Purpose

Analyze communication patterns for $ARGUMENTS.

This skill works best with Harmony's conversation data — transcripts, recordings, insights, and scorecards from real meetings. The more meetings analyzed, the stronger the pattern detection.

If the user specifies a person, time range, or meeting type, scope the analysis accordingly. If they ask about a specific behavior (e.g., "when do I avoid conflict?"), focus the analysis there. If no specific focus is given, provide a comprehensive communication profile.

Clarify Analysis Goals

If not specified, ask what they want to learn:

• Specific behaviors (conflict avoidance, interruptions, filler words)
• Communication effectiveness (clarity, directness, listening)
• Meeting facilitation skills (inclusion, time management, decision-making)
• Speaking patterns and ratios
• Growth areas for improvement
• Comparison across time periods

Analysis Dimensions

Analyze whichever dimensions are requested (or all, for a comprehensive profile):

1. Conflict Avoidance & Directness

• Hedging language ("maybe," "kind of," "I think," "sort of")
• Indirect phrasing instead of direct requests or feedback
• Changing subject when tension arises
• Agreeing without commitment ("yeah, but...")
• Not addressing obvious problems or elephant-in-the-room topics
• Deflecting accountability ("whatever you think is best")

2. Speaking Patterns

• Speaking time as percentage of meeting
• Interruptions given and received
• Average speaking turn length
• Question-to-statement ratio
• Monologue detection (unbroken speaking >2 minutes)

3. Filler Words & Verbal Habits

• Count of "um," "uh," "like," "you know," "actually," "basically," "literally"
• Frequency per minute or per speaking turn
• Situations where they increase (nervousness, uncertainty, complex topics)
• Hedge words that weaken statements ("just," "I feel like," "I'm not sure but")

4. Active Listening

• Questions that reference others' previous points
• Paraphrasing or summarizing others' ideas
• Building on others' contributions vs. redirecting to own agenda
• Asking clarifying questions vs. assuming
• Acknowledging contributions before moving on

5. Leadership & Facilitation

• Decision-making approach (directive vs. collaborative vs. avoidant)
• How disagreements are handled
• Inclusion of quieter participants
• Time management and agenda control
• Follow-up and action item clarity
• Tone-setting at meeting open and close

6. Sentiment & Tone

• Emotional consistency vs. volatility
• Positivity/negativity balance
• Energy level across different meeting types
• Warmth vs. distance with different audiences

Evidence Standard

Every finding must be grounded in specific, cited conversation moments:

### [Pattern Name]

**Finding**: [One-sentence summary]
**Frequency**: [X times across Y meetings]

**Example 1: [Meeting Name/Date]** — [Timestamp]

> [Actual quote from transcript]

**Why This Matters**:
[Impact on the conversation, the team, or the outcome]

**Better Approach**:
[Specific alternative phrasing or behavior — not generic advice]

Provide 2-3 strongest examples per pattern. More examples available on request.

Output Format

Analysis Summary

Analysis Period: [Date range] Meetings Analyzed: [X meetings] Total Duration: [X hours] Focus: [What was analyzed — comprehensive or specific behaviors]

Key Patterns Identified

For each pattern (3-5 patterns):

[Pattern Name]

• Observed: What was found, with frequency
• Impact: Why it matters for communication effectiveness
• Examples: 2-3 timestamped, cited examples with actual quotes
• Better Approach: Specific, actionable alternative for each example
• Recommendation: One concrete habit change to practice

Communication Strengths

1. [Strength 1 with specific example from a conversation]
2. [Strength 2 with specific example]
3. [Strength 3 with specific example]

Speaking Statistics

Growth Opportunities

Ranked by impact:

1. [Area 1]: [Specific, actionable advice with a practice exercise]
2. [Area 2]: [Specific, actionable advice with a practice exercise]
3. [Area 3]: [Specific, actionable advice with a practice exercise]

Trend Analysis (if multiple time periods)

Recommended Next Steps

1. [Concrete action to practice this week]
2. [Habit to build this month]
3. [Metric to track going forward]
4. [Suggested follow-up analysis in 30 days]

Example

Input: "Tell me all the times I avoided conflict in my meetings this month"

Output:

Analysis Summary

Analysis Period: February 1-28, 2026 Meetings Analyzed: 18 meetings Total Duration: 14.5 hours Focus: Conflict avoidance and indirect communication

Key Pattern: Hedging on Critical Feedback

Finding: You soften or avoid direct feedback 8 times across 7 meetings, particularly in 1:1s where performance issues need addressing.

