project

Parse the first word of the arguments as <subcommand>. Pass remaining words as <args>.

If no subcommand given, show:

Usage: /openltm:project <subcommand>

  init      — seed a new project goal into the LTM context system
               /openltm:project init

  analyze   — analyze project context before starting work
               /openltm:project analyze [topic or task description]

  register  — register or rename a project in the LTM registry
               /openltm:project register [name] [path]

---

init

1 — Verify project is registered:

cat ~/.claude/projects/registry.json

Match cwd. If missing, run /openltm:project register first.

2 — Check for existing goal:

bun --eval "
import { Database } from 'bun:sqlite';
const db = new Database(process.env.LTM_DB_PATH);
const row = db.query(\"SELECT content FROM context_items WHERE project_name=? AND type='goal' LIMIT 1\").get('<project>');
console.log(row ? row.content : '');
"

If a goal exists, show it and ask: "Replace it?"

3 — Ask for the goal:

"What is the current goal for <project>? (1–3 bullets, max 100 chars each)"

4 — Write:

bun --eval "
import { Database } from 'bun:sqlite';
const db = new Database(process.env.LTM_DB_PATH);
db.run(\"DELETE FROM context_items WHERE project_name=? AND type='goal'\", ['<project>']);
db.run('INSERT INTO context_items (project_name, type, content, created_at) VALUES (?, ?, ?, datetime(\"now\"))', ['<project>', 'goal', '<goal>']);
console.log('done');
"

5 — Confirm:

Project **<project>** seeded.
Goal: <goal>

Decisions, progress, and gotchas accumulate automatically via hooks.

Do NOT create context-goals.md or similar files. DB is the source of truth.

---

analyze

This command orchestrates context retrieval in the right order.

1 — Get project context: Call mcp__plugin_openltm_memory__context(project="<project>").

Returns: globals (importance ≥ 4) + scoped (importance ≥ 3).