markplane

name	markplane
description	Structured task-oriented memory via Markplane. Use when working with tasks, project state, or when you need to capture decisions and action items.

Markplane — Structured Task Memory

This workspace uses Markplane for structured, task-oriented memory. Markplane stores tasks, epics, plans, and notes as markdown files in .markplane/ with compressed AI-readable summaries in .context/.

Reading project state

The compressed summary is already in your system prompt (injected automatically). For deeper detail:

1. Read .markplane/.context/summary.md for current project overview
2. Read .markplane/.context/active-work.md for in-progress items
3. Read .markplane/.context/blocked-items.md for blocked items
4. Read .markplane/backlog/INDEX.md to browse all tasks
5. Read specific items by ID: .markplane/backlog/items/TASK-xxxxx.md

Working with items

Markplane items are markdown files with YAML frontmatter. You can read and write these files directly — especially the free-form markdown body below the frontmatter, which is where detailed context, notes, and implementation details belong.

Use MCP tools for structural operations:

• markplane_add — create a new item (generates ID, scaffolds template)
• markplane_update — change frontmatter fields (status, priority, assignee, etc.)
• markplane_link — manage cross-references (depends_on, blocks, related) and sync them in frontmatter
• markplane_query — find items by status, priority, type, tags
• markplane_show — read full item details by ID
• markplane_sync — regenerate INDEX.md and .context/ summaries

Read and edit files directly for content:

After creating an item with markplane_add, edit the markdown body directly to add context, notes, acceptance criteria, or any free-form detail. The MCP tools manage structure (IDs, frontmatter, cross-refs); you manage content.

Always run markplane_sync after creating or updating items so the context summaries reflect the latest state.

Examples of what's worth capturing

Use judgment about what deserves a structured item vs. what belongs in the daily log or conversation. Some examples:

• "We need to migrate the database before launch" -> task with priority and a dependency on the schema design task
• "We decided to use PostgreSQL over SQLite for the API" -> note (type: decision) with reasoning in the body
• "The deployment pipeline should do X, then Y, then Z" -> plan with steps and linked tasks for each phase
• "Ship the v2 API by end of Q2" -> epic with related tasks underneath
• "The client's timezone is UTC+9, they prefer async communication" -> this probably belongs in MEMORY.md, not Markplane — it's a preference, not a task

Not everything needs to be an item. If it's actionable, trackable, or a decision worth referencing later, it's a good candidate. If it's ephemeral context or a quick answer, let it live in the conversation.

Item types

Prefix	Type	Directory	Use for
EPIC	Epic	roadmap/	Long-term goals and themes
TASK	Task	backlog/	Action items, to-dos, bugs
PLAN	Plan	plans/	Multi-step implementation approaches
NOTE	Note	notes/	Decisions, research, reference info

Cross-references

Use [[TASK-xxxxx]] wiki-style links to connect related items. Use depends_on and blocks in frontmatter for dependency tracking (managed via markplane_link).

Coexistence with OpenClaw memory

Markplane complements, not replaces, OpenClaw's existing memory:

• MEMORY.md — identity, preferences, curated long-term context (keep using)
• memory/YYYY-MM-DD.md — daily logs, session transcripts (keep using)
• .markplane/ — structured tasks, decisions, project state (Markplane)

Daily logs capture what happened. Markplane captures what needs to happen.