - [Optional features and practical tools](https://work-guide.openai.chatgpt.site/features): understand available capabilities, consent boundaries, and when each tool helps. - [Remote and Voice walkthrough](https://work-guide.openai.chatgpt.site/remote): pair an approved computer and guide work from the ChatGPT mobile app. - [Focused prompt library](https://work-guide.openai.chatgpt.site/prompts): first tasks, Skills, heartbeats, Chief of Staff, and safe Site collaboration. - [Sites and build prompts](https://work-guide.openai.chatgpt.site/sites): HTML slides, dashboards, monthly close, and forecast reviews. - [Copyable Site prompts](https://work-guide.openai.chatgpt.site/download/site-prompts.md): the same five prompts as a portable Markdown file. - [Manager kickoff examples](https://work-guide.openai.chatgpt.site/manager#manager-kickoff-examples): practical fictional examples and downloadable PDF presentations. - [Detect SKILL.md](https://work-guide.openai.chatgpt.site/download/chief-of-staff/detect/SKILL.md): identify verified changes from approved sources. - [Incubate SKILL.md](https://work-guide.openai.chatgpt.site/download/chief-of-staff/incubate/SKILL.md): match existing work, owners, instructions, and approved tasks. - [Collaborate SKILL.md](https://work-guide.openai.chatgpt.site/download/chief-of-staff/collaborate/SKILL.md): follow one requested, time-limited conversation and draft an owner-led handoff. - [Reflect SKILL.md](https://work-guide.openai.chatgpt.site/download/chief-of-staff/reflect/SKILL.md): review completed evidence and propose one improvement. This shorter reference is useful after complete onboarding. Its links still require the actual account, workspace access, available capabilities, and specific human approval; a link does not grant a permission. ## Non-negotiable operating baseline 1. Organize projects around existing workstreams, owners, and trusted sources. 2. Separate global working preferences from project-level AGENTS.md instructions. 3. Use app shots, computer history, or connectors only when available and explicitly approved. 4. Create a personal or approved shared Skill after real work and corrections; test a fresh example and review it monthly. 5. Treat a heartbeat as a bounded, read-only receipt rather than proof of completion. 6. Use one optional Chief of Staff workflow with four responsibilities: Detect, Incubate, Collaborate, and Reflect. 7. Keep scheduling, outbound messages, remote changes, publication, and access behind separate explicit approval. 8. Verify what actually advanced; preserve the existing owner and ask when the evidence is incomplete. ## 1. Start with real work Start with one real job you repeat. Show ChatGPT what good looks like, which sources to trust, and what you need to approve. ### How to do it 1. Choose one real piece of work you already expect to repeat. 2. Explain who needs it, why it matters, and what a good result looks like. 3. Show which original files and sources to trust. 4. Say what ChatGPT can help prepare and what you still need to approve. ### Example Say you prepare a monthly review. Start by showing ChatGPT last month's review, the current source files, who will read the update, and what you need to check before sharing it. ### Human approval You decide what the work means, who receives it, and whether anything can be sent, published, deployed, or shared. ### Useful human guides - Get real work done with ChatGPT: https://work-guide.openai.chatgpt.site/guide/codex-as-work-hub - Build one reliable workflow in 30 days: https://work-guide.openai.chatgpt.site/guide/thirty-day-installation ### Common mistakes - Starting with a tool or a complicated system instead of a real job - Assuming one good answer means the work is now repeatable - Letting speed replace your decision to review or share ## 2. Give ChatGPT the right context Give ChatGPT the same useful background you would give a teammate: the goal, trusted sources, a real example, what to avoid, and what you need to approve. ### How to do it 1. Give the recurring work one clear project or home. 2. Share the goal, owner, audience, and what is out of scope. 3. Show which original sources to trust and include one real example. 4. Say which decisions and actions still need your approval. ### Example Keep your monthly review in one project with the trusted source files, an example of a good review, the current checklist, and a short update on where things stand. A new conversation can pick up from there. ### Human approval ChatGPT can organize and summarize the information you provide. It cannot replace the original source or share information more broadly without your approval. ### Useful human guides - Keep recurring work in one place: https://work-guide.openai.chatgpt.site/guide/durable-projects-and-task-architecture - Give ChatGPT Work the right project context: https://work-guide.openai.chatgpt.site/guide/agents-folders-and-proof - Pick the work back up where you left off: https://work-guide.openai.chatgpt.site/guide/continuity-and-recovery ### Common mistakes - Writing a Skill before sharing the real work and its context - Pasting every old conversation into a new request - Treating remembered information as proof that a fact is still current - Keeping the same work in several competing places ## 3. Build a workflow you can repeat Work through one real example, correct what matters, and capture the useful method while the context is still fresh. Test it on new examples before calling it reliable. ### How to do it 1. Do one real example with ChatGPT before trying to document the whole process. 2. Correct the sources, decisions, quality, and approval boundaries as you work. 3. Separate useful rules from details that belong only to this example. 