guides/prompt-guidance.md +0 −264 deleted
File Deleted View Diff
1# GPT-5.5 prompting guide
2
3Prompt GPT-5.5 with outcome-first goals, concise style controls, retrieval budgets, and validation loops.
4
5## New in GPT-5.5 vs GPT-5.4
6- Shorter, outcome-first prompts usually work better than process-heavy prompt stacks.
7- More efficient reasoning means `low` and `medium` effort should be re-evaluated before escalating.
8- Preambles, `phase` handling, and assistant-item replay remain important for tool-heavy Responses workflows.
9- Explicit personality, retrieval budgets, and validation rules help shape customer-facing and agentic UX.
10
11GPT-5.5 works best when prompts define the outcome and leave room for the model to choose an efficient solution path. Compared with earlier models, you can often use shorter, more outcome-oriented prompts: describe what good looks like, what constraints matter, what evidence is available, and what the final answer should contain.
12
13Avoid carrying over every instruction from an older prompt stack. Legacy prompts often over-specify the process because earlier models needed more help staying on track. With GPT-5.5, that can add noise, narrow the model's search space, or lead to overly mechanical answers.
14
15For more detail on GPT-5.5 behavior changes, start with the [Using GPT-5.5 guide](https://developers.openai.com/api/docs/guides/latest-model). This guide focuses on prompt changes that follow from those behavior changes.
16
17The patterns here are starting points. Adapt them to your product surface, tools, evals, and user experience goals.
18
19## Automated migration with Codex
20
21Codex can implement the changes from this guide with the [OpenAI Docs Skill](https://github.com/openai/skills/tree/main/skills/.curated/openai-docs).
22
23```text
24$openai-docs migrate this project to gpt-5.5
25```
26
27To use this skill in other coding agents, download it from the [OpenAI skills repository](https://github.com/openai/skills/tree/main/skills/.curated/openai-docs).
28
29## Personality and behavior
30
31GPT-5.5's default style is efficient, direct, and task-oriented. This is useful for production systems: responses stay focused, behavior is easier to steer, and the model avoids unnecessary conversational padding.
32
33For customer-facing assistants, support workflows, coaching experiences, and other conversational products, define both personality and collaboration style.
34
35- **Personality** controls how the assistant sounds: tone, warmth, directness, formality, humor, empathy, and level of polish.
36- **Collaboration style** controls how the assistant works: when it asks questions, when it makes assumptions, how proactive it should be, how much context it gives, when it checks work, and how it handles uncertainty or risk.
37
38Keep both short. Personality instructions should shape the user experience. Collaboration instructions should shape task behavior. Neither should replace clear goals, success criteria, tool rules, or stopping conditions.
39
40Example personality block for a steady task-focused assistant:
41
42```text
43# Personality
44You are a capable collaborator: approachable, steady, and direct. Assume the user is competent and acting in good faith, and respond with patience, respect, and practical helpfulness.
45
46Prefer making progress over stopping for clarification when the request is already clear enough to attempt. Use context and reasonable assumptions to move forward. Ask for clarification only when the missing information would materially change the answer or create meaningful risk, and keep any question narrow.
47
48Stay concise without becoming curt. Give enough context for the user to understand and trust the answer, then stop. Use examples, comparisons, or simple analogies when they make the point easier to grasp. When correcting the user or disagreeing, be candid but constructive. When an error is pointed out, acknowledge it plainly and focus on fixing it.
49
50Match the user's tone within professional bounds. Avoid emojis and profanity by default, unless the user explicitly asks for that style or has clearly established it as appropriate for the conversation.
51```
52
53Example personality block for an expressive collaborative assistant:
54
55```text
56# Personality
57Adopt a vivid conversational presence: intelligent, curious, playful when appropriate, and attentive to the user's thinking. Ask good questions when the problem is blurry, then become decisive once there is enough context.
58
59Be warm, collaborative, and polished. Conversation should feel easy and alive, but not chatty for its own sake. Offer a real point of view rather than merely mirroring the user, while staying responsive to their goals and constraints.
60
61Be thoughtful and grounded when the task calls for synthesis or advice. State a clear recommendation when you have enough context, explain important tradeoffs, and name uncertainty without becoming evasive.
62```
63
64For more expressive products, add warmth, curiosity, humor, or point of view explicitly, but keep the block short. Use personality to shape the experience, not to compensate for unclear goals or missing task instructions.
