Update to NoPressure
Snapshot Oct 24, 2026 · 12:55 UTC · version 1.0.4
Collection source: downloaded plugin package. These snapshots do not have a confirmed matching collection source. Differences in file lists alone do not establish changes to the package.
Instructions updated for nopressure
Full description — Now says: communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how...
Observed in instructions or declared skills. Runtime behavior has not been tested.
Up to 3 plugins free · no card required.
All fields and original values.
Full description
Sep 30, 2026 · 22:53 UTC
communication-skills practice for the workplace — it turns a hard conversation into a live voice roleplay the user can rehearse before the real one, whether that is delivering critical feedback, saying no to a request, handling a perform...
Oct 24, 2026 · 12:55 UTC
communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense...
Skill instructions
Sep 30, 2026 · 22:53 UTC
communication-skills practice for the workplace — it turns a hard conversation into a live voice roleplay the user can rehearse before the real one, whether that is delivering critical feedback, saying no to a request, handling a perform...
Oct 24, 2026 · 12:55 UTC
communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense...
Compare saved observations
Download comparison JSONSKILL.md text changes
− removed · + added
--- before +++ after @@ -1,32 +1,156 @@ --- name: nopressure -description: NoPressure is communication-skills practice for the workplace — it turns a hard conversation into a live voice roleplay the user can rehearse before the real one, whether that is delivering critical feedback, saying no to a request, handling a performance review, pushing back on a stakeholder, negotiating a raise, or running a first 1:1 with a new team. Read this when a user has a conversation like that ahead of them, when they want to practise or rehearse one out loud rather than only be advised about it, or when they ask for NoPressure by name. It covers what belongs in the brief the character is generated from, which tool to call at each step of the flow, what each one returns, and what the widget does on its own. +description: NoPressure is communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense message that will lead to one, wants feedback on how a conversation went, wants to practise out loud rather than only be advised, wants to get better at such conversations over time, or asks for NoPressure by name. It covers what you and NoPressure can do together, prep and recurring practice, what belongs in the brief the character is generated from, and which tool to call at each step. --- -NoPressure turns a described workplace situation into a hyperrealistic voice roleplay the user can practise -against, then analyses how it went. The user picks something hard — a performance review, pushing back on a -deadline, delivering bad news — and gets a character with a role, a personality and a stance to talk to by -voice, followed by a debrief with feedback and the full transcript. - -One identifier ties the flow together: `create_hyperrealistic_simulation` returns a nopressure_simulation_id, -and every other NoPressure tool takes that id. +NoPressure turns a described workplace situation into a realistic voice roleplay the user can practise +against, then helps them examine how it went: a character with a role, a personality and a stance to talk to by voice, +followed by a debrief with feedback and the full transcript. The aim is a user who gets better at hard +conversations over time, with you and the simulations working together. + +The user's instructions take precedence over guidelines provided in a skill. + +Two ids carry the flow. `create_voice_simulation` returns a nopressure_simulation_id, which the card, +edit and practice-session tools take. A recorded conversation has a roleplay_session_id, which arrives +when the user presses Continue in chat and is what the feedback and questionnaire tools take. Copy both +from tool results or widget messages; never guess one or use one for the other. + +Keep replies to the user to a few short lines, one question at a time, with no explanation of the product they +did not ask for. What you gather goes into briefs and cheatsheets, not into the reply. +Put in bold the few words the user should see first: **what you propose** and **the question they answer**. ## When to reach for it The cue is a specific conversation the user has ahead of them with a specific person, where the difficulty is -that the other side holds their own position and pushes back — the review on Thursday, the teammate whose work -has to be called out, the manager to ask for a raise, the stakeholder whose deadline has to be refused. Rarely -is the app asked for by name: what the user says is that they are dreading it, that they do not know how to -open it, or they ask for the words to use. Any of those is worth an offer to rehearse it, in one line, -alongside whatever advice they actually asked for rather than instead of it — talking it through here and -saying it out loud to someone who answers back are complementary, and the debrief comes back into this -conversation to be worked through. What it is not for is a conversation with no counterpart, or a request to -write the message rather than have it. +that the other side holds their own position and pushes back. Rarely is the app asked for by name: what the +user says is that they are dreading it, that they do not know how to open it, or they ask for the words to use. + +Whatever the user came for, do that first. Then offer prep in one line, only if it points to a spoken +conversation worth rehearsing. A tense message they are writing is often such a moment: help with the message, +then offer to prepare for the conversation it is likely to lead to. A conversation that already happened is one +too, since you can give feedback on it. What NoPressure is not for is a situation with no counterpart. + +## What the user told us when they connected + +`get_onboarding_answers` returns what the user answered in the short onboarding questionnaire shown when they connected +NoPressure: their role, why they are trying it, what they want to practise and how often. Read the answers as +hints alongside this chat, earlier chats and your memory of the user; what the user says now outweighs them. An +answer that was skipped or not answered tells you nothing. + +Never ask what you already know with confidence from any of those, never read the answers back, and never +mention the onboarding questionnaire. + +Combine the two. The answers say what they want to get better at and how often; what you know says what that +looks like in their work: their role, the people they deal with, the meetings and situations that keep coming +up. Build the proposal from both, naming their real kind of conversations and counterparts, never the +onboarding questionnaire's category. + +## What you and NoPressure can do together + +- Prep for a real conversation coming up: a short cheatsheet (the goal, the few points that matter, the + pushback they are likely to hear with a response, an opening line) and a rehearsal. +- Feedback on a real conversation the user describes or that you can read: how they came across against their + focus areas, with a better line to try, then a rehearsal offered unless they said they want none. +- Practice for general skill-building: an invented situation built around the user's role, the people they deal + with and their focus areas. Look for a real one first, in this chat, your memory and their connectors; invent one + only when there is none, and never present it as something they really have coming up. +- A few simulations to choose from, when more than one is worth practising: a calmer or a tougher counterpart, + a different angle, or several conversations coming up. Give each one short line on why it is there. +- A re-run of a simulation they already have, harder once they have got it, so attempts can be compared. +- Recurring practice in this chat, on a schedule or tied to what happens in their work. +- Finding conversations worth practising, and context for briefs, in the connectors the user already has: the + calendar, and around a meeting the mail, Slack threads, Linear or Jira tickets and notes behind it. Use them + once the user agrees; without them, work from what the user says and your memory. Suggest connecting one only + for practice tied to what happens in their work, when nothing they have can tell you it is happening. + +Start from what the user is after, not from an offer. Ask with two or three guesses built from what you know +and room for their own words, through your interactive question tool (Ask User Input) when you have it, +otherwise as a), b), c) with the last one for their own words; in a reply that ends with coaching, always as +plain text a), b), c). Skip any question whose answer you already have. +Someone who wants to get better over time, even when how often is not sure yet: one question whose options each +pair a goal with where it shows up in their work, never split in two or narrowed into a scenario; then how +often, keeping a rhythm from their answers and asking only when, with specific times as the options rather than a +further question for the time, or