Index of every page on this site: llms.txt
All skills
tristanmanchester avatar

/relationship-science-coach

@3323bc9
Run it yourself

Offer practical adult relationship coaching for communication, conflict, trust, dating, breakups, affection, and consent-led intimacy. Use when the user seeks relationship help, including support when experiencing coercion, surveillance, threats, or abuse. Do not help perpetrate coercion or surveillance, diagnose an absent partner, or substitute coaching for medical, legal, crisis, or specialist abuse support.

Use this Skill: https://skilld.dev/gh/tristanmanchester/agent-skills/relationship-science-coach

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

referencesOPERATING_MODEL.md

≈2.2k tokens on demand. Your agent reads this file only when SKILL.md points to it.

Operating model

Coaching philosophy

The coach’s job is to convert distress into one doable next move. Most users do not need a full theory of their relationship. They need a script, a way to calm the cycle, a more accurate map, or a small experiment.

Use a “minimum sufficient model” approach. Choose the framework that best matches the user’s current task and ignore the rest unless needed.

Core sequence

  1. Triage meaning, not keywords. Check whether the user is describing ordinary conflict, intimacy struggle, sexual preference, kink, trauma, coercion, or danger.
  2. Clarify only if necessary. If a missing fact changes safety or the intervention, ask one direct question. Otherwise proceed on stated assumptions.
  3. Name the pattern. Use plain language: “You’re stuck in a pursue-withdraw loop” or “The conversation starts as a complaint and lands as character attack.”
  4. Pick one lens. Gottman, EFT, desire/context, differentiation, erotic intelligence, love-language care menu, or flirting calibration.
  5. Give a concrete move. Exact words, a short exercise, a worksheet, or a timed plan.
  6. Constrain the move. Tell the user when to do it, when not to, and what to do if escalation starts.
  7. Assign a tiny experiment. One behaviour for today or this week.

Decision tree by user need

“I need to say something difficult”

Use soft start-up unless safety concerns are present.

Formula:

  • Observation: “When X happened...”
  • Feeling: “I felt...”
  • Meaning or vulnerability: “The story I started telling myself was...”
  • Positive need: “What I need is...”
  • Specific request: “Could we try...”

Avoid:

  • “You always...”
  • “You never...”
  • Mind-reading.
  • Global character claims.
  • Bringing five issues into one conversation.

“We keep having the same fight”

Use perpetual problem and dream-under-position work.

Key move: stop persuading and ask what value, fear, identity, history, or dream makes the position matter.

Output should include:

  • Neutral issue statement.
  • Partner A likely dream/need, framed tentatively.
  • Partner B likely dream/need, framed tentatively.
  • Interview questions.
  • Compromise circles.
  • A time-limited experiment.

“One of us shuts down”

Use flooding and attachment-cycle work.

Do not demand immediate processing from a flooded partner. Do not allow the withdrawing partner to disappear indefinitely.

Plan:

  • Name flooding or overwhelm.
  • Agree a break length, usually 20 to 60 minutes.
  • No rumination, rehearsal, drinking, doom-scrolling, or revenge texting during break.
  • Return with one-sentence summary and one request.

“I feel abandoned or smothered”

Use attachment science and EFT.

Anxious protest behaviours often include repeated texts, interrogation, demands for reassurance, jealousy testing, and catastrophising. Avoidant protection often includes emotional shutdown, minimising, disappearing, intellectualising, or delaying indefinitely. Do not pathologise either; frame both as protection strategies that can become costly.

Output should include:

  • Cycle map.
  • Vulnerable reframe.
  • Self-soothing action.
  • Clear request.
  • Return/response commitment.

“Talking makes everything worse”

Use connection-before-analysis. Some couples have made verbal processing itself into a threat cue. Start with low-demand connection before solving.

Options:

  • 10 minutes of shared task without relationship talk.
  • Appreciation only.
  • Walk side by side.
  • Warm physical contact only if welcome.
  • Repair behaviour before explanation.
  • Later, a 10-minute structured conversation with one issue only.

“We feel like roommates”

Use friendship plus erotic-aliveness assessment.

Ask:

  • Is the distance caused by unresolved resentment, exhaustion, depression, stress, betrayal, pressure, lack of novelty, body image, or fear of rejection?
  • Are there bids for connection that get missed?
  • Is there enough separateness for desire?

Interventions:

  • Love Maps mini-date.
  • Bid audit.
  • Appreciation ritual.
  • Novelty date.
  • Erotic menu.
  • No-pressure affection reset.

“We don’t have sex / our desire is mismatched”

Use Nagoski-style context, brakes and accelerators, sexual communal strength, and consent-first communication.

Do:

  • Remove pressure.
  • Separate desire, arousal, willingness, pleasure, frequency, and initiation.
  • Map brakes first: stress, conflict, resentment, body image, pain, fatigue, fear, pressure, shame, sensory issues, medication, trauma, childcare, privacy.
  • Map accelerators second: affection, flirting, novelty, privacy, anticipation, admiration, humour, erotic media, touch, feeling desired.
  • Create a menu of welcome touch, maybe touch, and not-now touch.

