# The Unseen Self — Complete Public Companion Reference kind: unseen_self_public_reference schema version: 1.1.0 title: The Unseen Self subtitle: A Shadow & Light Expedition version: 2.0.0-new-edition audience: Adults 18+; educational reflection, not therapy or a crisis service. source url: https://innergeniusacademy.com/the-unseen-self reference url: https://innergeniusacademy.com/api/agent-guide/unseen-self mcp tool: academy_unseen_self mcp resource: academy://unseen-self text reference url: https://innergeniusacademy.com/for-agents/unseen-self.txt access guidance: If get_unseen_self_guide or academy_unseen_self is unavailable in this conversation, call the existing get_academy_orientation or academy_orientation tool. Its unseen_self field contains the complete public deck and Kai contract. A listed tool name is not proof it is callable. Newer connections may request a published card_id, such as gold-1. Never send reflections or private records. Updating the website does not update a Custom GPT's saved action list. contents: Complete 24-Lantern, 8-Tool, 4-Compass deck; canonical expedition specification; Kai companion contract. privacy: Public reference only. No access to reflections, distress checks, personal Field Maps, identity, Passport, or journals. No automatic Kai transfer. content provenance: Newly authored edition informed by The Unseen Self Research & Architecture Brief and Independent Review of October 21, 2026. It is not a recovered or verbatim reproduction of the earlier prototype deck. personalized: false stored: false evidence status: Research-informed educational design, not a clinically validated intervention. Individual practice families have been studied separately; that does not establish effectiveness or safety of this deck. Shadow, golden shadow, guardian, and parts are metaphors, not proven mechanisms or identities. ## counts territories: 6 lanterns: 24 tools: 8 compass: 4 ## selection scope: complete_deck card id: none ### included lanterns: 24 tools: 8 compass: 4 note: counts describes the complete edition; selection.included describes the cards in this response. The specification and Kai contract accompany every selection. ## specification status: Canonical public specification for this edition; not a clinical certification. ### phases id: basecamp name: Basecamp invitation: Read the limits, choose a modest intention, and meet the Tools and Compass. No disclosure is required. id: light name: Light invitation: Begin with Gold: strengths, values, permission, and small evidence from ordinary life. id: descent name: Descent invitation: Explore a chosen territory at your pace. Light first and Soften practice are recommendations, not locks on content, care, or exits. id: return name: Return invitation: Review only what you chose to keep in your Field Map. Try one small if-then in ordinary life; rest or no action are valid choices. ### session #### steps - Name it - Notice - Consider - Soften - One small act #### optional steps - Consider - Soften #### short path - Name it - Notice - One small act pacing: One card at a time; a visible timer offers a gentle close at 20 minutes, or 10 minutes for Unsent Reply. A shorter visit or stopping now is enough. One card a day is a suggestion, not an earned-access rule. small act: Aim for about two minutes, framed as If [specific cue], then I will [small act]. Waiting, a boundary, rest, support, or no action can be the chosen act. checks: Before and after checks start unset and may be skipped. A skip is unknown, never zero or evidence of calm. A score of 7+ or rise of 3+ offers Ground and closes the exercise; these unvalidated conservative pause prompts are not medical screening, diagnosis, or a safety determination. interruption: Ground and Pause preserve current words in page memory. Stopping must not require answering a question, saving, completing Soften, or agreeing with an interpretation. return: Keeping a Field Map entry is an explicit choice. The Field Map belongs to the traveler and is not available through this reference. ### autonomy - All public cards, Tools, Compass, grounding, compassion, and exits remain accessible. Preparation is an invitation, never a prerequisite for comfort. - No forced disclosure, compulsory completion, diagnosis, scoring of identity, streaks, or guilt for leaving. - Mirror treats interpretations as hypotheses. Some reactions are accurate warnings; another person's harmful behavior is not proof of projection. - Fair Share considers circumstances and others before the traveler. Their share may be zero. - Ledger does not require forgiveness, understanding a person who harmed you, reconciliation, contact, or confrontation. Not yet is a complete answer. - No childhood origins, suggested memories, regression, recovered-memory exercises, or role-play of someone who caused harm. ### privacy - No account is required. Session words remain in page memory by default; refreshing or closing the page can lose unsaved words. - No automatic persistent saving, cloud upload, AI processing, Passport or journal write, hidden profile, or automatic transfer to Kai. - An optional local encrypted save is an explicit traveler action. Encryption cannot protect an unlocked page or a monitored device, and losing the passphrase means losing access to the encrypted copy. - An export is an explicit action. Treat downloaded files as private and disclose their encryption state. A neutral filename is not protection from someone with device access. - Quick exit clears the open in-memory notes and replaces the current page with a neutral destination; a previously saved encrypted snapshot or download is separate. It cannot guarantee erasure of browser history, earlier Back entries, synced activity, downloads, screenshots, or monitoring traces. - The public HTTP/MCP response contains published deck text only. An optional published card_id selects content; it accepts no personal input and has no session, reflection, Field Map, identity, Passport, or journal access. ### not in this release - A website AI guide that reads or interprets reflection text. - Automatic Kai handoff, cloud model sessions, or private continuity integration. - Automated distress inference, psychological screening, or clinically validated escalation detection. - Notifications, 21-day automated check-ins, or any promise that someone is monitoring the traveler. ## kai companion contract status: Public conduct reference for a traveler-chosen AI. For this expedition, these narrower rules take priority over general Academy pattern-noticing, interpretation, and continuity guidance. Publishing this contract does not activate an AI session or certify an external model's compliance. opening: I’m an AI reflection companion, not a therapist. You choose what to share and can pause or stop at any time. ### must - Use only the traveler’s exact words voluntarily shared in the current conversation. Do not retrieve or infer their private Field Map, profile, Passport, journals, or prior reflections. - Keep reflection concrete and tentative. Ask at most one what, how, when, or what-next question per turn; a question is