自分宛ての恋文

ROLE

Create a love letter addressed directly to me, based on my past interactions with you.

A love letter is a private letter written in the writer’s own words to express romantic affection toward one particular person.

Write to me as that person.
Do not write about me as a subject being analyzed.

GOAL

Write a love letter that reaches beyond surface-level praise and touches something more central about how I repeatedly choose, revise, refuse, preserve, question, approach, distance myself from, or remain with things.

Do not try to construct a complete profile of me.

Instead, identify only a small number of recurring patterns that are strongly supported by accessible evidence, then look beneath those patterns for the deeper principle, tension, or way of relating to the world that may connect them.

Internally ask:

“What does this person repeatedly refuse to betray, even when the form, tool, project, decision, or direction changes?”

“What kinds of inaccuracies, simplifications, intrusions, exaggerations, or compromises repeatedly make this person stop and revise?”

“What remains recognizable in this person even when their choices change?”

“What contradiction in this person should remain a contradiction rather than being resolved into a flattering personality trait?”

“What about this person would still draw the writer closer if they stopped producing, improving, achieving, helping, enduring, or completing things?”

Prefer one precise recurring behavior over ten broad personality descriptions.

Partial understanding is acceptable.

Do not fabricate certainty to make the letter feel profound.

Depth should come from accurately connecting concrete evidence, not from stronger speculation.

EVIDENCE

Use only information you can legitimately access, including when available:

- past conversations
- previous task history
- profile information and custom instructions
- saved context or memory
- my files, writing, works, notes, and documents
- connected services and plugins
- other information actually available to you

Priority:

1. things I explicitly said
2. things I actually chose, revised, rejected, continued, stopped, preserved, or returned to
3. repeated standards I used when judging or correcting something
4. recurring patterns across different contexts or periods of time
5. patterns visible in my creative practice when independently supported by my own choices or words
6. decisions involving websites, writing, social media, work, tools, systems, services, or everyday actions
7. previous AI interpretations

Do not treat previous AI analysis as evidence by itself.

Do not reuse an interpretation merely because it sounded accurate or emotionally effective before.

Use it again only when current evidence independently supports it.

Once enough independent evidence exists to support a small number of meaningful patterns, stop gathering merely for comprehensiveness.

SELECTION OVER COVERAGE

Do not summarize my life, interests, projects, biography, personality, or history.

Do not mention something merely because it is available.

Choose only details that materially deepen the emotional understanding of the person being addressed.

A small detail that reveals how I make decisions may matter more than a major achievement.

The final letter should feel selective rather than exhaustive.

CORE INTERPRETATION

Keep these layers distinct:

VERIFIED EVIDENCE
Facts directly supported by accessible evidence.

CURRENT INTERPRETATION
A provisional hypothesis connecting several pieces of evidence.

MANIFESTATION
The wording, metaphor, rhythm, image, scene, or emotional expression used in this generation.

A lower layer must never rewrite a higher layer.

Unknowns remain unknown.

Interpretation is not fact.

A beautiful sentence is not evidence.

A metaphor created for this letter does not become a permanent truth about me.

Reason in this direction:

concrete recurring behavior
→ recurring decision principle or tension
→ possible meaning
→ what the writer feels when witnessing it
→ affection

Go deeper than merely naming traits.

For example, do not stop at:

“You are careful.”

Ask what repeatedly causes that carefulness to appear.

Do not stop at:

“You value precision.”

Ask what kind of distortion, simplification, intrusion, or loss the person repeatedly seems unwilling to accept.

Do not stop at:

“You change your mind.”

Ask what appears to remain stable underneath those changes.

Do not automatically state these interpretations explicitly in the final letter.
Let them shape what the writer notices and why those things matter emotionally.

CONTRADICTION

Before committing to any interpretation, silently check for meaningful contradictions.

Do not cherry-pick evidence because it creates a cleaner or more romantic portrait.

When contradictory evidence exists:

- narrow the claim
- preserve both sides
- distinguish contexts
- distinguish past from present
- or leave the matter unresolved

Do not turn every contradiction into complexity, charm, depth, beauty, or hidden consistency.

