Learning Synthesis

Aggregate, analyze, and synthesize a week's learnings into themes, connections, and streak metrics. Used by: /founder-os:learn:weekly.

Purpose and Context

Transform raw daily learnings into actionable weekly intelligence. Detect recurring themes, identify cross-topic connections, track learning consistency via streaks, and compare activity trends to the previous week. Output is stored in a dedicated Weekly Insights Notion database and displayed as a formatted chat summary.

---

Notion Databases

Read From: [FOS] Learnings (fallback: Founder OS HQ - Learnings, then Learning Log Tracker - Learnings)

Search for "[FOS] Learnings" first. If not found, try "Founder OS HQ - Learnings". If not found, fall back to "Learning Log Tracker - Learnings". Query entries where Week matches the target week (format: YYYY-WNN). This retrieves all learnings for the synthesis window.

Write To: [FOS] Weekly Insights (fallback: Founder OS HQ - Weekly Insights, then Learning Log Tracker - Weekly Insights)

Use dynamic database discovery: search the workspace for a database titled "[FOS] Weekly Insights". If not found, try "Founder OS HQ - Weekly Insights". If not found, fall back to "Learning Log Tracker - Weekly Insights". If none is found, report: "Weekly Insights database not found. Ensure the Founder OS HQ workspace template is installed in your Notion workspace." Do not create the database automatically.

Schema

Property	Type	Notes
Week	title	ISO week identifier as YYYY-WNN. Primary key for upsert.
Summary	rich_text	Narrative 3-5 sentence synthesis of the week's learnings.
Top Themes	multi_select	2-4 most frequent topic categories from the week.
Learning Count	number	Total number of learnings logged during the week.
Most Active Topic	select	Single topic with the highest frequency for the week.
Key Connections	rich_text	Prose identifying 2-3 cross-learning patterns.
Learnings List	rich_text	Scannable bulleted index of all learning titles.
Streak Days	number	Consecutive weeks with at least one learning logged.
Vs Last Week	select	More active, Same pace, or Less active.
Source Mix	rich_text	Breakdown of learning sources (e.g., "Experience: 3, Reading: 2").
Generated At	date	ISO 8601 timestamp when the synthesis was generated.

Idempotent Upsert Rule

Search the database for a page whose Week (title) matches the target week string. If a match exists, update all properties. If no match exists, create a new page. This allows re-running /founder-os:learn:weekly for the same week to refresh the analysis.

---

Week Boundary Calculation

Determine the target week window:

• Default: the current ISO week (Monday 00:00 to Sunday 23:59)
• Override: --week=YYYY-WNN flag specifies an explicit week
• ISO weeks start on Monday and end on Sunday
• Calculate the Monday and Sunday dates from the ISO week number

Example: 2026-W10 → Monday 2026-03-02 through Sunday 2026-03-08.

---

Theme Detection

Analyze the Topics distribution across all learnings in the target week. For the full algorithm including frequency counting, minimum thresholds, and selection rules, read ${CLAUDE_PLUGIN_ROOT}/skills/learn/learning-synthesis/references/theme-detection-algorithm.md.

Quick Reference

• Count the frequency of each topic across all learnings in the week
• Select the 2-4 most frequent topics as Top Themes
• If fewer than 3 learnings, select the top 1-2 topics instead
• The single most frequent topic becomes Most Active Topic
• When there is a tie for Most Active Topic, prefer the topic that appeared in the most recent learning

---

Key Connections

Identify 2-3 cross-learning patterns where insights from different topics reinforce, contrast, or build upon each other.

Connection Detection

• Find pairs of learnings that share at least one Topic but differ in at least one other Topic
• Prioritize pairs where insights have complementary or contrasting content
• Describe each connection as a brief prose sentence explaining the relationship

Example: "Your Technical insight about error handling patterns reinforced your Process learning about code review checklists — both point toward catching issues earlier in the development cycle."

When fewer than 3 learnings exist for the week, or all learnings share identical Topics, note "No cross-topic connections detected this week."