Frequency: 8 instances in 18 meetings (44% of meetings contain at least one avoidance moment)

Example 1: 1:1 with Sarah — 00:14:32

"So, I was thinking... maybe we could, like, potentially consider looking at the timeline again? I mean, if you think that makes sense. But whatever you think is best!"

Why This Matters: Sarah's project was two weeks behind schedule. The hedging language ("maybe," "potentially," "if you think") and the deflection ("whatever you think") made it easy for her to miss the urgency entirely. She left the meeting without changing anything.

Better Approach: "Sarah, the project is two weeks behind the timeline we agreed on. I want to understand what's blocking progress and create a new timeline together today."

Example 2: Team standup — 00:08:15

"So, there's been some... feedback... about the deployment process. Nothing major! Just, you know, maybe something to think about when you get a chance."

Why This Matters: Three team members had escalated deployment failures that week. Framing it as "nothing major" and "something to think about" communicated that the issue was optional to address.

Better Approach: "We've had three deployment failures this week that impacted customers. I need us to do a root-cause analysis by Friday. Who can own that?"

Example 3: Product review — 00:32:10

"I mean, this design is... interesting. I have some thoughts but I don't want to slow things down. Let's just go with it and see what happens."

Why This Matters: You had specific UX concerns (mentioned in a Slack message the same day) but chose not to raise them to avoid slowing the meeting. The design shipped with those issues and required a redesign two weeks later.

Better Approach: "I have two specific concerns about the navigation flow. Can we spend five minutes on them now, or should we schedule a follow-up before we finalize?"

Communication Strengths

1. Excellent at reading the room — You consistently notice when team members are disengaged and bring them back in. (Team standup, Feb 12: "Amir, you look like you have a thought on this. What are you thinking?")
2. Strong at celebrating wins — You open 70% of team meetings with a specific shout-out to someone's work
3. Good question-asking — Your clarifying questions in product reviews are consistently cited as helpful by other participants

Growth Opportunities

1. Name the issue in the first sentence: Practice stating the problem before any softening language. Write it down before the meeting if needed.
2. Remove hedge words: For one week, catch yourself every time you say "maybe," "kind of," or "sort of" in a 1:1 and rephrase directly.
3. Separate caring from directness: Being kind and being direct are not opposites. "I want this to go well for you, and right now it's off track" is both caring and clear.

Recommended Next Steps

1. Before your next 1:1 with a performance concern, write down the direct version of what you need to say — then say it first, before any softening
2. Ask a trusted peer to flag you in real-time (Slack DM) when they notice hedging in team meetings
3. Re-run this analysis in 30 days to measure change in hedging frequency
4. Consider analyzing your 1:1s vs. group meetings separately — the pattern may differ by context

Edge Cases

• Single meeting provided: Provide analysis but note that patterns are more reliable across multiple meetings; recommend running again with more data
• User not identified in transcript: Ask which speaker they are, or look for the most frequent speaker / meeting organizer
• Transcripts without timestamps: Provide quote-based examples without timestamps; note that timestamped transcripts enable more precise analysis
• User asks about someone else's patterns: Analyze as requested but frame feedback constructively — this is coaching, not surveillance
• Sensitive conversations (HR, terminations, personal topics): Analyze communication patterns without commenting on the substance of sensitive discussions; focus on how, not what
• No clear patterns found: Report that honestly — not every analysis will surface dramatic findings. Highlight strengths and suggest specific behaviors to watch for next time