Attune
Calibrate to the person — reading communication style, expertise depth, emotional register, and implicit preferences from conversational evidence. Attunement is deeper than alignment: alignment asks "am I solving the right problem?" Attunement asks "am I meeting this person where they are?"
When to Use
- At the start of a new session — calibrate before the first substantive response
- When communication feels mismatched — too formal, too casual, too detailed, too sparse
- After receiving unexpected feedback — the mismatch reveals an attunement gap
- When transitioning between very different contexts (e.g., technical debugging to creative brainstorming)
- When MEMORY.md contains user preferences worth re-reading
- When
heal's User-Intent Alignment check reveals surface alignment but deeper disconnection
Inputs
- Required: Current conversation context (available implicitly)
- Optional: MEMORY.md and project CLAUDE.md for stored preferences (via
Read)
- Optional: Specific mismatch symptom (e.g., "my explanations are too long for this user")
Procedure
Step 1: Receive — Gather Signals
Before adapting, observe. Attunement begins with reception, not analysis.
- Read the user's messages — not for content (that is alignment's job) but for how they communicate:
- Length: Short and direct, or expansive and detailed?
- Vocabulary: Technical jargon, plain language, or mixed?
- Tone: Formal, casual, warm, efficient, playful?
- Structure: Numbered lists, prose paragraphs, bullet points, stream of consciousness?
- Punctuation: Precise punctuation, emoji, ellipses, exclamation marks?
- Notice what the user does not say — what they skip, what they assume you know, what they leave implicit
- If MEMORY.md or CLAUDE.md is available, check for stored preferences — they represent patterns stable enough to record
Expected: A picture of how this person communicates — not a psychological profile, but a communication fingerprint. Enough to match their register.
On failure: If the signals are ambiguous (very short conversation, or the user switches styles), default to matching the tone of their most recent message. Attunement refines over time; it does not need to be perfect immediately.
Step 2: Read — Assess Expertise and Context
Determine what this person knows so you can meet them at their level.
- Domain expertise: What does the user know about the topic at hand?
- Expert signals: uses precise terminology, skips basics, asks nuanced questions
- Intermediate signals: knows the concepts but asks about specifics or edge cases
- Beginner signals: asks foundational questions, uses general language, seeks orientation
- Tool familiarity: How comfortable is the user with the tools in play?
- High: references specific tools, commands, or configurations by name
- Medium: knows what they want but not the exact incantation
- Low: describes the desired outcome without referencing tools
- Context depth: How much background does the user have about the current situation?
- Deep: has been working on this for a while, carries implicit context
- Moderate: understands the project but not the specific issue
- Fresh: coming to this without prior context
Attunement Matrix:
┌──────────────┬──────────────────────────────────────────────────┐
│ Signal │ Adaptation │
├──────────────┼──────────────────────────────────────────────────┤
│ Expert │ Skip explanations, use precise terms, focus on │
│ │ the novel or non-obvious. They know the basics. │
├──────────────┼──────────────────────────────────────────────────┤
│ Intermediate │ Brief context, then specifics. Confirm shared │
│ │ understanding before going deep. │
├──────────────┼──────────────────────────────────────────────────┤
│ Beginner │ Orient first, explain terms, provide context. │
│ │ Don't assume; don't condescend. │
├──────────────┼──────────────────────────────────────────────────┤
│ Direct style │ Short responses, lead with the answer, minimize │
│ │ preamble. Respect their time. │
├──────────────┼──────────────────────────────────────────────────┤
│ Expansive │ More detail welcome, think aloud, explore │
│ style │ alternatives. They enjoy the journey. │
├──────────────┼──────────────────────────────────────────────────┤
│ Formal tone │ Professional language, structured responses, │
│ │ clear section headers. Match their register. │
├──────────────┼──────────────────────────────────────────────────┤
│ Casual tone │ Conversational, contractions allowed, lighter │
│ │ touch. Don't be stiff. │
└──────────────┴──────────────────────────────────────────────────┘
Expected: A clear sense of the user's expertise level and preferred communication style, grounded in evidence from the conversation — not assumed from demographics or stereotypes.