65
66## Improve time to first visible token with a preamble
67
68In streaming applications, users notice how long it takes before the first visible response appears. GPT-5.5 may spend time reasoning, planning, or preparing tool calls before emitting visible text.
69
70For longer or tool-heavy tasks, prompt the model to start with a short preamble: a brief visible update that acknowledges the request and states the first step. This can improve perceived responsiveness without changing the underlying task.
71
72Use this pattern when the task may take more than one step, require tool calls, or involve a long-running agent workflow.
73
74```text
75Before any tool calls for a multi-step task, send a short user-visible update that acknowledges the request and states the first step. Keep it to one or two sentences.
76```
77
78For coding agents that expose separate message phases, you can be more explicit:
79
80```text
81You must always start with an intermediary update before any content in the analysis channel if the task will require calling tools. The user update should acknowledge the request and explain your first step.
82```
83
84## Outcome-first prompts and stopping conditions
85
86GPT-5.5 is strongest when the prompt defines the target outcome, success criteria, constraints, and available context, then lets the model choose the path.
87
88For many tasks, describe the destination rather than every step. This gives the model room to choose the right search, tool, or reasoning strategy for the task.
89
90Prefer this:
91
92```text
93Resolve the customer's issue end to end.
94
95Success means:
96- the eligibility decision is made from the available policy and account data
97- any allowed action is completed before responding
98- the final answer includes completed_actions, customer_message, and blockers
99- if evidence is missing, ask for the smallest missing field
100```
101
102**Avoid unnecessary absolute rules.** Older prompts often use strict instructions like `ALWAYS`, `NEVER`, `must`, and `only` to control model behavior. Use those words for true invariants, such as safety rules, required output fields, or actions that should never happen. For judgment calls, such as when to search, ask for clarification, use a tool, or keep iterating, prefer decision rules instead.
103
104Avoid this style of instruction unless every step is truly required:
105
106```text
107First inspect A, then inspect B, then compare every field, then think through
108all possible exceptions, then decide which tool to call, then call the tool,
109then explain the entire process to the user.
110```
111
112Add explicit stopping conditions:
113
114```text
115Resolve the user query in the fewest useful tool loops, but do not let loop minimization outrank correctness, accessible fallback evidence, calculations, or required citation tags for factual claims.
116
117After each result, ask: "Can I answer the user's core request now with useful evidence and citations for the factual claims?" If yes, answer.
118```
119
120Define missing-evidence behavior:
121
122```text
123Use the minimum evidence sufficient to answer correctly, cite it precisely, then stop.
124```
125
126## Formatting
127
128GPT-5.5 is highly steerable on output format and structure. Use that control when it improves comprehension or product fit.
129
130Set `text.verbosity`, describe the expected output shape, and reserve heavier structure for cases where it improves comprehension or your product UI needs a stable artifact. The API default for `text.verbosity` is `medium`; use `low` when you prefer shorter, more concise responses.
131
132Plain conversational formatting:
133
134```text
135Let formatting serve comprehension. Use plain paragraphs as the default format for normal conversation, explanations, reports, documentation, and technical writeups. Keep the presentation clean and readable without making the structure feel heavier than the content.
136
137Use headers, bold text, bullets, and numbered lists sparingly. Reach for them when the user requests them, when the answer needs clear comparison or ranking, or when the information would be harder to scan as prose. Otherwise, favor short paragraphs and natural transitions.
138
139Respect formatting preferences from the user. If they ask for a terse answer, minimal formatting, no bullets, no headers, or a specific structure, follow that preference unless there is a strong reason not to.
140```
141
142Add explicit audience and length guidance:
143
144```text
145Write for a senior business audience. Keep the answer under 400 words. Use short paragraphs and only include bullets when they improve scannability. Prioritize the conclusion first, then the reasoning, then caveats.
146```
147
148For editing, rewriting, summaries, or customer-facing messages, tell the model what to preserve before asking it to improve style. This pattern is useful when you want polish without expansion.
149
150```text
151Preserve the requested artifact, length, structure, and genre first. Quietly improve clarity, flow, and correctness. Do not add new claims, extra sections, or a more promotional tone unless explicitly requested.
152```
153
154## Grounding, citations, and retrieval budgets
155
156For grounded answers, citation behavior should be part of the prompt. Define what needs support, what counts as enough evidence, and how the model should behave when evidence is missing. Absence of evidence shouldn't automatically become a factual "no." For more details and examples, see the [citation formatting guide](https://developers.openai.com/api/docs/guides/citation-formatting).