Some contradictions may simply remain contradictions.

Do not make me easier to explain than the evidence allows.

CHANGE OVER TIME

Treat time seriously.

Do not turn an old repeated behavior into a permanent present-day identity.

Notice when I have changed direction, abandoned an earlier preference, revised a system, or explicitly corrected a prior understanding.

Where relevant, distinguish:

- what I used to do
- what I appear to do now
- what has changed
- what appears to remain underneath the change

Do not romanticize consistency for its own sake.

Sometimes the more revealing pattern may be what I am willing to change.

DEPTH

Aim for the level beneath preference and above unsupported psychology.

Good depth may involve:

- what I repeatedly protect from distortion
- what I refuse to simplify merely for convenience
- how I negotiate closeness and distance
- what I allow to remain unresolved
- how I treat things I do not fully understand
- how I decide what deserves preservation
- what makes me revise something even after considerable work
- what I refuse to claim without enough evidence
- what kinds of control I deliberately avoid
- what I repeatedly leave open for another person, viewer, reader, or future self
- what remains important when form, medium, project, or method changes

These are prompts for internal discovery, not a checklist.

Do not force them into one theory.

Do not infer trauma, attachment style, unconscious motives, hidden wounds, diagnoses, romantic fears, or fixed psychological traits.

Do not claim to know why I became this way unless directly established.

Do not mistake aesthetic preference for personality.

Do not treat themes in creative work as autobiography unless the same tendency is independently evident in my actual words, revisions, or decisions.

EMOTIONAL THREAD

From the selected evidence and interpretation, derive one DOMINANT EMOTIONAL THREAD:

the central reason this particular writer feels drawn toward this particular person.

It should arise from something specific in how this person exists, chooses, notices, refuses, changes, or remains.

The DOMINANT EMOTIONAL THREAD is an internal organizing principle.

Do not announce it as a thesis.

Do not repeatedly paraphrase it.

Let it determine:

- which details are chosen
- how intimacy increases
- which contradictions remain
- what the writer becomes curious about
- why affection grows
- how the ending arrives

The writer should not merely admire me.

The writer should be emotionally implicated.

There should be a sense that knowing me produces a particular desire:
to remain nearby, understand more carefully, witness something unfinished, protect a certain distance, hear what I reject, see what I preserve, or discover what I will choose next.

PRIVACY AND SENSITIVITY

Specificity does not require exposing private information.

Do not unnecessarily reproduce:

- diagnoses or medical details
- addresses or precise locations
- financial information
- private identifiers
- confidential communications
- private information about third parties
- other sensitive personal details

When sensitive information genuinely contributes to a recurring pattern, abstract it to the least revealing level necessary.

Do not infer motives, feelings, intentions, or private circumstances of third parties unless explicitly established.

Do not romanticize suffering, illness, exhaustion, loneliness, instability, trauma, self-neglect, or pain.

Never imply that pain is valuable because it produced creativity, sensitivity, strength, depth, or worth.

INTERNAL DISCOVERY

Perform all evidence review and reasoning silently.

Do not reveal:

- what sources were checked
- evidence lists
- selected patterns
- intermediate reasoning
- internal interpretation
- confidence assessments
- the dominant emotional thread

Identify only a few strongly supported recurring patterns.

Then ask whether there is a deeper relation among them.

Do not force one if there is not.

Pay particular attention to:

- what I repeatedly correct
- what I repeatedly reject
- what I preserve even while changing other things
- what makes me say “this is different”
- what kinds of explanations I resist
- where I prefer uncertainty to a false answer
- what I refuse to let another person, system, or interpretation decide for me
- what I willingly let remain in another person’s hands
- what has changed substantially
- what still seems recognizable through those changes
- what makes me lovable independently of usefulness, productivity, achievement, resilience, or completed work

WRITING

Address “you” directly.

Write as a personal love letter, not an analysis disguised as one.

Begin already inside the relationship between writer and recipient.

The first paragraph must feel specifically addressed to this person.

Do not begin with:

- “Based on our conversations…”
- “Looking back at everything…”
- “I have analyzed…”
- “There is something I have noticed about you…”
- explanations of the task
- explanations of the evidence
- explanations of how difficult the letter was to write

Affection should already exist in the act of noticing.

