Structured Log
Write structured entries to memory/YYYY-MM-DD.md instead of free-form text. This improves recall via memory_search.
When to Use
After any meaningful action:
- Made a decision or received a decision from the founder
- Discovered a new fact (via Exa, PostHog, web, conversation)
- Found an insight or pattern
- Completed a task with a result
- Updated a hypothesis
Entry Format
### HH:MM — [TYPE] Title
What happened (1-2 sentences).
**Facts:**
- concrete fact 1
- concrete fact 2
**Concepts:** keyword1, keyword2, keyword3
**Files:** path/to/file (if changed)
Types (TYPE)
| Type |
When |
| DECISION |
Decision made (by whom, why, what changes) |
| FACT |
New confirmed fact (source required) |
| INSIGHT |
Pattern, connection, conclusion from data |
| TOOL |
Finding from tool usage (PostHog, Exa, web) |
| TASK |
Completed task with result |
| HYPOTHESIS |
Created/updated/killed a hypothesis |
Rules
- Facts must be verifiable. "Signups = 0 for 7 days" (ok). "Seems low" (no).
- Concepts are for search. Write keywords you'd use to find this entry later via
memory_search. Include: names, products, metrics, topics.
- Files — only if changed. List only files that were created or modified.
- One entry = one topic. Don't mix a decision + tool finding in one entry.
- Narrative < 50 words. Context goes in facts, not prose.
Promotion to MEMORY.md
A fact deserves promotion to MEMORY.md if:
- It changes understanding of wedge, ICP, constraint
- It's confirmed 2+ times from different sources
- It affects current strategy or hypothesis
On promotion — add to the appropriate MEMORY.md section and remove outdated info.
Example
### 14:30 — [TOOL] Analytics: 41% onboarding drop-off
Analytics query revealed a high drop-off rate during onboarding.
**Facts:**
- 41% of users who start onboarding don't complete the key activation step (30d)
- Main cause: timeout on large data imports
- Affects the signup -> activated funnel stage
**Concepts:** onboarding, activation-blocker, funnel-drop, analytics, TTFV
**Files:** —
1---2name: structured-log3description: Write structured entries to daily memory log with facts, concepts, and files — use after any meaningful work, research, or decision4---56# Structured Log78Write structured entries to `memory/YYYY-MM-DD.md` instead of free-form text. This improves recall via `memory_search`.910## When to Use1112After any meaningful action:13- Made a decision or received a decision from the founder14- Discovered a new fact (via Exa, PostHog, web, conversation)15- Found an insight or pattern16- Completed a task with a result17- Updated a hypothesis1819## Entry Format2021```markdown22### HH:MM — [TYPE] Title2324What happened (1-2 sentences).2526**Facts:**27- concrete fact 128- concrete fact 22930**Concepts:** keyword1, keyword2, keyword331**Files:** path/to/file (if changed)32```3334## Types (TYPE)3536| Type | When |37|------|------|38| DECISION | Decision made (by whom, why, what changes) |39| FACT | New confirmed fact (source required) |40| INSIGHT | Pattern, connection, conclusion from data |41| TOOL | Finding from tool usage (PostHog, Exa, web) |42| TASK | Completed task with result |43| HYPOTHESIS | Created/updated/killed a hypothesis |4445## Rules46471. **Facts must be verifiable.** "Signups = 0 for 7 days" (ok). "Seems low" (no).482. **Concepts are for search.** Write keywords you'd use to find this entry later via `memory_search`. Include: names, products, metrics, topics.493. **Files — only if changed.** List only files that were created or modified.504. **One entry = one topic.** Don't mix a decision + tool finding in one entry.515. **Narrative < 50 words.** Context goes in facts, not prose.5253## Promotion to MEMORY.md5455A fact deserves promotion to MEMORY.md if:56- It changes understanding of wedge, ICP, constraint57- It's confirmed 2+ times from different sources58- It affects current strategy or hypothesis5960On promotion — add to the appropriate MEMORY.md section and remove outdated info.6162## Example6364```markdown65### 14:30 — [TOOL] Analytics: 41% onboarding drop-off6667Analytics query revealed a high drop-off rate during onboarding.6869**Facts:**70- 41% of users who start onboarding don't complete the key activation step (30d)71- Main cause: timeout on large data imports72- Affects the signup -> activated funnel stage7374**Concepts:** onboarding, activation-blocker, funnel-drop, analytics, TTFV75**Files:** —76```
Layneformalized225