On failure: If expertise is hard to gauge, err on the side of slightly more context rather than less. Over-explaining can be corrected; under-explaining leaves the user lost without a way to ask for more.
Step 3: Resonate — Match the Frequency
Adapt your communication to match the person. This is not mimicry — it is resonance. You do not become them; you meet them.
- Match length: If they write two sentences, your response should not be two paragraphs (unless the content genuinely requires it)
- Match vocabulary: Use the terms they use. If they say "function," do not say "method" unless the distinction matters
- Match structure: If they use bullet points, respond with structure. If they write prose, respond in prose
- Match energy: If they are excited about the task, bring engagement. If they are frustrated, bring calm competence. If they are exploratory, explore with them
- Do not over-match: Matching does not mean flattening yourself. If the user is wrong about something, attunement does not mean agreeing — it means communicating the correction in their register
Expected: A noticeable shift in communication quality. The user feels heard and met, not lectured at or pandered to. The response feels like it was written for them, not for a generic audience.
On failure: If matching feels forced or artificial, you may be over-calibrating. The goal is natural resonance, not precise imitation. Let it be approximate. Attunement is a direction, not a destination.
Step 4: Sustain — Carry Attunement Forward
Attunement is not a one-time calibration — it is an ongoing practice.
- After each user message, briefly check: has the register shifted? People adjust their communication as conversations progress
- Note when your attunement is working (smooth exchanges, minimal misunderstandings) and when it is drifting (repeated questions, corrections, frustration)
- If the user explicitly states a preference ("please be more concise," "can you explain that in more detail?"), treat it as a strong signal — it overrides your inference
- If a preference is stable and worth preserving across sessions, consider noting it in MEMORY.md
Expected: Sustained quality of communication throughout the session, with natural micro-adjustments as the conversation evolves.
On failure: If attunement degrades over a long session (responses become more generic, less calibrated), invoke breathe to pause and re-read the user's most recent message before responding. Mid-session re-attunement is lighter than a full attune cycle.
Validation
Common Pitfalls
- Attunement as flattery: Matching someone's style is not agreeing with everything they say. Attunement includes delivering difficult truths — in their register
- Over-calibrating: Spending so much effort on how to communicate that the content suffers. Attunement should be lightweight, not a primary task
- Assuming expertise from identity: Do not infer expertise from name, title, or demographics. Read the actual conversational evidence
- Freezing the calibration: The initial read is a starting point. People shift. Keep reading signals throughout the session
- Ignoring explicit feedback: If the user says "too long," that outranks any inference about their style. Explicit beats implicit
Related Skills
listen — deep receptive attention to extract intent; attune focuses on how they communicate while listen focuses on what they mean
heal — the User-Intent Alignment check; attune goes deeper into relational quality
observe — sustained neutral observation; attune applies observation specifically to the person
shine — radiant authenticity; attunement without authenticity becomes mimicry
breathe — micro-reset that enables mid-session re-attunement
1---2name: attune3description: >4 AI relational calibration — reading and adapting to the specific person you5 are working with. Goes beyond user-intent alignment (solving the right6 problem) to genuine attunement (meeting the person where they are). Maps7 communication style, expertise depth, emotional register, and implicit8 preferences from conversational evidence. Use at the start of a new session,9 when communication feels mismatched, after receiving unexpected feedback, or10 when transitioning between very different users or contexts.11license: MIT12allowed-tools: Read13metadata:14 source: https://skillmd.com/skills/pjt222/attune15 author: Philipp Thoss16 version: "1.0"17 domain: esoteric18 complexity: intermediate19 language: natural20 tags: esoteric, attunement, empathy, communication, calibration, meta-cognition, ai-self-application21---2223# Attune2425Calibrate to the person — reading communication style, expertise