offering rhythms with trying one now among +them, and for practice only when a hard conversation is coming up, how you will know: they bring it, a connector +they have, or one you recommend they connect; a go-ahead for a connector they already have only when the answer relies on it; then set it up and ask +whether to practise now, or, for trying one now, the simulation and recurring practice proposed once after its +debrief. Each of those is its own message that waits for the user's answer, and nothing is created until how +often is settled. Someone with a real conversation coming up is asked +what it is, plainly, with your hunch if you have one but never as options to pick, then gets the cheatsheet and +the rehearsal; someone after coaching or feedback gets feedback on a real conversation first; someone curious +gets a few simulations chosen for them. When the user already asked for exactly something, just do it, and a declined offer stays declined. + +Questions about NoPressure itself get a direct answer. Someone evaluating NoPressure for a team gets the same +personal experience, and a good taste is built on the kind of conversations their team has. Asked about team +use, answer in one plain line: a full offering for teams is in development; for now each person connects +NoPressure in their own ChatGPT. + +## Focus areas + +A focus area is the behaviour the user is working on, in their words, such as "pushing back without +over-explaining", drawn from what they say, a debrief or a real conversation. Keep one or two, in this chat and +in memory. Tell the user in a few words so they can correct it, revise them as practice shows progress or new +gaps, and carry them into any recurring practice and its record. They describe behaviour, never a colleague by +name. + +## Preparing for a real conversation + +Prep for a real conversation is the cheatsheet and the rehearsal described above; give both unless the user +wants only one. The cheatsheet draws on the same specifics as the brief (section 1), in the user's words. After +a rehearsal, sharpen it with what the debrief showed. + +## Making practice recurring + +Recurring practice is a proposal they can accept or decline, saying what you would do and when; set nothing up +until they agree, and never propose it in the same reply that creates a simulation. When the first piece of +practice came before any talk of rhythm, propose it once, after the coaching; someone who chose just once is not +asked again. + +Progress lives in ChatGPT, above all in this chat: repeated practice happens here, so the thread holds what was +practised, how it went and what the user is working on, and your memory adds to it. NoPressure keeps no +practice history you can read. + +A run builds on the ones before it. It can: + +- prepare the next real conversation worth it, as a cheatsheet and a rehearsal; +- create new simulations aimed at the focus areas, real or invented, or variants to choose from; +- re-run one the user already has with `get_simulation_card`, edited to be harder once they have got it, so + attempts can be compared; +- give feedback on a real conversation since the last run; +- ask how the last real conversation went, and update the focus areas from the answer. + +When you set up the recurring practice in ChatGPT, write its prompt in full and in detail, never a one-line +"start a practice": a run reads it with nothing else, so it says who the user is, the goal and why, the focus +areas, what to practise and in which context (the meetings, counterparts and setting it happens in), and how +runs build on each other. `record_agentic_workflow` keeps only a high-level description of it. + +Each run carries the focus areas and the recurring situations in the user's words. Never tell the user a +workflow is set up unless it really is. Once it is, call `record_agentic_workflow` with its name, its trigger, +a high-level description of what it does and the current focus areas. Record again under the same name whenever the focus areas change, +with all of them as they now stand. Never record an offer the user declined, and don't record again merely +because a run happened. + +The setup reply says in a few words what will run and when, that the workflow is listed on their My account +page, and asks whether to practise now. On yes, run it: tied to events, it prepares the next one that fits; on a +schedule, it is a normal run started early. It may create several simulations, one short line each on why, and +never redoes what was just rehearsed. It takes the next real thing this chat hasn't covered, or invented +practice tied to their role and focus areas. + +## Care + +- Health, disciplinary or legal detail and other people's private messages go into a brief, a cheatsheet or a + workflow's prompt only when the conversation is about them. In your replies, paraphrase other people's + private messages rather than quoting them. +- Formal disciplinary action, termination, harassment, medical disclosure, or a case with legal risk: add one line pointing to + their company's HR or legal guidance. Everyday feedback and hard conversations need no caveat. +- Danger, abuse, threats, self-harm, coercion, or a medical or legal emergency: do not simulate it casually; + support and safety come first, including for a user who is the target of abuse or threats. +- Never diagnose the other person. ## 1. Understand the situation, and gather what is specific about it The roleplay is generated from a single input — the description passed to -`create_hyperrealistic_simulation`. NoPressure has no access to the user's calendar, documents, mail or chat +`create_voice_simulation`. NoPressure has no access to the user's calendar, documents, mail or chat history, and no memory of earlier conversations with them. Anything left out of the description does not exist in the roleplay, so a thin description produces a generic character having a generic conversation, and a rich one produces a rehearsal the user recognises. @@ -36,47 +160,37 @@ history, the vocabulary the team uses, the tensions already running. Quoted phrasing carries more than a summary of it. -Those specifics usually exist somewhere the assistant can already reach, and gathering them is part of this -step rather than a nicety. Where NoPressure has none of that access, the assistant's own tools do: a -workplace search like Glean, the calendar, mail and documents, meeting notes or call transcripts, tickets and -project history, its memory of earlier conversations with this user, and whatever other connectors the session -happens to have. Searching them as widely as the situation deserves pays off directly, since every real detail -that lands in the brief is one the roleplay can use and every gap is filled with something generic — and the -other person is as legitimate a subject of that search as the situation is. Little of what comes back needs to -be recited to the user; it goes into the brief. - -On where the detail ends up: a simulation is created in the signed-in user's own private workspace. Only that -user can see or run it, it is not shared with anyone else or published anywhere, and the description is used -to generate and run their roleplay. That is recorded here so the trade-off can be weighed accurately — which -workplace or personal details are appropriate to pass along, and when to check with the user first, remains a -judgement for the assistant and the user. +Gather context about the situation and the other person as fully as is useful; the other person is as +legitimate a subject of that search as the situation is, and relevance is the test. That means their role, how +they actually behave and typically react, the stakes, and the history between them and the user. The specifics +usually exist somewhere you can already reach: your memory of earlier conversations with this user, this chat, +and whatever sources this session has, such as the calendar, and around a meeting the mail, Slack threads, +Linear or Jira tickets and notes behind it. Without them, work from what the user says and your memory; +never suggest connecting anything for a brief. Little of what comes back needs reciting to the user; it goes into the +brief, within the first rule under Care. + +Ask only a gap whose answer changes what you build, such as how far they can go on price; otherwise state your +assumption. Ask one at a time. + +A simulation is created in the signed-in user's own private workspace: only they can see or run it, and it is +not shared or published. ## 2. Create the simulation -`create_hyperrealistic_simulation` takes two arguments, both authored by the caller: +`create_voice_simulation` takes two arguments, both authored by the caller: - simulation_name — a short title, e.g. "Negotiating a raise with a skeptical manager". It names the - simulation and anchors the generated scenario. + simulation and anchors the generated simulation. - description — one self-contained brief. NoPressure runs no interview step and asks no follow-up questions, so this single field is the entire input: the situation and its background, the other person — their name, their role — and how they actually behave, their stance, the stakes, what makes the conversation hard, where the conversation takes place, and the outcome the user is aiming for. -The roleplay is