Ceos Checkup

Step 3: Trend Analysis

For each component, show direction vs. the previous checkup:

Trend Analysis (Latest vs Previous)
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

| Component | Previous | Current | Change | Trend |
|-----------|----------|---------|--------|-------|
| Vision    | 4.0      | 4.3     | +0.3   | ↑     |
| People    | 3.5      | 4.1     | +0.6   | ↑↑    |
| Data      | 3.2      | 4.0     | +0.8   | ↑↑    |
| Issues    | 4.0      | 4.5     | +0.5   | ↑     |
| Process   | 3.0      | 3.5     | +0.5   | ↑     |
| Traction  | 3.8      | 4.3     | +0.5   | ↑     |
| OVERALL   | 3.8      | 4.2     | +0.4   | ↑     |

Trend indicators:

• ↑↑ = improved by 0.5+
• ↑ = improved by 0.1-0.4
• → = unchanged (within 0.1)
• ↓ = declined by 0.1-0.4
• ↓↓ = declined by 0.5+

Step 4: Flag Stale Checkups

Calculate days since the most recent checkup. If > 120 days:

📅 Last checkup was [N] days ago (recommended: quarterly / every 90 days).
   Consider running a new checkup to track progress.

Step 5: Component Deep Dive

Ask: "Want to drill into a specific checkup or component?"

If the user selects a checkup, display the full file with all participant ratings and action items.

If the user selects a component, show that component's score history across all checkups as a progression.

Mode: Compare

Use when analyzing alignment gaps between team members' ratings on a specific checkup.

Step 1: Select Checkup

Default to the most recent checkup. If multiple exist, ask: "Which checkup? (Default: most recent — [date])"

Load the selected checkup file and parse participant ratings.

If the checkup has only one participant: "Alignment analysis requires 2+ participants. This checkup has only one. Run a new checkup with multiple team members, or select a different checkup."

Step 2: Calculate Per-Question Statistics

For each of the 20 questions, calculate across all participants:

• Mean: Average rating
• Min: Lowest rating
• Max: Highest rating
• Range: Max - Min
• Spread: Whether ratings are clustered or dispersed

Step 3: Identify Alignment Gaps

Flag questions where the team diverges significantly:

• High variance: Range >= 3 (e.g., one person rated 5, another rated 2)
• Moderate variance: Range = 2
• Aligned: Range <= 1

Sort by range (highest first) to surface the biggest gaps.

Step 4: Display Alignment Gap Table

Alignment Analysis — [Date] Checkup
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Participants: Brad, Sarah, Mike

HIGH VARIANCE (Range >= 3) — Discuss These
| # | Question (abbreviated)       | Brad | Sarah | Mike | Avg | Range | Flag |
|---|------------------------------|------|-------|------|-----|-------|------|
| 7 | Proven customer process      | 5    | 2     | 3    | 3.3 | 3     | 🔴   |
| 16| Core Processes documented    | 4    | 1     | 3    | 2.7 | 3     | 🔴   |

MODERATE VARIANCE (Range = 2)
| # | Question (abbreviated)       | Brad | Sarah | Mike | Avg | Range | Flag |
|---|------------------------------|------|-------|------|-----|-------|------|
| 3 | Core Focus clear             | 5    | 3     | 4    | 4.0 | 2     | ⚠️   |

ALIGNED (Range <= 1)
| # | Question (abbreviated)       | Brad | Sarah | Mike | Avg | Range |
|---|------------------------------|------|-------|------|-----|-------|
| 1 | Clear vision in writing      | 5    | 5     | 4    | 4.7 | 1     |
| 11| Leadership team trust        | 5    | 4     | 5    | 4.7 | 1     |
| ... (remaining aligned questions)

Step 5: Alignment Summary

Alignment Summary
━━━━━━━━━━━━━━━━
High variance (Range >= 3):  2 questions — need team discussion
Moderate variance (Range 2): 1 question — worth reviewing
Aligned (Range <= 1):        17 questions — team consensus

Top Discussion Topics:
1. Q7: "Proven customer process" — Range 3 (Brad: 5, Sarah: 2)
   → Team sees this very differently. Discuss what "proven process" means.
2. Q16: "Core Processes documented" — Range 3 (Sarah: 1, Brad: 4)
   → Consider running ceos-process to document core processes.

Step 6: Drill Down

Ask: "Want to discuss any specific question or create issues from these gaps?"

