Edit: handover_prompts/notion_access_resolution_prompt.md

Note Content (Markdown)

# Handover Prompt: Notion Easier Agency Access Resolution & Content Extraction

## Context Reset - 100% Fresh Context Assumption
This handover prompt assumes no prior knowledge of the current session state. The agent should verify everything from scratch.

## Objective
Resolve Notion access permissions for the Easier Agency page/database and extract its content (Marketing, Sales, Relationships, Fulfilment, Operations departments) for gbrain ingestion.

## Phase 1: Verify MCP Connection & Integration Status

### 1.1 Check Available MCP Servers
```
hermes mcp list
```
Expected output should show:
- Notion server connected via npx notion-mcp-server
- 22 tools available

### 1.2 Test Basic Notion API Access
```
hermes mcp execute notion get_self '{}'
```
Should return bot user info for "Hermes" integration.

### 1.3 Verify API Key Validity
Search for known accessible content to confirm token works:
```
hermes mcp execute notion search_pages '{"query": "Hermes WebUI", "pageSize": 5}'
```
Should return the Hermes WebUI page if token is valid.

## Phase 2: Locate & Diagnose Easier Agency Access Issue

### 2.1 Search for Easier Agency
```
hermes mcp execute notion search_pages '{"query": "Easier Agency", "pageSize": 10}'
```
Expected results should include:
- Easier Agency (page or database)
- Easier Marketing 
- Possibly other related items

Note the IDs returned, particularly:
- Page ID: likely `eb7df9ac-6a51-49a1-8af5-5fcae6ff2674`
- Database parent ID: likely `86b9c355cc0c4438b964beb0141ecab9`

### 2.2 Test Direct Access Attempts
Try to access the suspected page ID:
```
hermes mcp execute notion get_page '{"page_id": "eb7df9ac-6a51-49a1-8af5-5fcae6ff2674"}'
```

Try to access the suspected database ID:
```
hermes mcp execute notion get_database '{"database_id": "86b9c355cc0c4438b964beb0141ecab9"}'
```

Both should return `object_not_found` errors if not shared.

### 2.3 Check Software Logins Database (Alternative Path)
Since Easier Agency appears in Software Logins per earlier findings:
```
hermes mcp execute notion get_data_source '{"data_source_id": "d6f9882d-56c7-493e-a026-8b6012e1b30b"}'
```
This should succeed (Software Logins is already accessible).

Then query for Easier Agency within Software Logins:
```
hermes mcp execute notion query_database '{
  "data_source_id": "d6f9882d-56c7-493e-a026-8b6012e1b30b",
  "filter": {"property": "Page", "title": {"equals": "Easier Agency"}},
  "pageSize": 1
}'
```
This should return the Easier Agency database entry but NOT the actual page content.

## Phase 3: Guide User to Share Content (Critical Step)

At this point, the agent should instruct the human user:

**ACCESS PERMISSION REQUIRED**

The Notion integration token is valid and connected, but the specific Easier Agency content has not been shared with the "Hermes" integration.

**User Action Required:**
1. Open Notion and navigate to the Easier Agency page
2. Click the ••• menu (top-right)
3. Select "Add connections" or "Connect"
4. Ensure the integration named **"Hermes"** is selected
5. Confirm the sharing

**Alternative if above doesn't work:**
1. Open the Software Logins database (which contains Easier Agency entry)
2. Share that database with the "Hermes" integration
3. This may grant access to the Easier Agency entry within it

**Verification:** After sharing, wait 30 seconds then retry the get_page/get_database calls above.

## Phase 4: Extract Easier Agency Directory Structure (Post-Access)

Once access is granted, extract the full directory:

### 4.1 Fetch Root Page Content
```
hermes mcp execute notion get_page '{
  "page_id": "eb7df9ac-6a51-49a1-8af5-5fcae6ff2674",
  "include_properties": true
}'
```

### 4.2 Get Page Content as Markdown
```
hermes mcp execute notion get_page_markdown '{
  "page_id": "eb7df9ac-6a51-49a1-8af5-5fcae6ff2674"
}'
```

### 4.3 Get Child Blocks (Directory Structure)
```
hermes mcp execute notion get_block_children '{
  "block_id": "eb7df9ac-6a51-49a1-8af5-5fcae6ff2674",
  "pageSize": 100
}'
```

### 4.4 Process Each Department Block
For each top-level block representing a department (Marketing, Sales, Relationships, Fulfilment, Operations):
- Extract the heading/title
- Get the description/SOP content
- If it links to sub-pages or databases, follow those links
- Extract all relevant content recursively

## Phase 5: Extract Supporting Content (Already Accessible)

While waiting for Easier Agency access, extract these verified accessible sources:

### 5.1 Software Logins Database (Metadata Only - NO Credentials)
```
hermes mcp execute notion query_database '{
  "data_source_id": "d6f9882d-56c7-493e-a026-8b6012e1b30b",
  "pageSize": 100
}'
```
Extract: Service name, In use? status, Cost/month, Activity, Username, Notes
**Do NOT extract:** Password/SSO fields (these contain credentials)