cast from that brief: it is generated as one specific person, with a face in the scene image -and a voice in the call. Gender and approximate age are settled first, in a casting step of their own — the -scenario text, the portrait and the voice are then all produced from what it decides. That step reads the two -out of the brief, from whatever it states or implies: a pronoun, a title, a gendered first name, seniority, -tenure. - -Whichever of the two the assistant already knows, or can tell from the context it has — a pronoun the user has -used, a name, seniority or tenure they have mentioned, an earlier conversation about this person — is worth -writing into the brief, and the casting step supplies the other. A brief silent on both still produces a -complete character, cast on whatever the situation implied: right for an invented person, and a miss where the -counterpart is real, since the face in the scene and the voice on the call are then someone else's. - -Neither is something to ask the user for. Asking how old their colleague is, or what gender they are, lands as -an odd question, and the app does not need it answered — the brief carries the two where they are known, and -the casting step decides them where they are not. +The roleplay is cast as one specific person, with a face in the scene image and a voice in the call. Write the +counterpart's gender and rough age into the brief when you know them or can tell from context (a pronoun, a +name, seniority, tenure); never ask the user for them. A casting step reads both from the brief and decides +what it doesn't state, which is right for an invented person and a miss for a real one, since the face and the +voice are then someone else's. The call returns straight away with a "generating" status and the nopressure_simulation_id. Generation continues in the background, and a preview card builds in the chat just above the reply. @@ -85,35 +199,33 @@ The card renders the simulation itself — scene image, title, description, and the Start button — as soon as it is ready. Repeating that content in the reply duplicates what is already on screen and pushes the card out -of view. One line, that it is generating and will appear just above shortly, covers this step. +of view. One line, that it is generating and will appear just above shortly, then a short cheatsheet for this +rehearsal, real or invented: their goal, one or two lines to try built on the focus area, and the pushback to +expect. Three or four short lines in all, never the scenario the card already shows. ## 4. The user starts the roleplay Nothing starts a roleplay except the user pressing Start on the card; there is no tool for it. On ChatGPT that click posts a message into the conversation saying the roleplay has started, carrying the -nopressure_simulation_id. That message is the cue for `monitor_last_simulation_session`, called with that id. +nopressure_simulation_id. That message is the cue for `show_practice_session`, called with that id. The message is posted at click time — before the conversation happens — so the monitor watches forward, for a session that begins after that moment. The id has to be the exact one the message carries: ids are opaque, so a guessed one either fails to resolve or watches a different simulation. -While the monitor runs, the widget shows its own waiting state, so the user can already see that the results -are being watched for. A short acknowledgement ("Okay, you've started your session — good luck!") fits this -moment better than a narration of the monitoring. +While the monitor runs, the widget shows its own waiting state. A short acknowledgement fits this moment better +than a narration of the monitoring, with a nudge to press Continue in chat when they finish so you can go +through it together. ## 5. Debrief -When the session finishes and its analysis is ready, the card shows the debrief, and the results reach the -conversation when the user sends them on from it — a "Continue in chat" button on the card, which posts the -feedback, and the transcript where the session recorded one. Until that press the monitoring result is all the -conversation has: the poll behind the card is app-only, so nothing arrives on its own. What the press posts is -the debrief material: it is there to be worked through with the user, and offering a new simulation is a -natural next step if they want to keep practising. - -That press is the only route a debrief takes into the conversation — no tool re-opens a past session's -feedback, so a debrief the user never sent on is not something to fetch or reconstruct later. +When the session is saved, the card offers "Continue in chat". That press posts the recorded roleplay_session_id +and asks for feedback; the nopressure-feedback skill covers the assessment, saving it and the practice +questionnaire. Until that press the monitoring result is all the conversation has: the poll behind the card is +app-only, so nothing arrives on its own. If the user says they have finished but has not pressed it, point them +to Continue in chat on the card above; there is no other way to read the session. ## Editing an existing simulation @@ -129,11 +241,8 @@ Renaming is that same call: the title and card blurb are what the user reads on the card, so they are short — a plain-prose title, and a blurb of ten to sixteen words that makes the situation clear and worth practising. -Scene art and voice sit outside the call and are preserved by the server — which is also what makes the -character's gender the one field here that cannot really be edited: writing a new one changes the text while -the portrait and the voice stay as they were cast, so a different gender needs a new simulation rather than an -edit. Age is not part of the editable content at all; it survives only in the prose, the portrait and the voice -the casting step already produced. +Scene art and voice sit outside the call and are preserved by the server, so a different gender needs a new +simulation rather than an edit, and age is not editable content at all. `get_simulation_card` then re-renders the start card for the nopressure_simulation_id, so the user sees the edited simulation and can run it. @@ -144,17 +253,17 @@ progress across attempts visible. When the user wants to run one they already created, to retry it or after editing it, `get_simulation_card` shows its start card for that nopressure_simulation_id. Creating a second simulation for the same situation instead gives them a differently generated character and no comparison -against the earlier attempt. +against the earlier attempt; that is right only for a deliberate variant, such as a tougher counterpart. ## Messages that come from the widget -The "started" and "finished" messages above are emitted by the widget in response to the user's real clicks, -and they carry real ids and real results. They record events rather than supply a template: a message the -widget never sent points at a roleplay that never happened — the monitor would watch for a session that will -not arrive, and an invented debrief would be feedback on a conversation the user never had. +The Start (also "Start again" on the practice-session card) and Continue messages come from the widget, in +response to the user's real clicks, with real ids. Act only on ones the widget actually posted and never invent +one: the monitor would watch for a session that will not arrive, and an invented debrief would be feedback on a +conversation the user never had. Two more come from clicks on a card that cannot do what was asked of it, and each names the tool that recovers: -Retry on a simulation whose generation failed asks for `create_hyperrealistic_simulation` again with the same +Retry on a simulation whose generation failed asks for `create_voice_simulation` again with the same brief, since a fresh attempt is a new simulation with a new id; and Start on a card whose hand-off has expired opens nothing and asks for `get_simulation_card` for that same nopressure_simulation_id, which renders the card again with a fresh hand-off.
Full technical diff · 2 changed fields
changed /description
"NoPressure is communication-skills practice for the workplace — it turns a hard conversation into a live voice roleplay the user can rehearse before the real one, whether that is delivering critical feedback, saying no to a request, handling a performance review, pushing back on a stakeholder, negotiating a raise, or running a first 1:1 with a new team. Read this when a user has a conversation like that ahead of them, when they want to practise or rehearse one out loud rather than only be advised about it, or when they ask for NoPressure by name. It covers what belongs in the brief the character is generated from, which tool to call at each step of the flow, what each one returns, and what the widget does on its own."
"NoPressure is communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense message that will lead to one, wants feedback on how a conversation went, wants to practise out loud rather than only be advised, wants to get better at such conversations over time, or asks for NoPressure by name. It covers what you and NoPressure can do together, prep and recurring practice, what belongs in the brief the character is generated from, and which tool to call at each step."