If the user wants to create issues, suggest using ceos-ids with the specific gap as the issue topic.

Output Format

Run Mode

1. Progressive display: Show questions component-by-component with running scores
2. Component completion: Show subtotal after each component
3. Final summary: Overall score, component scores table, strengths, opportunities
4. Action items: Table with owner, due date, related component

Review Mode

1. History table: Date, participants, overall score, per-component scores, status
2. Trend analysis: Previous vs. current with direction arrows
3. Stale checkup warning: If > 120 days since last

Compare Mode

1. Alignment gap table: Questions sorted by variance (highest first)
2. Per-participant ratings: Side-by-side columns
3. Summary statistics: Count of high/moderate/aligned questions
4. Discussion topics: Top gaps with context for team conversation

Guardrails

• Always show the complete file before writing. Never save a checkup file without displaying it to the user for review and approval.
• Validate ratings are 1-5 integers. If a participant gives a rating outside 1-5, re-prompt: "Ratings must be 1-5 (1=strongly disagree, 5=strongly agree)." Do not accept decimals.
• Present questions in component order. Always group questions by component (Vision, People, Issues, Traction, Process, Data) rather than randomizing or reordering.
• Don't skip questions. All 20 questions should be asked. If a participant cannot answer, store null and note it. Don't silently omit questions.
• Don't auto-invoke other skills. When low scores suggest using another skill (e.g., "Process is weak, run ceos-process"), mention the option but let the user decide. Say "Would you like to run ceos-process?" rather than invoking it automatically.
• Sensitive data warning. On first use in a session, remind the user: "Organizational Checkup ratings contain sensitive assessments of company health. Ensure this repository is private, not public."
• Use alignment gaps as conversation starters. Compare mode surfaces disagreements. Frame these constructively: "The team sees this differently — worth discussing" rather than "These people are wrong."
• Don't interpret scores prescriptively. The skill calculates and displays scores, but the meaning and action belong to the leadership team. Suggest relevant skills, don't mandate actions.
• Flag stale checkups. In Review mode, flag if the most recent checkup is > 120 days old. This is informational, not blocking.
• Questions are canonical. Do not modify, reword, or add custom questions. This skill implements the standard EOS Organizational Checkup.

Integration Notes

V/TO (ceos-vto)

• Direction: Read
• What data: data/vision.md — Core Values, Core Focus, 10-Year Target, 1-Year Plan
• Purpose: Provides context for Vision component questions (Q1-6). Core Values referenced in Run mode setup. V/TO changes may shift how the team rates Vision questions.

Accountability Chart (ceos-accountability)

• Direction: Read
• What data: data/accountability.md — seats and owners
• Purpose: Suggests participants for checkup sessions. The Accountability Chart defines who the "leadership team" is. Also provides context for People component questions (Q7-10).

Rocks (ceos-rocks)

• Direction: Read
• What data: data/rocks/ — current quarter's Rocks
• Purpose: Provides context for Traction component questions (Q13-15). Current Rock status shows whether the team is executing with discipline.

Scorecard (ceos-scorecard)

• Direction: Read
• What data: data/scorecard/ — metrics and weekly entries
• Purpose: Provides context for Data component questions (Q17-20). Active metrics count and Scorecard health inform the Data component discussion.

People Analyzer (ceos-people)

• Direction: Read
• What data: data/people/ — person evaluations
• Purpose: Provides context for People component questions (Q8-10). Current "right people, right seats" percentage gives context for how the team rates People health.

IDS (ceos-ids)

• Direction: Read
• What data: data/issues/open/ and data/issues/resolved/
• Purpose: Provides context for Issues component questions (Q11-12). Open issues count and resolution patterns inform the Issues component discussion. Low-scoring components may generate new issues via ceos-ids.

Annual Planning (ceos-annual)

• Direction: Related
• What data: data/annual/ — annual planning records
• Purpose: Annual planning includes an organizational checkup as Section 4. Checkup results from ceos-checkup can be referenced during annual planning to track year-over-year progress.

Quarterly Planning (ceos-quarterly-planning)

• Direction: Related
• What data: data/quarterly/ — quarterly planning records
• Purpose: Quarterly planning may reference checkup trends to inform Rock priorities. Low-scoring components suggest where to focus next quarter's Rocks.

Write Principle

Only ceos-checkup writes to data/checkups/. Other skills (particularly ceos-annual and ceos-quarterly-planning) may read checkup data for trend analysis and planning context, but they do not create or modify checkup files.