### 5.2 API Keys Database (Metadata Only - NO Raw Keys)
```
hermes mcp execute notion query_database '{
  "data_source_id": "f1be5677-deda-4ce4-bd64-7a0e12439022",
  "pageSize": 100
}'
```
Extract: Service name, Status, Last Rotated, Used By, Notes
**Do NOT extract:** API Key field (contains raw secrets)

### 5.3 Outreach Tracker Database
```
hermes mcp execute notion query_database '{
  "data_source_id": "2ee0308b-4848-80e2-b2d4-000bf535a2b4",
  "pageSize": 100
}'
```
Extract all fields (Name, Created, Chase 1/2, Chase 2, Response?, QA Note)

### 5.4 ContentOS Pages (Full Extraction)
Get markdown for:
- ContentOS PRD: `3210308b-4848-8163-b899-d9631caa16ae`
- ContentOS Pipeline: `3230308b-4848-8048-bc54-fa1729e7eda1`  
- ContentOS Phased Tasks: `3230308b-4848-80bf-aac4-cbb64c785e3a`

### 5.5 OUTBOUND Master Copies (Full Extraction)
Get markdown for:
- OUTBOUND 1: `3250308b-4848-8166-a323-d3eaa96741af`
- OUTBOUND 2: `3250308b-4848-8103-afd5-e928c7aca221`

### 5.6 Company Records
- Easier Marketing: `2250308b-4848-80ad-ba84-000be1cc02fe`
- Hermes WebUI: `3750308b-4848-8028-928f-c1787a1b7e0d`

## Phase 6: Prepare Content for gbrain Ingestion

### 6.1 Content Format Standardization
For each extracted piece, prepare in gbrain format:
```
# [Title]

[Compiled truth - current synthesis, key points, actionable information]

---

[Timeline - append-only evidence log with sources and dates]
```

### 6.2 Apply User Steering Guidance
- **Primary ICP Focus**: Tag and emphasize content related to UK 7-figure ecom brands (profit-focused)
- **Secondary/Passive**: Label SaaS/service business content as secondary attraction strategy
- **ContentOS Adaptation**: Identify Airtable dependencies for future replacement with Hermes-based automations
- **Post Type Refinement**: Analyze existing content types and distill to ~5 core categories
- **Retirement Notices**: Flag any Airtable-reliant processes for migration

### 6.3 File Organization Structure
Propose gbrain vault structure:
```
people/
  anthony-stratton.md
organisations/
  easier-agency.md
  easier-marketing.md
projects/
  contentos-pipeline.md
  outreach-framework.md
  easier-agency-departments/
    marketing.md
    sales.md
    relationships.md
    fulfilment.md
    operations.md
processes/
  contentos-workflows.md
  outreach-sops.md
decisions/
  icp-strategy-decision.md
  tech-stack-decision.md
etc.
```

## Phase 7: Verification & Handoff

### 7.1 Internal Validation
- Confirm no sensitive data (passwords, keys, tokens) is included
- Verify directory structure matches user's expectation
- Check that ICP focus is properly captured and labeled

### 7.2 Handover Package
Provide:
1. Extracted content in gbrain-ready format
2. Summary of what was accessed vs. what remained inaccessible
3. List of any content that requires user verification/correction
4. Recommended next steps for gbrain ingestion

## Success Criteria
- [ ] MCP Notion connection verified functional
- [ ] Easier Agency page/database access granted via sharing
- [ ] Five departments extracted: Marketing, Sales, Relationships, Fulfilment, Operations
- [ ] Supporting content extracted (metadata only for sensitive databases)
- [ ] Content formatted for gbrain with compiled truth + timeline
- [ ] User steering applied (ICP focus, ContentOS adaptation, etc.)
- [ ] No sensitive data leaked in output

## Troubleshooting Guide

### If Access Still Denied After Sharing:
1. Wait 60-90 seconds for permissions to propagate
2. Verify exact IDs being accessed match what was shared
3. Check if there are duplicate "Easier Agency" entries
4. Try sharing the parent workspace instead
5. As fallback, have user export as Markdown and share via file upload

### Large Content Handling:
- Extract department by department if timeout risk
- Use reasonable pagination limits (pageSize: 50-100)
- Prioritize text over complex embedded databases initially
- For databases, extract schema + sample records first

## Estimated Time
- Access diagnosis: 5-10 minutes
- User sharing action: 2-5 minutes (user-dependent)
- Permission propagation: 1-2 minutes
- Content extraction: 15-30 minutes
- Preparation: 10-20 minutes
- Total: 30-65 minutes (mostly waiting on user action)

---
*This handover prompt is designed for Claude or another AI agent to execute with minimal guidance. The agent should verify each step before proceeding and report back at major phase completions.*