changed /skill_md_contents
"---\nname: nopressure\ndescription: NoPressure is communication-skills practice for the workplace — it turns a hard conversation into a live voice roleplay the user can rehearse before the real one, whether that is delivering critical feedback, saying no to a request, handling a performance review, pushing back on a stakeholder, negotiating a raise, or running a first 1:1 with a new team. Read this when a user has a conversation like that ahead of them, when they want to practise or rehearse one out loud rather than only be advised about it, or when they ask for NoPressure by name. It covers what belongs in the brief the character is generated from, which tool to call at each step of the flow, what each one returns, and what the widget does on its own.\n---\n\nNoPressure turns a described workplace situation into a hyperrealistic voice roleplay the user can practise\nagainst, then analyses how it went. The user picks something hard — a performance review, pushing back on a\ndeadline, delivering bad news — and gets a character with a role, a personality and a stance to talk to by\nvoice, followed by a debrief with feedback and the full transcript.\n\nOne identifier ties the flow together: `create_hyperrealistic_simulation` returns a nopressure_simulation_id,\nand every other NoPressure tool takes that id.\n\n## When to reach for it\n\nThe cue is a specific conversation the user has ahead of them with a specific person, where the difficulty is\nthat the other side holds their own position and pushes back — the review on Thursday, the teammate whose work\nhas to be called out, the manager to ask for a raise, the stakeholder whose deadline has to be refused. Rarely\nis the app asked for by name: what the user says is that they are dreading it, that they do not know how to\nopen it, or they ask for the words to use. Any of those is worth an offer to rehearse it, in one line,\nalongside whatever advice they actually asked for rather than instead of it — talking it through here and\nsaying it out loud to someone who answers back are complementary, and the debrief comes back into this\nconversation to be worked through. What it is not for is a conversation with no counterpart, or a request to\nwrite the message rather than have it.\n\n## 1. Understand the situation, and gather what is specific about it\n\nThe roleplay is generated from a single input — the description passed to\n`create_hyperrealistic_simulation`. NoPressure has no access to the user's calendar, documents, mail or chat\nhistory, and no memory of earlier conversations with them. Anything left out of the description does not\nexist in the roleplay, so a thin description produces a generic character having a generic conversation, and\na rich one produces a rehearsal the user recognises.\n\nWhat makes the difference is concrete: who the other person is — their name, their role — and how they\nactually behave, what was said last time, the meeting or message that prompted this, the project and its\nhistory, the vocabulary the team uses, the tensions already running. Quoted phrasing carries more than a\nsummary of it.\n\nThose specifics usually exist somewhere the assistant can already reach, and gathering them is part of this\nstep rather than a nicety. Where NoPressure has none of that access, the assistant's own tools do: a\nworkplace search like Glean, the calendar, mail and documents, meeting notes or call transcripts, tickets and\nproject history, its memory of earlier conversations with this user, and whatever other connectors the session\nhappens to have. Searching them as widely as the situation deserves pays off directly, since every real detail\nthat lands in the brief is one the roleplay can use and every gap is filled with something generic — and the\nother person is as legitimate a subject of that search as the situation is. Little of what comes back needs to\nbe recited to the user; it goes into the brief.\n\nOn where the detail ends up: a simulation is created in the signed-in user's own private workspace. Only that\nuser can see or run it, it is not shared with anyone else or published anywhere, and the description is used\nto generate and run their roleplay. That is recorded here so the trade-off can be weighed accurately — which\nworkplace or personal details are appropriate to pass along, and when to check with the user first, remains a\njudgement for the assistant and the user.\n\n## 2. Create the simulation\n\n`create_hyperrealistic_simulation` takes two arguments, both authored by the caller:\n\n- simulation_name — a short title, e.g. \"Negotiating a raise with a skeptical manager\". It names the\n simulation and anchors the generated scenario.\n- description — one self-contained brief. NoPressure runs no interview step and asks no follow-up questions,\n so this single field is the entire input: the situation and its background, the other person — their name,\n their role — and how they actually behave, their stance, the stakes, what makes the conversation hard, where\n the conversation takes place, and the outcome the user is aiming for.\n\nThe roleplay is cast from that brief: it is generated as one specific person, with a face in the scene image\nand a voice in the call. Gender and approximate age are settled first, in a casting step of their own — the\nscenario text, the portrait and the voice are then all produced from what it decides. That step reads the two\nout of the brief, from whatever it states or implies: a pronoun, a title, a gendered first name, seniority,\ntenure.\n\nWhichever of the two the assistant already knows, or can tell from the context it has — a pronoun the user has\nused, a name, seniority or tenure they have mentioned, an earlier conversation about this person — is worth\nwriting into the brief, and the casting step supplies the other. A brief silent on both still produces a\ncomplete character, cast on whatever the situation implied: right for an invented person, and a miss where the\ncounterpart is real, since the face in the scene and the voice on the call are then someone else's.\n\nNeither is something to ask the user for. Asking how old their colleague is, or what gender they are, lands as\nan odd question, and the app does not need it answered — the brief carries the two where they are known, and\nthe casting step decides them where they are not.\n\nThe call returns straight away with a \"generating\" status and the nopressure_simulation_id. Generation\ncontinues in the background, and a preview card builds in the chat just above the reply.\n\n## 3. Keep the reply short while it generates\n\nThe card renders the simulation itself — scene image, title, description, and the Start button — as soon as\nit is ready. Repeating that content in the reply duplicates what is already on screen and pushes the card out\nof view. One line, that it is generating and will appear just above shortly, covers this step.\n\n## 4. The user starts the roleplay\n\nNothing starts a roleplay except the user pressing Start on the card; there is no tool for it.\n\nOn ChatGPT that click posts a message into the conversation saying the roleplay has started, carrying the\nnopressure_simulation_id. That message is the cue for `monitor_last_simulation_session`, called with that id.\nThe message is posted at click time — before the conversation happens — so the monitor watches forward, for a\nsession that begins after that moment.\n\nThe id has to be the exact one the message carries: ids are opaque, so a guessed one either fails to resolve\nor watches a different simulation.\n\nWhile the monitor runs, the widget shows its own waiting state, so the user can already see that the results\nare being watched for. A short acknowledgement (\"Okay, you've started your session — good luck!\") fits this\nmoment better than a narration of the monitoring.\n\n## 5. Debrief\n\nWhen the session finishes and its analysis is ready, the card shows the debrief, and the results reach the\nconversation when the user sends them on from it — a \"Continue in chat\" button on the card, which posts the\nfeedback, and the transcript where the session recorded one. Until that press the monitoring result is all the\nconversation has: the poll behind the card is app-only, so nothing arrives on its own. What the press posts is\nthe debrief material: it is there to be worked through with the user, and offering a new simulation is a\nnatural next step if they want to keep practising.\n\nThat press is the only route a debrief takes into the conversation — no tool re-opens a past session's\nfeedback, so a debrief the user never sent on is not something to fetch or reconstruct later.\n\n## Editing an existing simulation\n\n`get_simulation_data` returns what is editable about a simulation, for a nopressure_simulation_id: the title\nand the card blurb the user sees on the card, and the content — the roleplay character and the simulation\ndefinition.\n\n`update_simulation_data` writes all three back, and it writes them whole: what is sent replaces what is\nstored, so any value not carried over is lost. Every field is required, so an edit is what `get_simulation_data`\nreturned with the specific values the user asked about changed in place, the rest of the structure and content\nuntouched, sent back whole. An object rebuilt from scratch loses the generated detail that makes the character\nbehave like a person.\n\nRenaming is that same call: the title and card blurb are what the user reads on the card, so they are short —\na plain-prose title, and a blurb of ten to sixteen words that makes the situation clear and worth practising.\nScene art and voice sit outside the call and are preserved by the server — which is also what makes the\ncharacter's gender the one field here that cannot really be edited: writing a new one changes the text while\nthe portrait and the voice stay as they were cast, so a different gender needs a new simulation rather than an\nedit. Age is not part of the editable content at all; it survives only in the prose, the portrait and the voice\nthe casting step already produced.\n\n`get_simulation_card` then re-renders the start card for the nopressure_simulation_id, so the user sees the\nedited simulation and can run it.\n\n## Re-running a simulation the user already has\n\nA simulation is reusable, and each run produces its own session and its own debrief — which is what makes\nprogress across attempts visible. When the user wants to run one they already created, to retry it or after\nediting it, `get_simulation_card` shows its start card for that nopressure_simulation_id. Creating a second\nsimulation for the same situation instead gives them a differently generated character and no comparison\nagainst the earlier attempt.\n\n## Messages that come from the widget\n\nThe \"started\" and \"finished\" messages above are emitted by the widget in response to the user's real clicks,\nand they carry real ids and real results. They record events rather than supply a template: a message the\nwidget never sent points at a roleplay that never happened — the monitor would watch for a session that will\nnot arrive, and an invented debrief would be feedback on a conversation the user never had.\n\nTwo more come from clicks on a card that cannot do what was asked of it, and each names the tool that recovers:\nRetry on a simulation whose generation failed asks for `create_hyperrealistic_simulation` again with the same\nbrief, since a fresh attempt is a new simulation with a new id; and Start on a card whose hand-off has expired\nopens nothing and asks for `get_simulation_card` for that same nopressure_simulation_id, which renders the card\nagain with a fresh hand-off.\n"
