storage-basicmemory（存储基本内存）

Core principle: All project artifacts must be stored in basicmemory under a [ProjectId], with explicit linking and status tracking to maintain single source of truth across concurrent projects.

[!NOTE] Before starting anything, ensure you have identified the correct [ProjectId] for the project you are working on.

Requirements:

• use basicmemory mcp tools to read and write [Project Artifacts].
• Recognise types of [Project Artifacts]: [Spec], [Research], [Descision], [Epic], [Story], [Task], [Retrospective].
• Store [Project Artifacts] in basicmemory under the correct project context using [ProjectId].

How to interact with [Project Artifacts] storage?

CRITICAL All [Project Artifacts] are interacted with via basicmemory mcp tools. FAILURE MODE Interacting with [Project Artifacts] via the file system directly is not allowed and will lead to disorganization and loss of data.

• basicmemory_read_note - Read markdown notes
• basicmemory_read_content - Read file raw content by path
• basicmemory_view_note - View formatted notes
• basicmemory_write_note - Create/update markdown notes
• basicmemory_edit_note - Edit existing notes with operations
• basicmemory_move_note - Move notes to new locations
• basicmemory_delete_note - Delete notes by title
• basicmemory_canvas - Create Obsidian canvas files
• basicmemory_search_notes - Search across knowledge base
• basicmemory_search - Search for content across knowledge base
• basicmemory_fetch - Fetch full contents of search results
• basicmemory_recent_activity - Get recent activity
• basicmemory_build_context - Build context from memory URIs
• basicmemory_list_memory_projects - List all available projects
• basicmemory_create_memory_project - Create new projects
• basicmemory_delete_project - Delete projects
• basicmemory_list_directory - List directory contents
• basicmemory_sync_status - Check file sync status

Planning Artifact Frontmatter

Frontmatter for planning artifacts is important. It must remain at the top of the markdown file as YAML, NOT as markdown code blocks.

Understanding Frontmatter

Templates contain YAML frontmatter blocks at the top, between --- delimiters:

---
epic_id: 1
title: Epic Title
status: planned
priority: high
estimated_effort: 2 weeks
linked_spec:
  - type: spec
    target: 2.1.1-spec-title
---
# Rest of markdown content here

This is actual YAML frontmatter, not a markdown code block. When you create artifacts, this structure must be preserved exactly.

Frontmatter Preservation Workflow

When creating or updating planning artifacts via basicmemory_write_note: Step 1: Parse Template File

• Extract the frontmatter block (everything between first --- and second ---)
• Extract the markdown body (everything after the closing ---)
• Keep them separate Step 2: Assemble Content for basicmemory The content passed to basicmemory_write_note must be:

---
field1: value1
field2: value2
links:
  - type: epic
    target: 2.1.1-epic-title
---
# Markdown Heading
Markdown body content here...

Critical requirements:

• Content MUST start with ---
• Frontmatter block MUST end with ---
• Blank line MUST separate frontmatter from markdown body
• All frontmatter fields from template MUST be preserved
• No ## Frontmatter markdown sections Step 3: Call basicmemory_write_note

basicmemory_write_note(
    title="artifact-title",
    folder="2-epics",  # or 1-prds, 3-research, etc.
    content=assembled_content,  # With frontmatter at top
    entity_type="epic",  # or spec, research, story, task, decision
    tags=extracted_tags,  # Optional, if tags are in frontmatter
    project=project_id
)

Step 4: basicmemory Handles the Rest basicmemory automatically:

• Detects the YAML frontmatter (via has_frontmatter())
• Parses all frontmatter fields
• Removes frontmatter from markdown body (via remove_frontmatter())
• Preserves all fields in the final artifact as YAML frontmatter
• Converts back to proper markdown with YAML frontmatter

Field Mapping

Template frontmatter fields become YAML frontmatter in the artifact:

• epic_id: 1 → stored in frontmatter as epic_id
• status: planned → stored as status
• priority: high → stored as priority
• estimated_effort: 2 weeks → stored as estimated_effort
• linked_spec: [...] → stored as YAML list in frontmatter
• derived_from: [...] → stored as YAML list in frontmatter These are structured metadata fields, not markdown content.

Common Mistake: Markdown Code Blocks

❌ WRONG - Creates markdown code block, loses frontmatter:

---
title: Artifact Title
type: epic
permalink: some-path
---
# Artifact Title
## Frontmatter
\`\`\`yaml
epic_idid: 1
status: planned
\`\`\`
Rest of content...

✅ CORRECT - Preserves frontmatter as YAML:

---
title: Artifact Title
type: epic
permalink: some-path
epic_id: 1
status: planned
---
# Artifact Title
Rest of content...

Implementation Best Practices

1. Always parse frontmatter carefully - Extract the YAML block as-is from the template
2. Preserve field order - YAML field order matters for readability
3. Validate YAML syntax - Ensure frontmatter is valid YAML before passing to basicmemory
4. Include all template fields - Don't drop fields; merge with generated fields instead
5. Use proper YAML lists - Use - for list items in frontmatter, not JSON arrays
6. Test round-trip - After creation, verify the artifact has frontmatter, not markdown code blocks

[ProjectId] Naming and Format

ProjectId Convention:

• Format: slugified-project-name (kebab-case, lowercase alphanumeric + hyphens)
• Source: Derived from git repository name or project name
• Generated by: ./scripts/get_project_id.sh (automatically slugifies)
• Examples:

- Repository: github.com/username/dotfiles → ProjectId: dotfiles - Project name: "User Authentication System" → ProjectId: user-authentication-system - Example name "My App v2" → ProjectId: my-app-v2

Creating new Projects

If you need to create a new project in basicmemory, always ensure that

• Follow the [ProjectId] naming convention.
• Confirm the project does not already exist by listing existing projects first.
• Store the project at ~/Notes/Projects/[ProjectId]
• Never store the project in the same folder as the repo you are working on.