---

Streak Calculation

Track consecutive weeks of learning activity. For the full backwards-scan algorithm, gap detection, and display formatting, read ${CLAUDE_PLUGIN_ROOT}/skills/learn/learning-synthesis/references/streak-calculation.md.

Quick Reference

• Query the Weekly Insights database ordered by Week title descending
• Starting from the target week, scan backwards checking for consecutive week entries
• A streak breaks when a week has no corresponding entry (gap detected)
• Store the streak count in Streak Days (named for the metric, counts weeks)
• Display format: "N weeks" with fire emoji 🔥 for streaks of 4+ weeks
• First-ever weekly summary: display "Starting your streak! 🌱"

---

Trend Comparison

Compare the target week's Learning Count to the previous week's Learning Count using 20% thresholds:

Condition	Vs Last Week Value
Current > Previous × 1.2	More active
Current < Previous × 0.8	Less active
Otherwise	Same pace

When the previous week has no entry in the Weekly Insights database, skip the comparison and note "No previous week data for comparison."

---

Source Mix

Count the occurrences of each Source Type across the week's learnings. Format as a comma-separated breakdown:

Experience: 3, Reading: 2, Conversation: 1

Order by frequency descending. Omit source types with zero occurrences.

---

Learnings List

Generate a scannable bulleted list of all learning titles from the week:

• Granular Try-Catch Blocks for Async Errors
• Kickoff Questionnaires Reduce Client Scope Creep
• Notion API Pagination Requires Has_More Check

Order by Logged At ascending (chronological).

---

Summary Generation

Write a 3-5 sentence narrative synthesis covering:

1. Total learning volume and how it compares to the trend
2. Dominant themes and what they suggest about current focus
3. Notable connections or patterns across learnings
4. One forward-looking insight or recommendation

Write in a professional but conversational tone. Use specific references to individual learnings when relevant.

---

Output Format

Chat Output (always displayed)

## 📊 Weekly Learning Insights — [YYYY-WNN]

**[N] learnings** logged this week | Streak: [N] weeks [emoji] | [Vs Last Week]

### 🎯 Top Themes
[Topic1], [Topic2], [Topic3]
Most active: [Topic] ([count] learnings)

### 📝 Summary
[3-5 sentence narrative synthesis]

### 🔗 Key Connections
- [Connection 1 prose]
- [Connection 2 prose]

### 📋 All Learnings
[Bulleted list of titles with dates]

### 📊 Source Mix
[Source breakdown]

Notion Output (when --output includes notion or both)

Store all computed values in the Weekly Insights database using the schema above.

File Output (when --output=PATH specified)

Write the chat format to the specified file path.

---

Empty Week Handling

When no learnings exist for the target week:

1. Do not create a Weekly Insights entry
2. Display: "No learnings logged for [YYYY-WNN]. Use /founder-os:learn:log to capture insights throughout the week."
3. Show the current streak status (which will have broken if this is the current week)

---

Edge Cases

Notion Unavailable

Report the error clearly. Display whatever synthesis is possible from cached data, or report that synthesis requires Notion access.

Single Learning Week

Generate the full synthesis even for a single learning. Theme detection selects the learning's topics. Connections section notes "Only one learning this week — log more for cross-topic connections." Streak still counts.

Very High Volume Weeks

For weeks with more than 20 learnings, the Learnings List in chat output should cap at 15 entries with a "... and N more" suffix. The full list is stored in Notion regardless.

---

Additional Resources

Reference Files

For detailed algorithms, consult:

• ${CLAUDE_PLUGIN_ROOT}/skills/learn/learning-synthesis/references/theme-detection-algorithm.md — Frequency counting methodology, minimum thresholds based on learning count, Top Themes selection rules, and tie-breaking logic
• ${CLAUDE_PLUGIN_ROOT}/skills/learn/learning-synthesis/references/streak-calculation.md — Backwards-scan algorithm for consecutive week detection, gap identification, display formatting rules, and edge cases for streak reset