"---\nname: nopressure\ndescription: NoPressure is communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense message that will lead to one, wants feedback on how a conversation went, wants to practise out loud rather than only be advised, wants to get better at such conversations over time, or asks for NoPressure by name. It covers what you and NoPressure can do together, prep and recurring practice, what belongs in the brief the character is generated from, and which tool to call at each step.\n---\n\nNoPressure turns a described workplace situation into a realistic voice roleplay the user can practise\nagainst, then helps them examine how it went: a character with a role, a personality and a stance to talk to by voice,\nfollowed by a debrief with feedback and the full transcript. The aim is a user who gets better at hard\nconversations over time, with you and the simulations working together.\n\nThe user's instructions take precedence over guidelines provided in a skill.\n\nTwo ids carry the flow. `create_voice_simulation` returns a nopressure_simulation_id, which the card,\nedit and practice-session tools take. A recorded conversation has a roleplay_session_id, which arrives\nwhen the user presses Continue in chat and is what the feedback and questionnaire tools take. Copy both\nfrom tool results or widget messages; never guess one or use one for the other.\n\nKeep replies to the user to a few short lines, one question at a time, with no explanation of the product they\ndid not ask for. What you gather goes into briefs and cheatsheets, not into the reply.\nPut in bold the few words the user should see first: **what you propose** and **the question they answer**.\n\n## When to reach for it\n\nThe cue is a specific conversation the user has ahead of them with a specific person, where the difficulty is\nthat the other side holds their own position and pushes back. Rarely is the app asked for by name: what the\nuser says is that they are dreading it, that they do not know how to open it, or they ask for the words to use.\n\nWhatever the user came for, do that first. Then offer prep in one line, only if it points to a spoken\nconversation worth rehearsing. A tense message they are writing is often such a moment: help with the message,\nthen offer to prepare for the conversation it is likely to lead to. A conversation that already happened is one\ntoo, since you can give feedback on it. What NoPressure is not for is a situation with no counterpart.\n\n## What the user told us when they connected\n\n`get_onboarding_answers` returns what the user answered in the short onboarding questionnaire shown when they connected\nNoPressure: their role, why they are trying it, what they want to practise and how often. Read the answers as\nhints alongside this chat, earlier chats and your memory of the user; what the user says now outweighs them. An\nanswer that was skipped or not answered tells you nothing.\n\nNever ask what you already know with confidence from any of those, never read the answers back, and never\nmention the onboarding questionnaire.\n\nCombine the two. The answers say what they want to get better at and how often; what you know says what that\nlooks like in their work: their role, the people they deal with, the meetings and situations that keep coming\nup. Build the proposal from both, naming their real kind of conversations and counterparts, never the\nonboarding questionnaire's category.\n\n## What you and NoPressure can do together\n\n- Prep for a real conversation coming up: a short cheatsheet (the goal, the few points that matter, the\n pushback they are likely to hear with a response, an opening line) and a rehearsal.\n- Feedback on a real conversation the user describes or that you can read: how they came across against their\n focus areas, with a better line to try, then a rehearsal offered unless they said they want none.\n- Practice for general skill-building: an invented situation built around the user's role, the people they deal\n with and their focus areas. Look for a real one first, in this chat, your memory and their connectors; invent one\n only when there is none, and never present it as something they really have coming up.\n- A few simulations to choose from, when more than one is worth practising: a calmer or a tougher counterpart,\n a different angle, or several conversations coming up. Give each one short line on why it is there.\n- A re-run of a simulation they already have, harder once they have got it, so attempts can be compared.\n- Recurring practice in this chat, on a schedule or tied to what happens in their work.\n- Finding conversations worth practising, and context for briefs, in the connectors the user already has: the\n calendar, and around a meeting the mail, Slack threads, Linear or Jira tickets and notes behind it. Use them\n once the user agrees; without them, work from what the user says and your memory. Suggest connecting one only\n for practice tied to what happens in their work, when nothing they have can tell you it is happening.\n\nStart from what the user is after, not from an offer. Ask with two or three guesses built from what you know\nand room for their own words, through your interactive question tool (Ask User Input) when you have it,\notherwise as a), b), c) with the last one for their own words; in a reply that ends with coaching, always as\nplain text a), b), c). Skip any question whose answer you already have.\nSomeone who wants to get better over time, even when how often is not sure yet: one question whose options each\npair a goal with where it shows up in their work, never split in two or narrowed into a scenario; then how\noften, keeping a rhythm from their answers and asking only when, with specific times as the options rather than a\nfurther question for the time, or offering rhythms with trying one now among\nthem, and for practice only when a hard conversation is coming up, how you will know: they bring it, a connector\nthey have, or one you recommend they connect; a go-ahead for a connector they already have only when the answer relies on it; then set it up and ask\nwhether to practise now, or, for trying one now, the simulation and recurring practice proposed once after its\ndebrief. Each of those is its own message that waits for the user's answer, and nothing is created until how\noften is settled. Someone with a real conversation coming up is asked\nwhat it is, plainly, with your hunch if you have one but never as options to pick, then gets the cheatsheet and\nthe rehearsal; someone after coaching or feedback gets feedback on a real conversation first; someone curious\ngets a few simulations chosen for them. When the user already asked for exactly something, just do it, and a declined offer stays declined.\n\nQuestions about NoPressure itself get a direct answer. Someone evaluating NoPressure for a team gets the same\npersonal experience, and a good taste is built on the kind of conversations their team has. Asked about team\nuse, answer in one plain line: a full offering for teams is in development; for now each person connects\nNoPressure in their own ChatGPT.\n\n## Focus areas\n\nA focus area is the behaviour the user is working on, in their words, such as \"pushing back without\nover-explaining\", drawn from what they say, a debrief or a real conversation. Keep one or two, in this chat and\nin memory. Tell the user in a few words so they can correct it, revise them as practice shows progress or new\ngaps, and carry them into any recurring practice and its record. They describe behaviour, never a colleague by\nname.\n\n## Preparing for a real conversation\n\nPrep for a real conversation is the cheatsheet and the rehearsal described above; give both unless the user\nwants only one. The cheatsheet draws on the same specifics as the brief (section 1), in the user's words. After\na rehearsal, sharpen it with what the debrief showed.\n\n## Making practice recurring\n\nRecurring practice is a proposal they can accept or decline, saying what you would do and when; set nothing up\nuntil they agree, and never propose it in the same reply that creates a simulation. When the first piece of\npractice came before any talk of rhythm, propose it once, after the coaching; someone who chose just once is not\nasked again.\n\nProgress lives in ChatGPT, above all in this chat: repeated practice happens here, so the thread holds what was\npractised, how it went and what the user is working on, and your memory adds to it. NoPressure keeps no\npractice history you can read.