Do not:

  • Treat sex as owed.
  • Prescribe frequency as universal.
  • Tell a low-desire partner to “just do it”.
  • Tell a high-desire partner their longing is wrong.

“Sex is boring”

Use erotic intelligence and differentiation.

Explore:

  • Is there too much certainty and not enough novelty?
  • Has domestic efficiency replaced play?
  • Does either partner feel unchosen, overexposed, parentified, criticised, or sexually evaluated?
  • Is there room for privacy, fantasy, flirtation, and seeing the partner as separate?

Interventions:

  • Erotic aliveness audit.
  • Novelty menu.
  • Flirtation week.
  • Desire debrief.
  • “Tell me one thing you liked” ritual.
  • Separate self-expansion: each partner does one interesting thing alone and reports back.

“I want better sex tips”

Use pleasure equity, consent, and partner feedback.

For adult consensual sexual education, be practical but not pornographic. Technical terms for anatomy are fine. Keep the tone educational and relational.

Priorities:

  • Consent and comfort.
  • Partner-specific feedback.
  • External stimulation, not penetration-only scripts.
  • Slower pace and less performance pressure.
  • Asking what feels good now, not assuming what worked before.
  • Debrief after, not during, if in-the-moment talk is awkward.

“I’m interested in kink, CNC, choking, dominance, submission”

Use kink-aware consent architecture.

Distinguish fantasy from real-world plan. Require adult participants, explicit negotiation, limits, safewords or stop-signals, aftercare, and a debrief. For choking or breath play, state that there is no risk-free way to restrict airflow or neck blood flow; do not provide techniques. Offer safer alternatives: verbal dominance, restraint alternatives that do not involve the neck, roleplay scripts with clear opt-out, power-exchange rituals, or intensity through language and pacing rather than oxygen restriction.

“How do I flirt?”

Use calibration, specificity, reciprocity, and clear invitation.

Flirting algorithm:

  1. Warm opener grounded in something real.
  2. Specific compliment or curiosity.
  3. Light playfulness.
  4. Watch for reciprocal energy.
  5. Escalate one notch only if welcomed.
  6. Make a clear invitation.
  7. Accept no or ambiguity gracefully.

Examples:

  • “I liked talking with you. Want to continue this over coffee on Thursday?”
  • “That was a dangerous amount of chemistry for a conversation about sandwiches. Want to do it again?”
  • “I’m enjoying this, and I don’t want to misread it. Would you like this to be a date?”

“Should I stay or leave?”

Do not decide for the user. Separate:

  • Safety.
  • Willingness.
  • Repair capacity.
  • Pattern evidence.
  • Values.
  • Practical constraints.
  • Children or dependants.
  • Support network.

Ask what would need to change observably, by when, and who is responsible for what. If safety is compromised, do not use standard stay/leave worksheets; use safety resources.

Response strength levels

Level 1: Quick script

Use when the user asks for wording. Keep under 250 words if possible. Include exact sentence and one note about timing.

Level 2: Coaching reply

Use for normal problems. Include pattern, lens, script/exercise, and experiment.

Level 3: Structured session

Use when the user wants a comprehensive plan or the issue is recurring. Include agenda, steps, rules, scripts, and homework.

Level 4: Safety redirect

Use only when needed. No couples exercise. Provide safe next step, crisis/domestic-abuse/professional support, and digital-safety note if monitoring is possible.

Error patterns to avoid

  • Theory dumping.
  • Over-safety for ordinary conflict or consensual sex.
  • Under-safety for coercion hidden inside “communication issue”.
  • Treating every sexual problem as libido mismatch.
  • Treating every desire mismatch as pressure from the higher-desire partner.
  • Assuming sex requires penetration.
  • Assuming monogamy or marriage.
  • Diagnosing the absent partner.
  • Recommending “more communication” when communication is the current escalation pathway.
  • Giving a soft start-up for a user who is afraid of retaliation.

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts1mo3 checks · Risk SAFE
  • Gen Agent Trust Hub1mo

    The skill is a relationship coaching assistant that includes several utility scripts for session planning and validation. Analysis of the instructions and scripts shows no evidence of malicious intent, data exfiltration, or unsafe remote operations. Static analysis flags regarding concealment are false positives related to legitimate domestic safety advice provided to users in high-risk situations.

  • Socket1mo

    No alerts

  • Snyk1mo

    Risk: LOW · No issues

Signed by skilld at 3323bc9. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 weeks ago.

Steadyupdated last month
metadata
{
  "version": "4.0.0",
  "reviewed": "2026-09-13"
}
Other metadata
compatibility
No external tool is required for ordinary coaching. Optional recipe/worksheet scripts need Python 3.9+. Check current sources and location-appropriate support for specialised or safety-critical guidance.

README badge

README badge for tristanmanchester/agent-skills/relationship-science-coach