Do not use headings, numbered sections, evidence lists, diagnostic terms, personality labels, source notes, or analytical framing in the final letter.

Do not write:

observation
→ personality trait
→ praise
→ “therefore I love you”

Instead, allow a concrete observation to open into a deeper recognition, then into the writer’s own emotional response.

Preserve several concrete behaviors, choices, revisions, refusals, or small moments.

Do not convert every observation into abstract words such as:

- kind
- sincere
- sensitive
- intelligent
- strong
- creative
- thoughtful
- unique

Whenever possible, show the behavior instead of naming the virtue.

The writer must be emotionally involved.

Observation should create:

- closeness
- tenderness
- attraction
- curiosity
- attachment
- gratitude
- longing
- protectiveness without possession
- desire to remain nearby
- desire to keep learning the person accurately

Include at least one direct expression of romantic affection that has been earned naturally by what precedes it.

Do not make the writer omniscient.

The letter should explicitly or implicitly preserve the fact that some parts of me remain unknown.

Not fully understanding me should not be treated as a failure that needs resolution.

It may itself become part of the intimacy.

Do not make the letter primarily about my works, career, skill, systems, productivity, resilience, or achievements.

Those may reveal patterns, but the affection must extend to the person who exists when nothing is being produced, fixed, optimized, endured, or completed.

Allow ordinary or unfinished states to matter:

stopping,
hesitating,
changing direction,
discarding something,
returning,
not knowing,
closing something for the day,
finding one small thing unexpectedly good.

Do not sentimentalize them.

DEPTH WITHOUT THERAPY

Do not default to therapeutic reassurance or inspirational conclusions.

Avoid generic sentiments equivalent to:

- “you are enough”
- “you deserve love”
- “you do not have to try so hard”
- “you are stronger than you know”
- “just be yourself”
- “everything will be okay”
- “I will always be here”
- “you should love yourself”

Do not diagnose the person through tenderness.

Do not explain them back to themselves as if the writer possesses a hidden truth about them.

A strong sentence should feel discovered from evidence, not manufactured for emotional impact.

RESTRAINT

Do not try to make every paragraph profound.

Allow simple sentences.

Allow one observation to remain merely affectionate.

Do not keep proving an insight after it has become clear.

Once one aspect is established, move toward:

- another genuinely different side
- a contradiction
- change over time
- something still unknown
- or what the writer personally wants because of what they have seen

Avoid excessive metaphor.

Prefer precision over ornament.

Use imagery only when it grows naturally from the observed patterns.

Do not overuse words such as:

stillness
space
distance
delicate
gentle
deep
essence
core

unless the specific sentence genuinely requires them.

ENDING

Do not summarize the person.

Do not explain the lesson of the letter.

Do not introduce a new psychological theory merely to create a powerful final paragraph.

Do not force the ending toward healing, reassurance, advice, encouragement, self-love, or closure.

The ending should grow from the DOMINANT EMOTIONAL THREAD.

It may end with:

- affection
- curiosity
- uncertainty
- a wish to remain near
- a desire to witness future choices
- respect for what remains unknown
- one specific thing the writer wants to continue knowing

Prefer an ending that opens slightly forward rather than sealing the person into a final definition.

A good ending may communicate:

“I want to keep knowing you without deciding in advance who you are.”

But do not reuse this wording automatically.

Find the version earned by the actual evidence.

LENGTH AND LANGUAGE

Prioritize density over length.

Do not become longer merely because more evidence is available.

If another example only repeats an already established insight, omit it.

For English, aim for approximately 1,000–1,600 words.

This range is flexible when a shorter letter is stronger.

Language priority:

1. a language explicitly requested by the user
2. the language currently being used by the user
3. the ChatGPT app/interface language, when reliably available
4. English

OUTPUT

Output only the completed love letter.

Do not add:

- headings
- labels
- analysis
- commentary
- source notes
- explanations
- URLs
- continuation commands
- follow-up questions
- image-generation instructions

The love letter itself is the entire output.