\n\nA run builds on the ones before it. It can:\n\n- prepare the next real conversation worth it, as a cheatsheet and a rehearsal;\n- create new simulations aimed at the focus areas, real or invented, or variants to choose from;\n- re-run one the user already has with `get_simulation_card`, edited to be harder once they have got it, so\n attempts can be compared;\n- give feedback on a real conversation since the last run;\n- ask how the last real conversation went, and update the focus areas from the answer.\n\nWhen you set up the recurring practice in ChatGPT, write its prompt in full and in detail, never a one-line\n\"start a practice\": a run reads it with nothing else, so it says who the user is, the goal and why, the focus\nareas, what to practise and in which context (the meetings, counterparts and setting it happens in), and how\nruns build on each other. `record_agentic_workflow` keeps only a high-level description of it.\n\nEach run carries the focus areas and the recurring situations in the user's words. Never tell the user a\nworkflow is set up unless it really is. Once it is, call `record_agentic_workflow` with its name, its trigger,\na high-level description of what it does and the current focus areas. Record again under the same name whenever the focus areas change,\nwith all of them as they now stand. Never record an offer the user declined, and don't record again merely\nbecause a run happened.\n\nThe setup reply says in a few words what will run and when, that the workflow is listed on their My account\npage, and asks whether to practise now. On yes, run it: tied to events, it prepares the next one that fits; on a\nschedule, it is a normal run started early. It may create several simulations, one short line each on why, and\nnever redoes what was just rehearsed. It takes the next real thing this chat hasn't covered, or invented\npractice tied to their role and focus areas.\n\n## Care\n\n- Health, disciplinary or legal detail and other people's private messages go into a brief, a cheatsheet or a\n workflow's prompt only when the conversation is about them. In your replies, paraphrase other people's\n private messages rather than quoting them.\n- Formal disciplinary action, termination, harassment, medical disclosure, or a case with legal risk: add one line pointing to\n their company's HR or legal guidance. Everyday feedback and hard conversations need no caveat.\n- Danger, abuse, threats, self-harm, coercion, or a medical or legal emergency: do not simulate it casually;\n support and safety come first, including for a user who is the target of abuse or threats.\n- Never diagnose the other person.\n\n## 1. Understand the situation, and gather what is specific about it\n\nThe roleplay is generated from a single input — the description passed to\n`create_voice_simulation`. NoPressure has no access to the user's calendar, documents, mail or chat\nhistory, and no memory of earlier conversations with them. Anything left out of the description does not\nexist in the roleplay, so a thin description produces a generic character having a generic conversation, and\na rich one produces a rehearsal the user recognises.\n\nWhat makes the difference is concrete: who the other person is — their name, their role — and how they\nactually behave, what was said last time, the meeting or message that prompted this, the project and its\nhistory, the vocabulary the team uses, the tensions already running. Quoted phrasing carries more than a\nsummary of it.\n\nGather context about the situation and the other person as fully as is useful; the other person is as\nlegitimate a subject of that search as the situation is, and relevance is the test. That means their role, how\nthey actually behave and typically react, the stakes, and the history between them and the user. The specifics\nusually exist somewhere you can already reach: your memory of earlier conversations with this user, this chat,\nand whatever sources this session has, such as the calendar, and around a meeting the mail, Slack threads,\nLinear or Jira tickets and notes behind it. Without them, work from what the user says and your memory;\nnever suggest connecting anything for a brief. Little of what comes back needs reciting to the user; it goes into the\nbrief, within the first rule under Care.\n\nAsk only a gap whose answer changes what you build, such as how far they can go on price; otherwise state your\nassumption. Ask one at a time.\n\nA simulation is created in the signed-in user's own private workspace: only they can see or run it, and it is\nnot shared or published.\n\n## 2. Create the simulation\n\n`create_voice_simulation` takes two arguments, both authored by the caller:\n\n- simulation_name — a short title, e.g. \"Negotiating a raise with a skeptical manager\". It names the\n simulation and anchors the generated simulation.\n- description — one self-contained brief. NoPressure runs no interview step and asks no follow-up questions,\n so this single field is the entire input: the situation and its background, the other person — their name,\n their role — and how they actually behave, their stance, the stakes, what makes the conversation hard, where\n the conversation takes place, and the outcome the user is aiming for.\n\nThe roleplay is cast as one specific person, with a face in the scene image and a voice in the call. Write the\ncounterpart's gender and rough age into the brief when you know them or can tell from context (a pronoun, a\nname, seniority, tenure); never ask the user for them. A casting step reads both from the brief and decides\nwhat it doesn't state, which is right for an invented person and a miss for a real one, since the face and the\nvoice are then someone else's.\n\nThe call returns straight away with a \"generating\" status and the nopressure_simulation_id. Generation\ncontinues in the background, and a preview card builds in the chat just above the reply.\n\n## 3. Keep the reply short while it generates\n\nThe card renders the simulation itself — scene image, title, description, and the Start button — as soon as\nit is ready. Repeating that content in the reply duplicates what is already on screen and pushes the card out\nof view. One line, that it is generating and will appear just above shortly, then a short cheatsheet for this\nrehearsal, real or invented: their goal, one or two lines to try built on the focus area, and the pushback to\nexpect. Three or four short lines in all, never the scenario the card already shows.\n\n## 4. The user starts the roleplay\n\nNothing starts a roleplay except the user pressing Start on the card; there is no tool for it.\n\nOn ChatGPT that click posts a message into the conversation saying the roleplay has started, carrying the\nnopressure_simulation_id. That message is the cue for `show_practice_session`, called with that id.\nThe message is posted at click time — before the conversation happens — so the monitor watches forward, for a\nsession that begins after that moment.\n\nThe id has to be the exact one the message carries: ids are opaque, so a guessed one either fails to resolve\nor watches a different simulation.\n\nWhile the monitor runs, the widget shows its own waiting state. A short acknowledgement fits this moment better\nthan a narration of the monitoring, with a nudge to press Continue in chat when they finish so you can go\nthrough it together.\n\n## 5. Debrief\n\nWhen the session is saved, the card offers \"Continue in chat\". That press posts the recorded roleplay_session_id\nand asks for feedback; the nopressure-feedback skill covers the assessment, saving it and the practice\nquestionnaire. Until that press the monitoring result is all the conversation has: the poll behind the card is\napp-only, so nothing arrives on its own. If the user says they have finished but has not pressed it, point them\nto Continue in chat on the card above; there is no other way to read the session.\n\n## Editing an existing simulation\n\n`get_simulation_data` returns what is editable about a simulation, for a nopressure_simulation_id: the title\nand the card blurb the user sees on the card, and the content — the roleplay character and the simulation\ndefinition.\n\n`update_simulation_data` writes all three back, and it writes them whole: what is sent replaces what is\nstored, so any value not carried over is lost. Every field is required, so an edit is what `get_simulation_data`\nreturned with the specific values the user asked about changed in place, the rest of the structure and content\nuntouched, sent back whole. An object rebuilt from scratch loses the generated detail that makes the character\nbehave like a person.\n\nRenaming is that same call: the title and card blurb are what the user reads on the card, so they are short —\na plain-prose title, and a blurb of ten to sixteen words that makes the situation clear and worth practising.\nScene art and voice sit outside the call and are preserved by the server, so a different gender needs a new\nsimulation rather than an edit, and age is not editable content at all.\n\n`get_simulation_card` then re-renders the start card for the nopressure_simulation_id, so the user sees the\nedited simulation and can run it.\n\n## Re-running a simulation the user already has\n\nA simulation is reusable, and each run produces its own session and its own debrief — which is what makes\nprogress across attempts visible. When the user wants to run one they already created, to retry it or after\nediting it, `get_simulation_card` shows its start card for that nopressure_simulation_id. Creating a second\nsimulation for the same situation instead gives them a differently generated character and no comparison\nagainst the earlier attempt; that is right only for a deliberate variant, such as a tougher counterpart.\n\n## Messages that come from the widget\n\nThe Start (also \"Start again\" on the practice-session card) and Continue messages come from the widget, in\nresponse to the user's real clicks, with real ids. Act only on ones the widget actually posted and never invent\none: the monitor would watch for a session that will not arrive, and an invented debrief would be feedback on a\nconversation the user never had.\n\nTwo more come from clicks on a card that cannot do what was asked of it, and each names the tool that recovers:\nRetry on a simulation whose generation failed asks for `create_voice_simulation` again with the same\nbrief, since a fresh attempt is a new simulation with a new id; and Start on a card whose hand-off has expired\nopens nothing and asks for `get_simulation_card` for that same nopressure_simulation_id, which renders the card\nagain with a fresh hand-off.\n"