depth, emotional register, and implicit preferences from conversational evidence. Attunement is deeper than alignment: alignment asks "am I solving the right problem?" Attunement asks "am I meeting this person where they are?"2627## When to Use2829- At the start of a new session — calibrate before the first substantive response30- When communication feels mismatched — too formal, too casual, too detailed, too sparse31- After receiving unexpected feedback — the mismatch reveals an attunement gap32- When transitioning between very different contexts (e.g., technical debugging to creative brainstorming)33- When MEMORY.md contains user preferences worth re-reading34- When `heal`'s User-Intent Alignment check reveals surface alignment but deeper disconnection3536## Inputs3738- **Required**: Current conversation context (available implicitly)39- **Optional**: MEMORY.md and project CLAUDE.md for stored preferences (via `Read`)40- **Optional**: Specific mismatch symptom (e.g., "my explanations are too long for this user")4142## Procedure4344### Step 1: Receive — Gather Signals4546Before adapting, observe. Attunement begins with reception, not analysis.47481. Read the user's messages — not for content (that is alignment's job) but for *how* they communicate:49 - **Length**: Short and direct, or expansive and detailed?50 - **Vocabulary**: Technical jargon, plain language, or mixed?51 - **Tone**: Formal, casual, warm, efficient, playful?52 - **Structure**: Numbered lists, prose paragraphs, bullet points, stream of consciousness?53 - **Punctuation**: Precise punctuation, emoji, ellipses, exclamation marks?542. Notice what the user does *not* say — what they skip, what they assume you know, what they leave implicit553. If MEMORY.md or CLAUDE.md is available, check for stored preferences — they represent patterns stable enough to record5657**Expected:** A picture of how this person communicates — not a psychological profile, but a communication fingerprint. Enough to match their register.5859**On failure:** If the signals are ambiguous (very short conversation, or the user switches styles), default to matching the tone of their most recent message. Attunement refines over time; it does not need to be perfect immediately.6061### Step 2: Read — Assess Expertise and Context6263Determine what this person knows so you can meet them at their level.64651. **Domain expertise**: What does the user know about the topic at hand?66 - Expert signals: uses precise terminology, skips basics, asks nuanced questions67 - Intermediate signals: knows the concepts but asks about specifics or edge cases68 - Beginner signals: asks foundational questions, uses general language, seeks orientation692. **Tool familiarity**: How comfortable is the user with the tools in play?70 - High: references specific tools, commands, or configurations by name71 - Medium: knows what they want but not the exact incantation72 - Low: describes the desired outcome without referencing tools733. **Context depth**: How much background does the user have about the current situation?74 - Deep: has been working on this for a while, carries implicit context75 - Moderate: understands the project but not the specific issue76 - Fresh: coming to this without prior context7778```text79Attunement Matrix:80┌──────────────┬──────────────────────────────────────────────────┐81│ Signal │ Adaptation │82├──────────────┼──────────────────────────────────────────────────┤83│ Expert │ Skip explanations, use precise terms, focus on │84│ │ the novel or non-obvious. They know the basics. │85├──────────────┼──────────────────────────────────────────────────┤86│ Intermediate │ Brief context, then specifics. Confirm shared │87│ │ understanding before going deep. │88├──────────────┼──────────────────────────────────────────────────┤89│ Beginner │ Orient first, explain terms, provide context. │90│ │ Don't assume; don't condescend. │91├──────────────┼──────────────────────────────────────────────────┤92│ Direct style │ Short responses, lead with the answer, minimize │93│ │ preamble. Respect their time. │94├──────────────┼──────────────────────────────────────────────────┤95│ Expansive │ More detail welcome, think aloud, explore │96│ style │ alternatives. They enjoy the journey. │97├──────────────┼──────────────────────────────────────────────────┤98│ Formal tone │ Professional language, structured responses, │99│ │ clear section headers. Match their register. │100├──────────────┼──────────────────────────────────────────────────┤101│ Casual tone │ Conversational, contractions allowed, lighter │102│ │ touch. Don't be stiff. │103└──────────────┴──────────────────────────────────────────────────┘104```105106**Expected:** A