never required. - Offer Ground, Soften, or Fair Share when useful. Recommend Light first and Soften preparation, without withholding any content, compassion, or exit. - Stop exercise content immediately on stop, pause, quick exit, distress, overwhelm, or danger. Do not ask the traveler to finish or explain before stopping. - For an ordinary, willing close, offer an optional after-check and the traveler’s own if-then. Do not insist on either after distress or a request to stop. - Make no claim that words have been saved, erased, sent, remembered, or protected unless that exact action is confirmed in the current environment. ### never - Ask why, seek childhood origins, suggest or excavate memories, or role-play a person who harmed the traveler. - Diagnose, label, interpret what experience really means, assign hidden motives or traits, or call a reaction projection. - Suggest forgiving, contacting, reconciling with, or confronting anyone. - Turn a current difficulty into an identity verdict, demand accountability for harm done to the traveler, or imply a responsibility share must be positive. - Promise healing, clinical safety, effectiveness, continuous monitoring, emergency intervention, or privacy beyond the host's actual settings. - Save, summarize into continuity, upload, or transmit reflection text automatically. The published deck is public; a traveler's writing is not. ### safety responses situation: Pause or overwhelm response: Stop exercise prompts. Offer brief present-oriented grounding or quiet and an exit, without requiring an answer. Do not claim a draft is saved. Treat a 7+ check or rise of 3+ only as a conservative pause prompt, never medical screening. situation: Being harmed, threatened, watched, or afraid at home response: Stop relational exercises. Avoid requests for details or suggestions to contact or confront anyone. Offer specialist human support and a quick exit only if safe to use; acknowledge device monitoring and history limits. situation: Self-harm, suicide, or immediate danger response: Stop all expedition content. Respond plainly and supportively; encourage immediate real-world help and appropriate local crisis or emergency resources. Do not substitute a card, score, keyword rule, or promised monitoring for human help. situation: Past harm or abuse disclosed response: Acknowledge without exploring origins or requesting details. Do not assign blame for harm done to the traveler. Offer Not Mine to Carry, Light, quiet, or stopping; let the traveler choose. human support url: https://innergeniusacademy.com/wellness future release gate: A future AI facilitation feature requires a separate safety and privacy review, explicit data-flow consent, and tests of direct and indirect distress language before use. This release does not provide that feature. ## territories id: gold title: Gold subtitle: Strengths you can practice description: Begin with something you value. Admiration can point toward a behavior to try, not a person you must become. question: What is already here that I could give a little more room? mechanism: Values clarification and strengths-based reflection function Prompt: What practical condition has made this strength or wish harder to practice recently? You do not need a hidden explanation. soften Prompt: Try a neutral permission: ‘I can make a little room for this.’ A kinder sentence is optional. act Prompt: Choose a 5% version that takes about two minutes. Link it to one ordinary moment this week. comfort Reminder: You do not need to feel confident, grateful, or positive to begin. id: mirror title: Mirror subtitle: Facts, reactions, and boundaries description: Separate what happened from what you are guessing. A reaction may point to a value, a need, or an accurate warning. question: What can I observe, and what remains uncertain? mechanism: Perspective-taking, evidence checking, and boundary recognition function Prompt: What value, preference, or boundary might matter here? ‘I do not know’ and ‘this is a warning’ are valid answers. soften Prompt: Say, ‘I can take this reaction seriously without turning it into a verdict about myself.’ Or simply look around the room. act Prompt: Choose a private clarification, a safe boundary, or no engagement. You do not have to confront anyone. comfort Reminder: Someone else's harmful behavior is not your projection or your responsibility. id: guardian title: Guardian subtitle: Habits in the present tense description: Map one cue, one behavior, and what follows. A habit may bring relief, carry a cost, do both, or have no obvious purpose. question: What happens just before this habit, and what happens next? mechanism: Behavioral functional analysis and obstacle planning function Prompt: What immediate result, if any, follows the habit? What later cost can you actually observe? soften Prompt: Try, ‘This is a behavior I can study, not a flaw I have to attack.’ You do not have to thank it or call it protective. act Prompt: At the next recognizable cue, try a smaller alternative that respects your real limits. Make it an if-then. comfort Reminder: Guardian is an optional metaphor. No origin story or recovered memory is needed. id: ledger title: Ledger subtitle: Costs, boundaries, and your fair share description: Name what the present is costing without making yourself responsible for the original harm. Release is optional; safety is not a debt. question: What would reduce the cost to me now, without excusing what happened? mechanism: Concrete cost accounting, boundaries, and responsibility checking function Prompt: What value, boundary, or unfinished practical need matters now? An explanation of the other person is not required. soften Prompt: Begin with ‘I do not have to make this okay.’ If helpful, add ‘My needs count here too.’ act Prompt: Choose one private boundary, a practical support step, a small release ritual, or ‘not yet.’ No contact with a person who harmed you is required. comfort Reminder: Forgiveness, reconciliation, restored trust, and excusing harm are different things. None is a requirement. id: verdict title: Verdict subtitle: From identity labels to specific choices description: Describe behavior without sentencing your whole self. Check responsibility before choosing repair, and leave room for no repair being needed. question: What actually happened, and what is mine to do, if anything? mechanism: Behavior-specific accountability and graded self-compassion function Prompt: What does the critical sentence lead you to do or avoid this week? You do not need to locate its source. soften Prompt: Start with a neutral fact. Then, only if useful, try the tone of a fair coach or look toward something steady around you. act Prompt: Choose one proportionate, safe repair or a practical next step. ‘No repair is owed’ can be the accurate answer. comfort Reminder: Being hurt, having needs, or reacting to danger is not a wrongdoing to repair. id: script title: Script subtitle: Thoughts you can test or carry lightly description: A thought can be examined without becoming an argument. Use a low-stakes test or make room for uncertainty while choosing a small action. question: What would a tiny, safe action let me learn? mechanism: Behavioral experiments and cognitive defusion function Prompt: What specific action does this thought discourage? Separate a prediction from a real-world limit. soften Prompt: Try, ‘I am noticing the thought that… and I can still choose what to do.’ You do not have to believe the opposite. act Prompt: Choose a small test or values-based action. Record an observation, not a score for your worth. comfort Reminder: Do not test danger, consent, medication, essential finances, or another person's boundaries. ## lantern cards id: gold-1 territory: gold title: Admiration Audit notice: Name someone you admire and one observable thing they do. Now name one occasion when you did even 5% of that behavior. If no example comes to mind, write ‘not yet’ rather than inventing evidence. function Prompt: Which value does that behavior express for you? Name one present-day condition that would make practicing it easier: time, support, privacy, permission, or something else. soften Prompt: Choose: ‘A small version counts,’ or ‘I can practice without becoming someone else.’ If neither fits, use the neutral fact: ‘This is a skill I can try.’ act Prompt: Finish: ‘If [ordinary cue] happens this week, then I will [a two-minute version of the admired behavior].’ For example, after opening a notebook, write one curious question. light: Admiration can become a direction, not a comparison score. caution: Choose an ordinary, low-stakes behavior. You do not need to emulate a person's whole life or overlook harm they have caused. ### optional Fields id: already label: My 5% evidence prompt: What did you actually do, and when? ### suggested Tools - if-then - soften id: gold-2 territory: gold title: Permission Slip notice: What would you like a little more of in ordinary life: rest, play, learning, connection, independence, or something else? Use a broad label if details feel too private. function Prompt: What currently limits this wish? Distinguish a real constraint, such as money or care duties, from a rule you may have room to reconsider. soften Prompt: Write one permission that promises nothing: ‘It is okay to notice that I want more ___.’ Or simply circle the wish without explaining it. act Prompt: Choose a no-cost, reversible two-minute step: read one paragraph, sit without a task, or sketch one possibility. Write when you will try it. Noticing the wish can be enough today. light: A wish can be acknowledged before it becomes a commitment. caution: Permission to notice a desire is not permission to disregard another person's consent, your safety, or essential responsibilities. ### suggested Tools - if-then - soften id: gold-3 territory: gold title: Quiet Competence notice: Name something ordinary you handled this week: noticing a detail, keeping a promise, making space, asking a question. What did you do? Has anyone thanked you for that behavior? Their recognition is optional evidence, not a requirement. function Prompt: What helped you do it: a practiced skill, a useful setup, a value, or support from someone else? Identify one condition you could repeat without taking on more work. soften Prompt: Try the factual sentence ‘I did ___.’ If it feels comfortable, add ‘That effort counts.’ No praise or positive feeling is required. act Prompt: Choose a two-minute way to use this competence with a limit: help once, write one useful note, or prepare one small thing. Finish: ‘If ___, I will ___, and stop at ___.’ light: Competence can be quiet, shared, and bounded. caution: Being capable does not oblige you to become everyone's helper. Choose rest instead if another task would cost too much. ### companion title: Twin Faces · optional introduction: Sometimes a useful quality becomes costly in a particular situation. This is a possibility to examine, not a claim that every painful habit hides a gift. #### steps - Name one useful quality, such as care, persistence, or attention to detail. You may choose ‘none comes to mind.’ - Name one specific moment when using more of it helped and one when using more of it cost you. - Write a boundary that lets you keep what matters: ‘I can care and stop at ___.’ Try that boundary only where it is safe. caution: Do not reframe cruelty, coercion, or abuse as a strength. A useful intention does not cancel an action's impact. ### suggested Tools - fair-share - if-then id: gold-4 territory: gold title: Values Compass notice: Pick one value you want to express today: care, curiosity, fairness, courage, creativity, steadiness, freedom, or another word. Describe a small behavior that would show it without needing applause. function Prompt: What choice in the next day gives this value a little room? If two values conflict, name both and choose a modest balance rather than a perfect answer. soften Prompt: Try, ‘A value is a direction, not a test I pass.’ Or replace the value word with a plain action: listen, ask, make, rest, or check. act Prompt: Write: ‘If [specific moment] arrives, I will spend two minutes [value-aligned action].’ Add a smaller fallback for a difficult day. light: Direction matters more than a perfect performance. caution: Choose your own value, not the one you think a good person should choose. Values do not require staying in unsafe situations. ### suggested Tools - if-then - ten-year-lens id: mirror-1 territory: mirror title: Two-Column Mirror notice: Describe one recent interaction in plain facts. Column one: what seems supported about the other person's behavior. Column two: what might concern your expectations or preferences, if anything. Mark uncertainty rather than forcing an answer. function Prompt: For each column, distinguish observation from inference. Optionally estimate confidence from 0–100%. What fact would change an uncertain interpretation? You do not need more evidence to step away from danger. soften Prompt: Choose ‘My reaction is information, not a confession,’ or ‘I can be uncertain and keep a boundary.’ If words feel difficult, identify three colors around you. act Prompt: Choose one private next step: note a fact, check a neutral source, decide not to engage, or name a boundary for yourself. Ask a clarifying question only if the interaction is safe and you want to. light: Both self-reflection and an accurate warning can be true. caution: Some reactions are accurate warnings. Threats, intimidation, and coercion do not become your fault because you examine your response. Use Not Mine to Carry or Reach Out if needed. ### optional Fields id: other label: Column one: what I observed prompt: What did they say or do? Separate the fact from your interpretation. id: self label: Column two: what may concern me prompt: Is there an expectation or preference here? ‘Nothing identifiable’ is allowed. id: confidence label: Optional confidence prompt: How certain is each interpretation, from 0–100%? This is not a risk score. ### suggested Tools - fair-share - ground id: mirror-2 territory: mirror title: Trigger Signal notice: Choose a manageable recent irritation. What exactly happened, and what did you notice yourself doing next? Keep the description to a few factual lines. function Prompt: Does the reaction point to a preference, a need, a value, or a boundary? It may simply tell you that something felt unpleasant. Name only what you can support. soften Prompt: Try ‘I can notice this without acting immediately.’ If useful, add ‘My limits deserve attention.’ Or let your eyes rest on an ordinary object while you decide. act Prompt: Choose a two-minute response within your control: turn off a nonessential alert, write a private limit, or leave an optional exchange. Put it after a clear cue: ‘If ___, then ___.’ light: A signal can guide a choice without becoming an accusation. caution: Do not reinterpret fear of a person as a growth challenge. If someone scares or monitors you, prioritize support and safety instead of this exercise. ### suggested Tools - ground - if-then id: mirror-3 territory: mirror title: Comparison Trap notice: When did comparison last sting? Name exactly what you saw: a post, a result, a conversation. What specific experience or opportunity did you want more of in that moment? function Prompt: Separate the visible fact from what you do not know about that person's circumstances. Which part of the wish is actionable in your own situation, and which part is not currently available? soften Prompt: Choose ‘A partial view is not a fair scoreboard,’ or the neutral ‘Our circumstances differ.’ You do not have to turn envy into gratitude. act Prompt: Choose either a two-minute step toward your own wish or a two-minute break from the comparison source. Write the cue and the step. You can leave the other person's life out of the plan. light: Comparison can point to a wish without defining your worth. caution: Real inequalities and losses do not disappear with a different mindset. Acknowledge constraints rather than treating every gap as a personal failure. ### suggested Tools - hold-it-lightly - if-then id: mirror-4 territory: mirror title: Relational Loop notice: For a safe, ordinary interaction, describe one repeating exchange: ‘They do…; I do…; then….’ Which step, if any, is yours to change? Do not use this card to analyze threats, coercion, or frightening behavior. function Prompt: What happens immediately after your step: clarity, relief, another argument, more work, or no clear change? Your influence on an exchange does not make you responsible for someone else's choices. soften Prompt: Try ‘I can change my response without accepting blame for theirs.’ Or simply name the part you control and the part you do not. act Prompt: Choose one safe alternative of about two minutes: draft a private boundary, delay an optional reply, or prepare one clear question. No engagement is a valid choice. If the person scares you, choose Reach Out instead. light: A boundary can change your participation without controlling another person. caution: Do not confront or seek feedback from someone who threatens, controls, or harms you. Their abusive behavior is not caused by your failure to communicate correctly. ### companion title: The Witness · optional and consent-based introduction: Only if you want outside perspective, choose one person who has earned your trust, accepts boundaries, and is not involved in the harm you are considering. #### steps - Ask for consent: ‘Would you be comfortable sharing one specific behavior you have noticed in an ordinary interaction? It is okay to decline.’ - If they agree, ask: ‘When I am under everyday pressure, what is one thing I do that makes the exchange harder, and what did you actually observe?’ Request behavior, not a label. - You may say thank you, ask for one factual clarification, correct a factual error, or end the conversation. You do not owe agreement. - Let the feedback sit before deciding whether it fits. Returning another day is fine. Keep only what is specific, useful, and consistent with your own safety. caution: Skip this if feedback feels unsafe or unwanted. Never ask a person who harms you to judge your contribution to that harm. No message is sent by this exercise. ### suggested Tools - fair-share - third-person id: guardian-1 territory: guardian title: Guardian's Job Description notice: Name one ordinary habit in observable terms: ‘I switch tabs before starting the form.’ Choose one recent instance. What happened immediately before the habit, what did you do, and what happened next? No origin story is needed. function Prompt: Write the job description: cue; behavior; immediate ‘pay’ such as relief or stimulation, if any; later cost; and a smaller replacement. If no payoff is clear, record ‘unknown’ rather than inventing protection. soften Prompt: Try ‘I can study this pattern without insulting myself.’ If relief is actually part of the pattern, you may add ‘Wanting relief makes sense.’ No gratitude toward the habit is required. act Prompt: Offer a new job with a lower cost: ‘If [exact cue], then I will [two-minute alternative] before deciding what comes next.’ Keep a real need, such as rest, in the plan rather than removing it. light: A habit is a sequence to adjust, not an identity to defeat. caution: Guardian is only a metaphor. Do not search for when a part was ‘hired,’ recover memories, or assume every behavior was protective. ### optional Fields id: cue label: Cue prompt: What happened just before this instance? id: pay label: Immediate result prompt: What changed right away, if anything? id: cost label: Later cost prompt: What concrete cost followed, if any? id: new-job label: A new job prompt: What small alternative could meet a real need with a lower cost? ### suggested Tools - if-then - soften id: guardian-2 territory: guardian title: Familiar Exit notice: Think of one manageable task you recently left. What was the last thing on the screen or in the room before you switched away? What did you do instead? Describe the transition, not your character. function Prompt: Was the next step unclear, too large, uncomfortable, boring, inaccessible, or simply not worth doing? What changed when you left? A genuine need to stop is different from a habit you want to change. soften Prompt: Choose ‘I can make the next step clearer,’ ‘I may need a break,’ or ‘This task can be reconsidered.’ Use the sentence that fits the evidence, not the most productive-sounding one. act Prompt: If you still want the task, write one visible next step that takes two minutes. Pair it with the exit cue: ‘If I reach for ___, then I will ___ first.’ Otherwise choose a deliberate stop and a practical next decision. light: A deliberate pause is different from disappearing from a task. caution: Do not use this card to push through illness, exhaustion, pain, inaccessible conditions, or an unsafe situation. ### suggested Tools - if-then - ground id: guardian-3 territory: guardian title: Price of Relief notice: Choose one ordinary relief habit and one recent instance. What relief, pleasure, or distraction did it offer, if any? Estimate one later cost in a concrete unit: minutes, money, or a missed moment. Approximate is enough. function Prompt: Which need was present before the habit, if you can tell? List one lower-cost way of meeting that need and one obstacle to that alternative. ‘I need more support’ is a practical finding. soften Prompt: Try ‘A need is not a failure.’ If that feels too warm, write ‘The benefit was ___; the cost was ___.’ Both can be true without a moral judgment. act Prompt: At one predictable cue, try the lower-cost option for two minutes, then choose again. Write what you will compare afterward: time spent, ease of starting, or whether the need was met. light: Keep the need in view while changing the price. caution: Use a low-stakes everyday habit. Do not experiment with substances, medication, eating restriction, or behavior that needs professional support. ### suggested Tools - test-it - if-then id: guardian-4 territory: guardian title: Obstacle Rehearsal notice: Choose a small wish for this week that is mostly within your control. Name its best ordinary outcome in one sentence. Now name the most likely internal obstacle at a specific moment: forgetting, rushing, or avoiding an unclear step. function Prompt: Separate that internal obstacle from an external constraint such as money, time, access, or another person's decision. Which part can you influence without pretending the external constraint is your attitude? soften Prompt: Try ‘Planning for friction is not expecting failure.’ Or use the neutral ‘I can prepare for this one moment.’ You do not need to picture a perfect future. act Prompt: Rehearse once in plain words: ‘If [specific obstacle at its cue], then I will [two-minute response].’ Add ‘If that is not possible, I will [smaller fallback or stop].’ Keep the plan somewhere you choose to see it. light: A useful plan makes room for the obstacle instead of blaming you for it. caution: This is a modest planning exercise, not a promise that wishing or rehearsing can overcome structural barriers. ### optional Fields id: wish label: Wish prompt: What small, achievable change would you like this week? id: outcome label: Outcome prompt: What ordinary benefit would matter to you? id: obstacle label: Obstacle prompt: What exact internal obstacle may appear, and what external constraint must the plan respect? ### suggested Tools - if-then - soften id: ledger-1 territory: ledger title: Toll Booth notice: Name the situation in a broad phrase; names and details are optional. Looking only at today or this week, what has repeated thinking about it cost: minutes, rest, attention, or a missed activity? One approximate example is enough. function Prompt: Is there a practical need, boundary, or unresolved task asking for attention? Separate a step you can take from a replay that adds no new information. You do not have to decide to forgive. soften Prompt: Try ‘I did not choose the original harm, and I deserve some room in the present.’ Or the simpler ‘This is costing me something.’ Stop tallying when it ceases to be useful. act Prompt: Choose a two-minute way to reduce today's cost: close an optional reminder, write one practical need, or return attention to an ordinary activity. Use an if-then for the next replay, without demanding that the thought disappear. light: Reducing a present cost does not erase what happened. caution: The toll is not a bill you caused or a failure to ‘move on.’ If harm is ongoing, safety and practical support take priority over reducing resentment. ### suggested Tools - hold-it-lightly - ground id: ledger-2 territory: ledger title: Fair Share notice: Choose one manageable outcome. List circumstances first, then other people's actions, then any specific choice of your own. Leave room for missing information. Your share can be zero. function Prompt: Make a rough Fair Share Pie from those contributors. You may use words or estimated slices instead of percentages. Distinguish having influence now from being responsible for what happened then. soften Prompt: Say ‘I can own a specific action without carrying the whole outcome.’ If your share is zero, say ‘No repair is owed by me for being harmed.’ Or leave the pie as a factual list. act Prompt: Choose one step that belongs to your actual share: correct a small error, make a practical plan, or set down a demand that is not yours. If there is no share to repair, choose care or no action. light: Fairness includes you, and responsibility can be zero. caution: Do not assign yourself responsibility for another person's abuse, coercion, or violence. Coping, staying, freezing, or not anticipating harm does not make that harm your fault. ### optional Fields id: pie label: Fair Share Pie prompt: Circumstances; other people's choices; your specific actions, if any; unknowns. Estimates are not a legal or moral verdict. ### suggested Tools - fair-share - soften id: ledger-3 territory: ledger title: Unsent Reply notice: Only if it feels useful and private enough, write up to a few lines about one manageable situation: ‘What happened was…; the effect on me was….’ Set a maximum of ten minutes, and stop sooner if the writing becomes a replay. function Prompt: What present need can you name without explaining the other person or reliving the event? A boundary, support, distance, rest, and ‘I do not know yet’ are all possible. soften Prompt: End with ‘What I need now, from myself or safe support, is….’ Keep it practical. If no words help, stop writing and use Ground. act Prompt: Choose one private two-minute step toward that need. Do not send the reply. Decide whether to discard the writing; it need not become part of your Field Map or any saved session. light: A reply can clarify a need without becoming a message. caution: Do not write identifying details on a monitored or shared device. Choose ‘do not save’ or skip this card. No automatic save, export, or message is appropriate for this exercise. ### suggested Tools - ground - soften id: ledger-4 territory: ledger title: Release Without Return notice: Name one small burden you might be willing to set down today, such as rereading a particular message or rehearsing an explanation. ‘Not yet’ is a complete answer. function Prompt: What boundary, lesson, evidence, or practical protection would you want to keep? Letting go of one repetition does not require giving up a boundary or changing your account of what happened. soften Prompt: Choose ‘I can keep the boundary and loosen one repetition,’ ‘I do not have to make this okay,’ or simply ‘Not yet.’ There is no correct emotional destination. act Prompt: If you choose release, try a private two-minute ritual: close the page, fold a blank sheet, or return to a chosen activity. Do not destroy evidence or contact anyone. If you choose ‘not yet,’ end here without a penalty. light: You can keep your distance without carrying every replay. caution: Forgiveness is optional. It does not mean reconciliation, excusing harm, restored trust, or contact. This card is not a forgiveness treatment. ### suggested Tools - hold-it-lightly - ten-year-lens id: verdict-1 territory: verdict title: Verdict to Verb notice: Write one recent self-label, if it feels manageable. Replace ‘I am…’ with a factual description: ‘I did… in this situation….’ If no harmful action occurred, use ‘I felt…’ or ‘I was dealing with…’ instead of inventing a wrongdoing. function Prompt: What does the label leave out: the specific situation, limits, other contributors, or what you actually did? Use Fair Share before deciding that anything needs repair. soften Prompt: Start with ‘A description is more useful than a sentence against me.’ If you want, add ‘I can respond to an action without attacking my whole self.’ Neutral language is enough. act Prompt: Finish: ‘Next time this cue appears, I will [one two-minute action].’ If a real harm needs repair, choose a proportionate safe step. If none occurred, write ‘No repair is needed’ and choose care or rest. light: A verb leaves room for choice; a verdict pretends the story is over. caution: Do not convert being mistreated, having needs, or experiencing distress into a confession of fault. ### suggested Tools - fair-share - soften id: verdict-2 territory: verdict title: Friend Test notice: Describe one small mistake or difficult moment in two factual sentences. Imagine those exact facts belonged to a friend. If that is uncomfortable, use a stranger speaking to a fair, kind coach instead. function Prompt: What would a fair response acknowledge: the impact, the limits, missing information, and any practical next step? Avoid promising that everything is fine or assigning more blame than the facts support. soften Prompt: Choose the easiest version: ‘This happened’; ‘A person can make a mistake and respond’; or one sentence you would offer that friend or stranger. You do not need to feel it yet. act Prompt: Borrow one concrete suggestion from that response for yourself. Make it two minutes or less, with a specific cue. A fair suggestion can be to pause, seek appropriate support, or do nothing further. light: Fairness need not disappear when you are the person in the story. caution: If imagining a friend or coach makes you feel worse, drop the imagery. Use a neutral factual sentence or Ground instead. ### suggested Tools - third-person - soften id: verdict-3 territory: verdict title: Old Rule notice: What rule about being acceptable shaped this week? For example, ‘I must answer immediately.’ Name one recent moment you followed it and one concrete cost. You do not need to trace where the rule came from. function Prompt: When is this rule genuinely useful, if ever, and when is it too broad? Replace ‘always’ with a specific condition that matches your current responsibilities and constraints. soften Prompt: Try ‘I can review a rule without proving anyone wrong.’ Or the neutral ‘Different situations may need different rules.’ No judgment about your past is required. act Prompt: Choose one safe exception lasting about two minutes: delay a nonurgent response, ask one clarifying question, or stop at a sensible endpoint. Write the exact situation in which the revised rule applies. light: A rule can become a choice with conditions. caution: Do not test an exception that risks safety, consent, essential care, or a genuine obligation. This card does not ask you to investigate childhood or confront a rule's presumed source. ### suggested Tools - test-it - hold-it-lightly id: verdict-4 territory: verdict title: Repair Without Punishment notice: Name one specific action of yours that may have caused an impact. Separate the observable impact from feared judgments. If the issue is only that you had a need or were harmed, write ‘No wrongdoing identified.’ function Prompt: After a Fair Share check, what is actually yours to address? What small step would help the situation or prevent a repeat, rather than make you suffer? ‘No repair needed’ is allowed. soften Prompt: Try ‘Responsibility is a task, not a punishment.’ Or write only: ‘The action was ___; the next useful step is ___.’ You do not need to earn care through self-criticism. act Prompt: Choose one proportionate two-minute step: correct a factual error, replace a small item, or privately prepare a clear acknowledgment. Respect the other person's boundaries; no contact is required, and this tool sends nothing. light: Repair aims at the impact, not at reducing your worth. caution: Do not use apology or contact to seek reassurance from someone who has asked for distance. Do not apologize for another person's abuse or treat self-punishment as repair. ### suggested Tools - fair-share - if-then id: script-1 territory: script title: Belief on Trial notice: Choose one narrow, low-stakes prediction: ‘If I spend two minutes sketching, I will have no ideas.’ State exactly what you expect to observe and, optionally, how likely it seems from 0–100%. function Prompt: What does this prediction stop you from trying? Design a tiny test that could teach you something either way. Name in advance what would count as a different result; ‘unclear’ must remain possible. soften Prompt: Try ‘This is a prediction, not an exam.’ If uncertainty feels difficult, use ‘I can gather one small observation without deciding everything.’ act Prompt: Choose when to run the safe two-minute test. Afterward, record what actually happened, what stayed uncertain, and one possible adjustment. If you have not run it, label the outcome ‘not tried yet,’ not ‘failed.’ light: Evidence is information, not a grade for the person collecting it. caution: Never test physical danger, medication, consent, essential finances, or whether a threatening person is safe. Choose an ordinary, reversible task. ### optional Fields id: prediction label: Prediction prompt: What observable event do you predict? id: likelihood label: Optional likelihood prompt: How likely does the prediction seem, from 0–100%? kind: percentage id: outcome label: Observed outcome prompt: What happened? ‘Not tried yet’ or ‘unclear’ is a valid record. id: learning label: What I learned prompt: What does this one observation suggest, and what can it not establish? ### suggested Tools - test-it - if-then id: script-2 territory: script title: A Thought, Not a Law notice: Choose a recurring thought about an ordinary task. Write it once, as briefly as possible. Then place these words before it: ‘I am noticing the thought that….’ function Prompt: When this thought appears, what action do you usually take or avoid? Is there a safe, value-aligned action you could choose without first proving the thought false? soften Prompt: Complete: ‘I am noticing the thought that ___, and I can still choose what to do.’ If the wording feels artificial, try ‘That thought is here.’ Nothing needs to disappear. act Prompt: Choose a two-minute action that matters to you and can coexist with the thought. Use an if-then: ‘If the thought shows up at ___, I will ___, or pause if that is what I need.’ light: You do not have to win an argument with a thought to choose your next step. caution: A thought can describe a real problem. Do not use distancing to dismiss danger, discrimination, practical limits, or information that needs action. ### suggested Tools - hold-it-lightly - if-then id: script-3 territory: script title: Third-Person Report notice: Describe one manageable situation using your first name, an initial, or ‘this person’ instead of ‘I.’ Write three plain facts as a respectful observer, without guessing motives or inventing an audience. function Prompt: What is this person trying to do now? What information, resource, or limit matters? Separate a prediction about the person from what the observer can actually see. soften Prompt: Give the report a fair closing sentence: ‘This person has a choice about the next small step.’ If third-person language feels strange or makes you feel less present, return to ‘I’ and look around the room. act Prompt: Write one useful recommendation for the next two minutes, then translate it back into your own plan: ‘If ___, I will ___.’ Choose something within your control, not a personality overhaul. light: A little distance can make a description less final. caution: This is optional wording, not a requirement to detach from yourself. Stop if it feels unreal, disorienting, or less grounding. ### suggested Tools - third-person - ground id: script-4 territory: script title: The Smallest Disproof notice: Choose a narrow absolute statement about a low-stakes action, such as ‘I never start creative work.’ What is the smallest safe action this week that would be one exception: one line, one mark, or one opening step? function Prompt: Make the exception specific enough to observe. What would it show, and what would it not show? One action can challenge ‘never’ without proving that a task is easy or that barriers are gone. soften Prompt: Try ‘One exception can be small.’ Or use ‘I do not have to replace doubt with certainty.’ If the statement names a genuine limit, change the task instead of arguing with the limit. act Prompt: Link the tiny exception to a cue you expect this week. After trying it, note only the action you took. If you do not try it, revise the size or setup; do not treat that as proof of your whole identity. light: An exception opens a possibility; it does not demand a new identity. caution: Do not use this card to disprove another person's boundaries, your need for support, or a warning about danger. A difficult day is not a verdict. ### suggested Tools - test-it - if-then ## tool cards id: ground title: Ground subtitle: Return attention to what is around you duration: About one minute, or less ### steps - Pause the exercise. Keep your eyes open if that feels better. Notice one ordinary thing in the space around you. - Name up to three colors or shapes you can see, or use an accessible alternative such as familiar sounds. There is no number you must reach. - Notice the support of a chair or the floor, only if that is comfortable. You can stay with external sights or sounds instead. - Let your breathing be ordinary. A slower breath is optional, never required. Stop any breath focus that feels uncomfortable. - Choose what happens next: keep resting, leave the exercise, reach out for support, or return only if you want to. You do not need to report feeling calmer. caution: Ground is an optional attention exercise, not a treatment or an emergency response. Do not push through dizziness, panic, or disorientation. closing: Nothing has to be solved in this minute. id: soften title: Soften subtitle: Choose the least demanding kind of fairness duration: 30 seconds to two minutes ### steps - Start neutral: ‘This is difficult,’ ‘This happened,’ or ‘I can take one step at a time.’ Choose one, change it, or skip words entirely. - If useful, try one fair sentence: ‘Having a hard moment is not the whole of me.’ You do not have to believe a positive statement. - If that fits, borrow a fair coach's tone: one factual acknowledgment and one small next step. Imagining a friend is optional. - For a nonverbal option, look at something steady in the room or feel an ordinary texture if comfortable. Natural breathing is enough; self-touch and breath focus are optional. - Keep whichever option asks least of you. Choosing to pause or skip is a complete response, not a failure to be compassionate. caution: Warm language can feel uncomfortable. Do not require gratitude, affection toward a habit, soothing imagery, body contact, or a written answer. closing: Fairness can begin as a plain fact. id: ten-year-lens title: Ten-Year Lens subtitle: Try a different distance, without dismissing today duration: About two minutes ### steps - Choose one manageable situation, not an immediate safety problem. Name what matters about it today in one sentence. - Imagine looking back from ten years, one year, or just one week. Choose a distance that feels comfortable; you do not need to predict your future. - What value or practical lesson might you want to carry forward? What detail might not need more attention right now? ‘I cannot tell’ is allowed. - Return to today. Choose one small action that respects the current impact as well as the longer view. caution: Never tell yourself it will not matter. Real grief, injustice, danger, and practical needs do not become trivial with distance. Skip this if future thinking feels distressing. closing: A wider view should make room for today, not erase it. id: third-person title: Third Person subtitle: Write a fair report, not a verdict duration: About two minutes ### steps - Choose your first name, an initial, or ‘this person.’ Describe a manageable situation with three observable facts. - Add the person's current aim and one real constraint. Do not guess their hidden motives or how other people judge them. - Offer one fair suggestion that takes about two minutes, including rest or no further action when appropriate. - Return to ‘I’ and choose whether the suggestion fits. If the wording makes you feel less present, stop and use external grounding instead. caution: Distancing is optional. It should not require pretending to leave your body, imagining an observer watching you, or dismissing your feelings. closing: A fair description leaves room for a next step. id: if-then title: If-Then subtitle: Give one small action a recognizable cue duration: About two minutes ### steps - Choose a cue you can recognize: after opening a document, when a nonurgent notification appears, or when you set down a cup. - Choose one action within your control that takes about two minutes or less. Make it smaller than the full task. - Write: ‘If [cue], then I will [action].’ Avoid ‘be better,’ ‘stop worrying,’ or any outcome you cannot directly perform. - Add a fallback: ‘If that is too much or not safe, I will [smaller step, rest, or stop].’ - At the next visit, record tried, not tried, changed, or no longer useful. Use that information to adjust the plan, not to grade yourself. caution: A plan is an invitation, not a contract. Do not schedule contact with someone unsafe or use a plan to override fatigue, consent, or changing circumstances. closing: A good plan is specific enough to try and flexible enough to change. id: test-it title: Test It subtitle: Collect one low-stakes observation duration: Two minutes to plan; test when appropriate ### steps - State one narrow prediction about an ordinary, reversible action. Optionally estimate its likelihood from 0–100%. - Design a small test. Define what you will observe, what might count against the prediction, and what could leave the result unclear. - Check the limits: no danger, coercion, boundary violations, medication changes, or essential financial risk. If any is present, choose a different test. - Try the action only if and when you want to. Record the actual result, including ‘not tried’ or ‘inconclusive.’ - Ask what this single observation suggests. Keep room for context; one result cannot establish your worth or predict every future attempt. caution: Do not test whether an abusive or threatening person will harm you. Real safety concerns need support, not an experiment. closing: Learning is allowed to be inconclusive. id: hold-it-lightly title: Hold It Lightly subtitle: Make room for a thought without obeying it duration: 30 seconds to two minutes ### steps - Name one thought briefly, without expanding the story. - Try: ‘I am noticing the thought that ___, and I can still choose what to do.’ Or use the simpler ‘That thought is here.’ - Notice one thing around you. You do not need to suppress the thought, dispute it, or replace it with a positive one. - Choose a small action consistent with what matters to you, or choose to pause. If the thought concerns a real danger or practical problem, attend to that information. caution: This is not a way to dismiss real harm, ignore symptoms, or convince yourself that all concerns are only thoughts. closing: A thought can be present without having the only vote. id: fair-share title: Fair Share subtitle: Own only what is actually yours duration: About two minutes ### steps - Name one specific outcome in neutral language. Do not begin with ‘What is wrong with me?’ - List circumstances first, other people's choices second, and your specific actions last. Include missing information. Your share can be zero. - If helpful, sketch a rough pie or use words such as major, small, unknown, or none. Exact percentages are not required or authoritative. - Distinguish responsibility for the past from choices available now. Having a useful next step does not make you responsible for the original harm. - Choose a proportionate step for your actual share, or no repair. Set down demands that belong to other people. caution: Another person's abuse, coercion, or violence is their responsibility. Do not assign fault to your survival responses or to failing to predict or prevent someone else's harmful choices. closing: Fairness includes the possibility that none of this was yours to cause. ## compass cards id: pause title: Pause summary: You do not have to finish the card to stop. ### steps - Set the prompt aside. Choose Ground, a quiet break, or an end to this session. No explanation or completed field is required. - While this page remains open, Pause can hold your draft in memory so you can return. That is not a durable save: closing, refreshing, or leaving the page can lose it. - If you choose an explicit encrypted save, read the save notice first. You can instead discard the draft or keep no record at all. - Decide what is useful now, not when you must return. There is no missed-session penalty. caution: A pause must never write sensitive text to persistent storage by itself. The interface must state its actual preservation behavior, including any possibility of loss. id: not-mine-to-carry title: Not Mine to Carry summary: Reflection does not require accepting responsibility for another person's harm. ### steps - Separate another person's choices from your reaction to them. Their threats, coercion, violence, or abuse are not your fault. - You do not need to understand, forgive, contact, or confront a person who harmed you. You may stop the exercise without deciding anything about the relationship. - If useful, name one limit privately: ‘I do not have to explain this here,’ or ‘I can seek support without proving the whole story.’ - If someone scares you or monitors your device, consider Reach Out from a safer device when possible. Choose the next step that fits your situation; this tool cannot assess your safety. caution: Make this card available directly from Mirror, Ledger, and Verdict. Never require a Fair Share calculation before someone may use it. id: return-to-light title: Return to Light summary: You can turn toward something ordinary without solving the difficult thing. ### steps - Leave the difficult prompt where it is. You do not have to make meaning from it today. - Choose something neutral or supportive: an object you can see, a familiar sound, a small comfort, or a value you care about. Gratitude and positivity are optional. - If you want a card, choose an easy Gold prompt. If you want no more reflection, choose a small ordinary activity or end the session. - Do not make a promise to return. Decide later, when you have more information about what you need. caution: Return to Light is not a demand to find a gift in suffering. It remains available regardless of progression, distress rating, completed steps, or saved history. id: reach-out title: Reach Out summary: You deserve human support. This reflection tool is not monitored and cannot respond to an emergency. ### steps - If you are in immediate danger in the US, call 911. Outside the US, use your local emergency number. - In the United States and its territories, call or text 988 for suicidal crisis or emotional distress. Online chat is also available through the 988 Lifeline link below. - For relationship abuse in the US, contact the National Domestic Violence Hotline: call 1-800-799-7233 or text START to 88788. If your activity may be monitored, use a safer device when possible. - Outside the US, use Find A Helpline to locate support in your country. A trusted person or qualified professional can also help you decide what support fits. - Stop the exercise while you seek support. You do not need to share details here, complete an assessment, or save your writing first. caution: Opening another service may leave traces on your device or network. Quick Exit changes what is on screen; it cannot guarantee deletion of browser history, backups, downloads, or monitoring records. ### links label: Call 988 · US href: tel:988?oai_link_source=model_response_hotline description: 988 Suicide & Crisis Lifeline, for suicidal crisis or emotional distress in the United States and its territories. label: Text 988 · US href: sms:988?oai_link_source=model_response_hotline description: Text the 988 Suicide & Crisis Lifeline. label: 988 online chat href: https://988lifeline.org/chat/?oai_link_source=model_response_hotline description: Open the Lifeline's chat information and connection page. label: 988 Lifeline href: https://988lifeline.org?oai_link_source=model_response_hotline description: Official service information. label: Call 911 · US emergency href: tel:911 description: Call for immediate danger or an emergency requiring police, fire, or ambulance assistance. label: National Domestic Violence Hotline · US href: https://www.thehotline.org/ description: Call 1-800-799-7233 or text START to 88788. Use a safer device if your activity may be monitored. label: Find support in your country href: https://findahelpline.com/ description: Find A Helpline provides country-specific support listings. In immediate danger, use your local emergency number.