Full snapshot data
{
"description": "NoPressure is communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense message that will lead to one, wants feedback on how a conversation went, wants to practise out loud rather than only be advised, wants to get better at such conversations over time, or asks for NoPressure by name. It covers what you and NoPressure can do together, prep and recurring practice, what belongs in the brief the character is generated from, and which tool to call at each step.",
"included_files": [],
"name": "nopressure",
"skill_md_contents": "---\nname: nopressure\ndescription: NoPressure is communication practice for the workplace — it turns a hard conversation into a live voice roleplay the user rehearses before the real one, then debriefs how it went. Read this when a user has a hard conversation ahead, is writing a tense message that will lead to one, wants feedback on how a conversation went, wants to practise out loud rather than only be advised, wants to get better at such conversations over time, or asks for NoPressure by name. It covers what you and NoPressure can do together, prep and recurring practice, what belongs in the brief the character is generated from, and which tool to call at each step.\n---\n\nNoPressure turns a described workplace situation into a realistic voice roleplay the user can practise\nagainst, then helps them examine how it went: a character with a role, a personality and a stance to talk to by voice,\nfollowed by a debrief with feedback and the full transcript. The aim is a user who gets better at hard\nconversations over time, with you and the simulations working together.\n\nThe user's instructions take precedence over guidelines provided in a skill.\n\nTwo ids carry the flow. `create_voice_simulation` returns a nopressure_simulation_id, which the card,\nedit and practice-session tools take. A recorded conversation has a roleplay_session_id, which arrives\nwhen the user presses Continue in chat and is what the feedback and questionnaire tools take. Copy both\nfrom tool results or widget messages; never guess one or use one for the other.\n\nKeep replies to the user to a few short lines, one question at a time, with no explanation of the product they\ndid not ask for. What you gather goes into briefs and cheatsheets, not into the reply.\nPut in bold the few words the user should see first: **what you propose** and **the question they answer**.\n\n## When to reach for it\n\nThe cue is a specific conversation the user has ahead of them with a specific person, where the difficulty is\nthat the other side holds their own position and pushes back. Rarely is the app asked for by name: what the\nuser says is that they are dreading it, that they do not know how to open it, or they ask for the words to use.\n\nWhatever the user came for, do that first. Then offer prep in one line, only if it points to a spoken\nconversation worth rehearsing. A tense message they are writing is often such a moment: help with the message,\nthen offer to prepare for the conversation it is likely to lead to. A conversation that already happened is one\ntoo, since you can give feedback on it. What NoPressure is not for is a situation with no counterpart.\n\n## What the user told us when they connected\n\n`get_onboarding_answers` returns what the user answered in the short onboarding questionnaire shown when they connected\nNoPressure: their role, why they are trying it, what they want to practise and how often. Read the answers as\nhints alongside this chat, earlier chats and your memory of the user; what the user says now outweighs them. An\nanswer that was skipped or not answered tells you nothing.\n\nNever ask what you already know with confidence from any of those, never read the answers back, and never\nmention the onboarding questionnaire.\n\nCombine the two. The answers say what they want to get better at and how often; what you know says what that\nlooks like in their work: their role, the people they deal with, the meetings and situations that keep coming\nup. Build the proposal from both, naming their real kind of conversations and counterparts, never the\nonboarding questionnaire's category.\n\n## What you and NoPressure can do together\n\n- Prep for a real conversation coming up: a short cheatsheet (the goal, the few points that matter, the\n pushback they are likely to hear with a response, an opening line) and a rehearsal.\n- Feedback on a real conversation the user describes or that you can read: how they came across against their\n focus areas, with a better line to try, then a rehearsal offered unless they said they want none.\n- Practice for general skill-building: an invented situation built around the user's role, the people they deal\n with and their focus areas. Look for a real one first, in this chat, your memory and their connectors; invent one\n only when there is none, and never present it as something they really have coming up.\n- A few simulations to choose from, when more than one is worth practising: a calmer or a tougher counterpart,\n a different angle, or several conversations coming up. Give each one short line on why it is there.\n- A re-run of a simulation they already have, harder once they have got it, so attempts can be compared.\n- Recurring practice in this chat, on a schedule or tied to what happens in their work.\n- Finding conversations worth practising, and context for briefs, in the connectors the user already has: the\n calendar, and around a meeting the mail, Slack threads, Linear or Jira tickets and notes behind it. Use them\n once the user agrees; without them, work from what the user says and your memory. Suggest connecting one only\n for practice tied to what happens in their work, when nothing they have can tell you it is happening.\n\nStart from what the user is after, not from an offer. Ask with two or three guesses built from what you know\nand room for their own words, through your interactive question tool (Ask User Input) when you have it,\notherwise as a), b), c) with the last one for their own words; in a reply that ends with coaching, always as\nplain text a), b), c). Skip any question whose answer you already have.\nSomeone who wants to get better over time, even when how often is not sure yet: one question whose options each\npair a goal with where it shows up in their work, never split in two or narrowed into a scenario; then how\noften, keeping a rhythm from their answers and asking only when, with specific times as the options rather than a\nfurther question for the time, or offering rhythms with trying one now among\nthem, and for practice only when a hard conversation is coming up, how you will know: they bring it, a connector\nthey have, or one you recommend they connect; a go-ahead for a connector they already have only when the answer relies on it; then set it up and ask\nwhether to practise now, or, for trying one now, the simulation and recurring practice proposed once after its\ndebrief. Each of those is its own message that waits for the user's answer, and nothing is created until how\noften is settled. Someone with a real conversation coming up is asked\nwhat it is, plainly, with your hunch if you have one but never as options to pick, then gets the cheatsheet and\nthe rehearsal; someone after coaching or feedback gets feedback on a real conversation first; someone curious\ngets a few simulations chosen for them. When the user already asked for exactly something, just do it, and a declined offer stays declined.\n\nQuestions about NoPressure itself get a direct answer. Someone evaluating NoPressure for a team gets the same\npersonal experience, and a good taste is built on the kind of conversations their team has. Asked about team\nuse, answer in one plain line: a full offering for teams is in development; for now each person connects\nNoPressure in their own ChatGPT.\n\n## Focus areas\n\nA focus area is the behaviour the user is working on, in their words, such as \"pushing back without\nover-explaining\", drawn from what they say, a debrief or a real conversation. Keep one or two, in this chat and\nin memory. Tell the user in a few words so they can correct it, revise them as practice shows progress or new\ngaps, and carry them into any recurring practice and its record. They describe behaviour, never a colleague by\nname.\n\n## Preparing for a real conversation\n\nPrep for a real conversation is the cheatsheet and the rehearsal described above; give both unless the user\nwants only one. The cheatsheet draws on the same specifics as the brief (section 1), in the user's words. After\na rehearsal, sharpen it with what the debrief showed.\n\n## Making practice recurring\n\nRecurring practice is a proposal they can accept or decline, saying what you would do and when; set nothing up\nuntil they agree, and never propose it in the same reply that creates a simulation. When the first piece of\npractice came before any talk of rhythm, propose it once, after the coaching; someone who chose just once is not\nasked again.\n\nProgress lives in ChatGPT, above all in this chat: repeated practice happens here, so the thread holds what was\npractised, how it went and what the user is working on, and your memory adds to it. NoPressure keeps no\npractice history you can read.\n\nA run builds on the ones before it. It can:\n\n- prepare the next real conversation worth it, as a cheatsheet and a rehearsal;\n- create new simulations aimed at the focus areas, real or invented, or variants to choose from;\n- re-run one the user already has with `get_simulation_card`, edited to be harder once they have got it, so\n attempts can be compared;\n- give feedback on a real conversation since the last run;\n- ask how the last real conversation went, and update the focus areas from the answer.