insights

name	insights
description	Daily Zettelkasten insight generation — run CLI insight engine, create wiki pages for confidence>=0.7 insights, integrate to wiki. Schedule: 06:00 AM.
tags	["insights","zettelkasten","synthesis","daily"]
triggers	["cron",{"manual":"delegate_task"}]
updated	"2026-05-27T00:00:00.000Z"
created_by	agent

insights — Zettelkasten Insight Generator

Run the Zettelkasten insight generation engine and integrate high-confidence insights (≥0.7) into the wiki as synthesis pages.

Tool Protocol

Use terminal() with cat for ALL file reads in cron context. Use MCP tools for wiki operations.

Quick Start

1. Layer 2 Load: Read wiki/scratchpad/jobs/sheet.md and wiki/scratchpad/agent-sheets/insights/carryover.md
2. Layer 1 Start: Initialize or clear vault.md to act as your episodic scratchpad for this session
3. Run insight generation CLI (Step 1), logging status to vault.md
4. Read and evaluate generated insights (Step 2)
5. Create wiki pages for confidence ≥ 0.7 (Step 3)
6. Deliver (silent if no new pages created)
7. MOP Compression: Compress vault.md into carryover.md (Layer 1 → Layer 2)

Workflow

Step 1 — Run Insight Generation

Defense-in-depth timeout — the CLI can hang. Use triple timeout:

cd /home/ty/Repositories/ai_workspace/project-synapse-mcp && \
    timeout --kill-after=10s 580s uv run python scripts/generate_insights.py \
    --topic general --print --max-runtime 540 2>&1

Do NOT use MCP generate_insights() — it times out at 300s (the CLI has a much higher limit).

If the CLI times out (exit code 124 from timeout):

• This is normal and expected — note in carryover as "CLI watchdog timeout"
• Check if partial output was produced: cat data/insights/latest.json 2>/dev/null
• If no output, write carryover noting timeout and respond [SILENT]

Step 2 — Evaluate Generated Insights

cat /home/ty/Repositories/ai_workspace/project-synapse-mcp/data/insights/latest.json

For each insight in the output:

• Confidence ≥ 0.7: Create a wiki page (Step 3)
• Confidence < 0.7: Note in carryover only — no page created
• Duplicate check: synapse_recall(query="{insight title}") — skip if already exists

Step 3 — Create Wiki Pages

For each high-confidence insight:

1. Map title to slug:
   "Titans Memory Architecture" → "titans-memory-efficiency-insight"

2. wiki_write_page(path="wiki/synthesis/{slug}.md", content=...)
   Frontmatter:
   ---
   created: {today}
   updated: {today}
   type: synthesis
   summary: "{insight title}"
   tags: [insights, zettelkasten, {topic}]
   status: active
   confidence: {from insight}
   ---

3. Cross-link:
   wiki_search(query="{insight topic}")
   → Add [[wikilinks]] to related concepts/entities

4. Record to episodic memory:
   synapse_remember(content="Insight: {title} (confidence {score}) → wiki/synthesis/{slug}.md")

5. wiki_update_index()

Step 4 — Deliver

If new pages were created, deliver a summary:

**Insights — {YYYY-MM-DD}**

{N} insights generated | {N} pages created (confidence ≥ 0.7)

**New Pages:**
- [[slug]] — {title} (confidence: {score})

**Below Threshold:**
- {title} (confidence: {score}) — noted, no page

If no new pages created (timeout or all below threshold), respond [SILENT].

Final Step — MOP Compression (Layer 1 → Layer 2)

Read your vault.md (Episodic Trace) and compress it into wiki/scratchpad/agent-sheets/insights/carryover.md (Semantic State), adhering to the ~512 token bound:

---
created: { original date }
updated: { today's date }
type: carryover
summary: "{N} insights generated, {N} wiki pages created"
tags: [insights, carryover]
---

Include:

• What Was Done: Insights generated, pages created (title + confidence + slug)
• What Remains: - [ ] checklist (below-threshold insights worth revisiting, CLI issues)
• Kanban Status: Items already surfaced

Once compressed, clear or archive your vault.md so the next session starts fresh.

Critical Paths

• CLI script: /home/ty/Repositories/ai_workspace/project-synapse-mcp/scripts/generate_insights.py
• Output: /home/ty/Repositories/ai_workspace/project-synapse-mcp/data/insights/latest.json
• Wiki pages: wiki/synthesis/{slug}.md
• Reports: wiki/scratchpad/jobs/reports/insights/
• Carryover: wiki/scratchpad/agent-sheets/insights/carryover.md

MCP Tools

Tool	Purpose
synapse_remember	Record insights to episodic memory
synapse_recall	Check for duplicate insights
wiki_write_page	Create synthesis pages (confidence ≥ 0.7)
wiki_search	Find related pages for cross-linking
wiki_update_index	Refresh index after new pages

CRITICAL CONSTRAINT: DO NOT interact with the Kanban board or run kanban scripts. Output open items as - [ ] in the ## What Remains section of your carryover.md. The overseer will create Kanban tickets and assign them to you in jobs/sheet.md. You may run the official generate_insights.py engine as explicitly documented above, but do not author your own scripts. Use standard MCP tools exclusively for wiki operations.

Fallback Patterns

• CLI timeout (exit 124): Normal. Check for partial output, note in carryover, [SILENT]
• CLI crashes (other exit code): Note error in carryover, [SILENT]
• MCP unavailable: Note in carryover, skip wiki page creation, deliver insights as text report
• All insights below threshold: Note in carryover, [SILENT]

Quality Standards

• Only create pages for confidence ≥ 0.7
• Use proper slug mapping (descriptive, hyphen-separated)
• Tag with: insights, zettelkasten, {topic}
• Every new page must cross-link to at least 2 existing pages
• Record every insight to episodic memory (even below-threshold ones)

Kanban Queue

Drain your assigned cards at the start of every cycle:

kanban_list(lane="ready", assignee="<your-profile>")

If the queue is empty, proceed with the agent's normal work. Cross-agent cards (with tenant=, lane=triage) are routed by the Overseer — you do not need to action them directly.

Full contract (call signatures, intents, the tenant trap): overseer/references/kanban-coordination.md.