4. Capture what you learned while the real example and corrections are still in context. ### Example Work through this month's review using the actual sources. Correct the source order, flag missing information, and rewrite the summary for its audience. Capture those decisions before moving on to a fresh conversation. ### Human approval ChatGPT can help with the steps you have agreed on. New exceptions, important decisions, and anything that affects someone else come back to you. ### Useful human guides - Know when a workflow becomes a Skill: https://work-guide.openai.chatgpt.site/guide/artifact-to-automation - Let ChatGPT help you keep track of work: https://work-guide.openai.chatgpt.site/guide/personal-chief-of-staff ### Common mistakes - Writing a Skill upfront before ChatGPT has seen the real work - Leaving important corrections behind when the useful context is lost - Treating a first draft as proof that the work will succeed next time ## 4. Turn repeatable judgment into a Skill Draft the Skill after you have worked through a real example and corrected the judgment, while that context is still fresh. Validate the draft on fresh examples before you trust or reuse it. One good result is not enough. ### How to do it 1. Choose real work that repeats and matters when it goes wrong. 2. Give ChatGPT Work the sources, context, audience, and example you would give a teammate. 3. Work through the example and correct the decisions, quality, and approval boundaries. 4. Draft the Skill here, while the real work and your corrections are still fresh in context. 5. Test the draft on multiple real examples with fresh information; improve it before you trust or reuse it. ### Example After working through a real monthly review, ask ChatGPT Work to draft a Skill that remembers the trusted source, audience, quality checks, and your corrections. Leave this month's numbers out. Then run the draft against a new review before deciding that it works. ### Human approval You decide when a draft is ready to test, whether the results are reliable, and whether any action beyond preparing the work is approved. ### Useful human guides - Know when a workflow becomes a Skill: https://work-guide.openai.chatgpt.site/guide/artifact-to-automation ### Common mistakes - Writing the Skill upfront, before ChatGPT has seen the real work - Losing the useful context before drafting what the example taught - Treating one good result or a first draft as a validated Skill - Baking one example's facts or files into instructions meant to be reused - Trying to fix a broken source or tool by adding more instructions ## 5. Automate carefully Automate only after the workflow has worked across real examples. Start with a small, reviewable step and keep the important decisions with a person. ### How to do it 1. Choose one workflow you have already tested more than once. 2. Start with the smallest action you can review and undo. 3. Keep sending, publishing, access changes, and other sensitive actions under human approval. 4. Check what happens when a source is missing, outdated, duplicated, or unavailable. ### Example A weekly review can collect the right files and prepare a draft for you. You still check the result, decide who should see it, and approve anything that gets sent. ### Human approval Sending, publishing, deploying, deleting, changing access, and other consequential actions still require your explicit approval. ### Useful human guides - Let ChatGPT Work help without giving up control: https://work-guide.openai.chatgpt.site/guide/bounded-proactivity - Know what is current, checked, and approved: https://work-guide.openai.chatgpt.site/guide/currentness-lifecycle-approvals ### Common mistakes - Automating a process you have not actually tested - Treating urgency as permission to act - Running the same action twice without checking whether it already happened ## 6. Connect what works Connect the workflows that already work. Keep each project's owner, trusted sources, next steps, and required approvals clear. ### How to do it 1. Give each kind of work one clear project and owner. 2. Pass only the relevant context between projects. 3. State what is finished, what is current, and what still needs approval. 4. Notice repeated friction and improve one useful part at a time. 5. Build a Site when a recurring workflow needs one clear place to review, decide, and act. ### Example A team can connect planning, analysis, meetings, and presentations in a working Site without copying every source into one project. A clear header, tabs, useful toolbar, and grounded mascot help people review the work and find the next decision. Each owner keeps control of the sources and approves the next handoff. ### Human approval Connecting projects does not grant new permissions. Each owner still controls the sources, the work, and any action that needs approval. ### Useful human guides - Help a team work together without losing ownership: https://work-guide.openai.chatgpt.site/guide/team-operating-systems - Move work between the tools you already use: https://work-guide.openai.chatgpt.site/guide/cross-application-work - Chief of Staff: https://work-guide.openai.chatgpt.site/guide/cos-loop - Keep longer work safe and recoverable: https://work-guide.openai.chatgpt.site/guide/work-never-stops - Build a Site that becomes part of the work: https://work-guide.openai.chatgpt.site/guide/sites-as-working-artifacts ### Common mistakes - Giving one assistant every source and every decision - Losing track of who owns different work and information - Treating a draft or local preview as a finished, published result ## Approval and safety - Stay read-only unless the person explicitly approves a change. - Keep facts, assumptions, recommendations, and unanswered questions separate. - Never treat a draft, local preview, or hosted candidate as published work. - Verify owners, sources, and permissions before relying on a result. - Ask before sending, publishing, deploying, deleting, or changing access.