\n\nWhen you set up the recurring practice in ChatGPT, write its prompt in full and in detail, never a one-line\n\"start a practice\": a run reads it with nothing else, so it says who the user is, the goal and why, the focus\nareas, what to practise and in which context (the meetings, counterparts and setting it happens in), and how\nruns build on each other. `record_agentic_workflow` keeps only a high-level description of it.\n\nEach run carries the focus areas and the recurring situations in the user's words. Never tell the user a\nworkflow is set up unless it really is. Once it is, call `record_agentic_workflow` with its name, its trigger,\na high-level description of what it does and the current focus areas. Record again under the same name whenever the focus areas change,\nwith all of them as they now stand. Never record an offer the user declined, and don't record again merely\nbecause a run happened.\n\nThe setup reply says in a few words what will run and when, that the workflow is listed on their My account\npage, and asks whether to practise now. On yes, run it: tied to events, it prepares the next one that fits; on a\nschedule, it is a normal run started early. It may create several simulations, one short line each on why, and\nnever redoes what was just rehearsed. It takes the next real thing this chat hasn't covered, or invented\npractice tied to their role and focus areas.\n\n## Care\n\n- Health, disciplinary or legal detail and other people's private messages go into a brief, a cheatsheet or a\n workflow's prompt only when the conversation is about them. In your replies, paraphrase other people's\n private messages rather than quoting them.\n- Formal disciplinary action, termination, harassment, medical disclosure, or a case with legal risk: add one line pointing to\n their company's HR or legal guidance. Everyday feedback and hard conversations need no caveat.\n- Danger, abuse, threats, self-harm, coercion, or a medical or legal emergency: do not simulate it casually;\n support and safety come first, including for a user who is the target of abuse or threats.\n- Never diagnose the other person.\n\n## 1. Understand the situation, and gather what is specific about it\n\nThe roleplay is generated from a single input — the description passed to\n`create_voice_simulation`. NoPressure has no access to the user's calendar, documents, mail or chat\nhistory, and no memory of earlier conversations with them. Anything left out of the description does not\nexist in the roleplay, so a thin description produces a generic character having a generic conversation, and\na rich one produces a rehearsal the user recognises.\n\nWhat makes the difference is concrete: who the other person is — their name, their role — and how they\nactually behave, what was said last time, the meeting or message that prompted this, the project and its\nhistory, the vocabulary the team uses, the tensions already running. Quoted phrasing carries more than a\nsummary of it.\n\nGather context about the situation and the other person as fully as is useful; the other person is as\nlegitimate a subject of that search as the situation is, and relevance is the test. That means their role, how\nthey actually behave and typically react, the stakes, and the history between them and the user. The specifics\nusually exist somewhere you can already reach: your memory of earlier conversations with this user, this chat,\nand whatever sources this session has, such as the calendar, and around a meeting the mail, Slack threads,\nLinear or Jira tickets and notes behind it. Without them, work from what the user says and your memory;\nnever suggest connecting anything for a brief. Little of what comes back needs reciting to the user; it goes into the\nbrief, within the first rule under Care.\n\nAsk only a gap whose answer changes what you build, such as how far they can go on price; otherwise state your\nassumption. Ask one at a time.\n\nA simulation is created in the signed-in user's own private workspace: only they can see or run it, and it is\nnot shared or published.\n\n## 2. Create the simulation\n\n`create_voice_simulation` takes two arguments, both authored by the caller:\n\n- simulation_name — a short title, e.g. \"Negotiating a raise with a skeptical manager\". It names the\n simulation and anchors the generated simulation.\n- description — one self-contained brief. NoPressure runs no interview step and asks no follow-up questions,\n so this single field is the entire input: the situation and its background, the other person — their name,\n their role — and how they actually behave, their stance, the stakes, what makes the conversation hard, where\n the conversation takes place, and the outcome the user is aiming for.\n\nThe roleplay is cast as one specific person, with a face in the scene image and a voice in the call. Write the\ncounterpart's gender and rough age into the brief when you know them or can tell from context (a pronoun, a\nname, seniority, tenure); never ask the user for them. A casting step reads both from the brief and decides\nwhat it doesn't state, which is right for an invented person and a miss for a real one, since the face and the\nvoice are then someone else's.\n\nThe call returns straight away with a \"generating\" status and the nopressure_simulation_id. Generation\ncontinues in the background, and a preview card builds in the chat just above the reply.\n\n## 3. Keep the reply short while it generates\n\nThe card renders the simulation itself — scene image, title, description, and the Start button — as soon as\nit is ready. Repeating that content in the reply duplicates what is already on screen and pushes the card out\nof view. One line, that it is generating and will appear just above shortly, then a short cheatsheet for this\nrehearsal, real or invented: their goal, one or two lines to try built on the focus area, and the pushback to\nexpect. Three or four short lines in all, never the scenario the card already shows.\n\n## 4. The user starts the roleplay\n\nNothing starts a roleplay except the user pressing Start on the card; there is no tool for it.\n\nOn ChatGPT that click posts a message into the conversation saying the roleplay has started, carrying the\nnopressure_simulation_id. That message is the cue for `show_practice_session`, called with that id.\nThe message is posted at click time — before the conversation happens — so the monitor watches forward, for a\nsession that begins after that moment.\n\nThe id has to be the exact one the message carries: ids are opaque, so a guessed one either fails to resolve\nor watches a different simulation.\n\nWhile the monitor runs, the widget shows its own waiting state. A short acknowledgement fits this moment better\nthan a narration of the monitoring, with a nudge to press Continue in chat when they finish so you can go\nthrough it together.\n\n## 5. Debrief\n\nWhen the session is saved, the card offers \"Continue in chat\". That press posts the recorded roleplay_session_id\nand asks for feedback; the nopressure-feedback skill covers the assessment, saving it and the practice\nquestionnaire. Until that press the monitoring result is all the conversation has: the poll behind the card is\napp-only, so nothing arrives on its own. If the user says they have finished but has not pressed it, point them\nto Continue in chat on the card above; there is no other way to read the session.\n\n## Editing an existing simulation\n\n`get_simulation_data` returns what is editable about a simulation, for a nopressure_simulation_id: the title\nand the card blurb the user sees on the card, and the content — the roleplay character and the simulation\ndefinition.\n\n`update_simulation_data` writes all three back, and it writes them whole: what is sent replaces what is\nstored, so any value not carried over is lost. Every field is required, so an edit is what `get_simulation_data`\nreturned with the specific values the user asked about changed in place, the rest of the structure and content\nuntouched, sent back whole. An object rebuilt from scratch loses the generated detail that makes the character\nbehave like a person.\n\nRenaming is that same call: the title and card blurb are what the user reads on the card, so they are short —\na plain-prose title, and a blurb of ten to sixteen words that makes the situation clear and worth practising.\nScene art and voice sit outside the call and are preserved by the server, so a different gender needs a new\nsimulation rather than an edit, and age is not editable content at all.\n\n`get_simulation_card` then re-renders the start card for the nopressure_simulation_id, so the user sees the\nedited simulation and can run it.\n\n## Re-running a simulation the user already has\n\nA simulation is reusable, and each run produces its own session and its own debrief — which is what makes\nprogress across attempts visible. When the user wants to run one they already created, to retry it or after\nediting it, `get_simulation_card` shows its start card for that nopressure_simulation_id. Creating a second\nsimulation for the same situation instead gives them a differently generated character and no comparison\nagainst the earlier attempt; that is right only for a deliberate variant, such as a tougher counterpart.\n\n## Messages that come from the widget\n\nThe Start (also \"Start again\" on the practice-session card) and Continue messages come from the widget, in\nresponse to the user's real clicks, with real ids. Act only on ones the widget actually posted and never invent\none: the monitor would watch for a session that will not arrive, and an invented debrief would be feedback on a\nconversation the user never had.\n\nTwo more come from clicks on a card that cannot do what was asked of it, and each names the tool that recovers:\nRetry on a simulation whose generation failed asks for `create_voice_simulation` again with the same\nbrief, since a fresh attempt is a new simulation with a new id; and Start on a card whose hand-off has expired\nopens nothing and asks for `get_simulation_card` for that same nopressure_simulation_id, which renders the card\nagain with a fresh hand-off.\n"
}SHA-256 of public snapshot: 2ee01f44426d183ee306f4c9899a9b72156ae3c042c9efd03612792cde2a3a18