orient

Topic-focused session orientation across graph, knowledge, journal, and research.

Requirements

Optional — MCP Memory Server enables graph search. Works without it using knowledge files and research index only.

What It Does

When you invoke /orient:

1. Parses your input into topic keywords, entity types, and/or time qualifiers
2. Searches the MCP Memory Server graph for matching entities (if connected)
3. Scans knowledge files and journal entries for relevant sections
4. Checks the research index for related materials
5. Produces a focused briefing with suggested follow-up queries

Output Structure

Section	What it covers
Graph	Matching entities from MCP Memory Server with top observations
Knowledge	Relevant excerpts from knowledge files (section-level, not full files)
Journal	Matching entries from journal files
Research	Listed research files (not read — offered for follow-up)
Suggested Queries	Follow-up graph queries to go deeper

Empty sections are omitted automatically.

Usage

Orient on a topic

/orient membrain
/orient data engineering

Searches graph, knowledge, and research for everything related to the topic.

Orient on entity types

/orient desires
/orient desires and tensions

Searches the graph for specific entity types (Desire, Tension, Evidence, Project, etc.).

Orient on recent activity

/orient recent
/orient recent evidence
/orient last week

Shows recent journal entries, optionally filtered to a specific topic.

Compound queries

/orient recent membrain
/orient the streaming migration project

Combines time qualifiers and topic keywords for targeted results.

Graceful Degradation

If MCP Memory Server is not connected, the graph section is skipped and noted. Knowledge files and research still work — you always get something useful.

Scope Controls

Strict limits keep orientation fast:

• Max 3 search_nodes calls, 2 open_nodes calls
• Max 10 graph entities shown, 3 observations each
• Max 3 knowledge files read, 3 journal entries
• Max 5 research entries listed (not read)
• If >15 graph results, asks you to narrow the query

Output stays between ~800-2000 tokens.

note

This plugin assumes a specific project structure (knowledge/, knowledge/journal/, research/INDEX.md) and MCP Memory Server entity types. Fork and adapt the SKILL.md if your project uses different conventions.

Design Notes

Why topic-focused?

Session orientation is all-or-nothing — you either load everything or manually query. /orient fills the gap: say /orient membrain and get only what's relevant across all knowledge sources.

Why strict caps?

Orientation should be fast. Dumping 50 graph entities defeats the purpose. The caps keep output small enough to orient, not enough to overwhelm.