157
158### Add an explicit retrieval budget
159
160Retrieval budgets are stopping rules for search. They tell the model when enough evidence is enough.
161
162```text
163For ordinary Q&A, start with one broad search using short, discriminative keywords. If the top results contain enough citable support for the core request, answer from those results instead of searching again.
164
165Make another retrieval call only when:
166- The top results do not answer the core question.
167- A required fact, parameter, owner, date, ID, or source is missing.
168- The user asked for exhaustive coverage, a comparison, or a comprehensive list.
169- A specific document, URL, email, meeting, record, or code artifact must be read.
170- The answer would otherwise contain an important unsupported factual claim.
171
172Do not search again to improve phrasing, add examples, cite nonessential details, or support wording that can safely be made more generic.
173```
174
175## Creative drafting guardrails
176
177For drafting tasks, tell the model which claims must come from sources and which parts may be creatively written. This is especially important for slides, launch copy, customer summaries, talk tracks, leadership blurbs, and narrative framing.
178
179```text
180For creative or generative requests such as slides, leadership blurbs, outbound copy, summaries for sharing, talk tracks, or narrative framing, distinguish source-backed facts from creative wording.
181
182- Use retrieved or provided facts for concrete product, customer, metric, roadmap, date, capability, and competitive claims, and cite those claims.
183- Do not invent specific names, first-party data claims, metrics, roadmap status, customer outcomes, or product capabilities to make the draft sound stronger.
184- If there is little or no citable support, write a useful generic draft with placeholders or clearly labeled assumptions rather than unsupported specifics.
185```
186
187## Frontend engineering and visual taste
188
189For frontend work, refer to the [example instructions](https://developers.openai.com/api/docs/guides/frontend-prompt) for practical ways to steer UI quality. They cover product and user context, design-system alignment, first-screen usability, familiar controls, expected states, responsive behavior, and common generated-UI defaults to avoid, such as generic heroes, nested cards, decorative gradients, visible instructional text, and broken layouts.
190
191## Prompt the model to check its work
192
193Give GPT-5.5 access to tools that let it check outputs when validation is possible.
194
195For coding agents, ask for concrete validation commands:
196
197```text
198After making changes, run the most relevant validation available:
199- targeted unit tests for changed behavior
200- type checks or lint checks when applicable
201- build checks for affected packages
202- a minimal smoke test when full validation is too expensive
203
204If validation cannot be run, explain why and describe the next best check.
205```
206
207For visual artifacts, ask for inspection after rendering:
208
209```text
210Render the artifact before finalizing. Inspect the rendered output for layout, clipping, spacing, missing content, and visual consistency. Revise until the rendered output matches the requirements.
211```
212
213For engineering and planning tasks, make implementation plans traceable:
214
215```text
216For implementation plans, include:
217- requirements and where each is addressed
218- named resources, files, APIs, or systems involved
219- state transitions or data flow where relevant
220- validation commands or checks
221- failure behavior
222- privacy and security considerations
223- open questions that materially affect implementation
224```
225
226## Phase parameter
227
228Starting with GPT-5.4, long-running or tool-heavy Responses workflows can use assistant-item `phase` values to distinguish intermediate updates from final answers. GPT-5.5 uses the same pattern.
229
230If you use `previous_response_id`, the API preserves prior assistant state automatically. If your application manually replays assistant output items into the next request, preserve each original `phase` value and pass it back unchanged. This matters most when a response includes preambles, repeated tool calls, or a final answer after intermediate assistant updates.
231
232```text
233If manually replaying assistant items:
234- Preserve assistant `phase` values exactly.
235- Use `phase: "commentary"` for intermediate user-visible updates.
236- Use `phase: "final_answer"` for the completed answer.
237- Do not add `phase` to user messages.
238```
239
240## Suggested prompt structure
241
242Use this structure as a starting point for complex prompts. Keep each section short. Add detail only where it changes behavior.
243
244```text
245Role: [1-2 sentences defining the model's function, context, and job]
246
247# Personality
248[tone, demeanor, and collaboration style]
249
250# Goal
251[user-visible outcome]
252
253# Success criteria
254[what must be true before the final answer]
255
256# Constraints
257[policy, safety, business, evidence, and side-effect limits]
258
259# Output
260[sections, length, and tone]
261
262# Stop rules
263[when to retry, fallback, abstain, ask, or stop]
264```