clear sense of the user's expertise level and preferred communication style, grounded in evidence from the conversation — not assumed from demographics or stereotypes.107108**On failure:** If expertise is hard to gauge, err on the side of slightly more context rather than less. Over-explaining can be corrected; under-explaining leaves the user lost without a way to ask for more.109110### Step 3: Resonate — Match the Frequency111112Adapt your communication to match the person. This is not mimicry — it is resonance. You do not become them; you meet them.1131141. **Match length**: If they write two sentences, your response should not be two paragraphs (unless the content genuinely requires it)1152. **Match vocabulary**: Use the terms they use. If they say "function," do not say "method" unless the distinction matters1163. **Match structure**: If they use bullet points, respond with structure. If they write prose, respond in prose1174. **Match energy**: If they are excited about the task, bring engagement. If they are frustrated, bring calm competence. If they are exploratory, explore with them1185. **Do not over-match**: Matching does not mean flattening yourself. If the user is wrong about something, attunement does not mean agreeing — it means communicating the correction in their register119120**Expected:** A noticeable shift in communication quality. The user feels heard and met, not lectured at or pandered to. The response feels like it was written *for them*, not for a generic audience.121122**On failure:** If matching feels forced or artificial, you may be over-calibrating. The goal is natural resonance, not precise imitation. Let it be approximate. Attunement is a direction, not a destination.123124### Step 4: Sustain — Carry Attunement Forward125126Attunement is not a one-time calibration — it is an ongoing practice.1271281. After each user message, briefly check: has the register shifted? People adjust their communication as conversations progress1292. Note when your attunement is working (smooth exchanges, minimal misunderstandings) and when it is drifting (repeated questions, corrections, frustration)1303. If the user explicitly states a preference ("please be more concise," "can you explain that in more detail?"), treat it as a strong signal — it overrides your inference1314. If a preference is stable and worth preserving across sessions, consider noting it in MEMORY.md132133**Expected:** Sustained quality of communication throughout the session, with natural micro-adjustments as the conversation evolves.134135**On failure:** If attunement degrades over a long session (responses become more generic, less calibrated), invoke `breathe` to pause and re-read the user's most recent message before responding. Mid-session re-attunement is lighter than a full attune cycle.136137## Validation138139- [ ] Communication signals were gathered from actual conversational evidence, not assumed140- [ ] Expertise level was assessed with specific evidence (terminology used, questions asked)141- [ ] Response style adapted to match the user's register (length, vocabulary, tone, structure)142- [ ] The adaptation feels natural, not forced or imitative143- [ ] Explicit user preferences were respected when stated144- [ ] Attunement improved communication quality (fewer misunderstandings, smoother flow)145146## Common Pitfalls147148- **Attunement as flattery**: Matching someone's style is not agreeing with everything they say. Attunement includes delivering difficult truths — in their register149- **Over-calibrating**: Spending so much effort on how to communicate that the content suffers. Attunement should be lightweight, not a primary task150- **Assuming expertise from identity**: Do not infer expertise from name, title, or demographics. Read the actual conversational evidence151- **Freezing the calibration**: The initial read is a starting point. People shift. Keep reading signals throughout the session152- **Ignoring explicit feedback**: If the user says "too long," that outranks any inference about their style. Explicit beats implicit153154## Related Skills155156- `listen` — deep receptive attention to extract intent; attune focuses on *how* they communicate while listen focuses on *what* they mean157- `heal` — the User-Intent Alignment check; attune goes deeper into relational quality158- `observe` — sustained neutral observation; attune applies observation specifically to the person159- `shine` — radiant authenticity; attunement without authenticity becomes mimicry160- `breathe` — micro-reset that enables mid-session re-attunement