name: standup description: Use when generating standup reports, summarizing recent work across projects, or preparing for status meetings. Searches development memories from all Miller-registered projects. user-invocable: true arguments: "[time_range]"

Standup Summary

Generate narrative standup summaries from development memories across all registered projects.

Arguments

• No argument: Yesterday (all day) + today so far
• 3d, 7d: Last N days through now
• 2025-01-20: Since specific date through now

Process

digraph standup {
  rankdir=TB;
  node [shape=box];

  read_registry [label="Read ~/.miller/project_registry.json"];
  check_exists [label="Project path exists?" shape=diamond];
  scan_memories [label="Scan .memories/YYYY-MM-DD/*.md\nin time range"];
  scan_plans [label="Scan .memories/plans/*.md\nfor completed [x] tasks"];
  aggregate [label="Group by project"];
  summarize [label="Synthesize narrative"];

  read_registry -> check_exists;
  check_exists -> scan_memories [label="yes"];
  check_exists -> read_registry [label="no, skip"];
  scan_memories -> scan_plans;
  scan_plans -> aggregate;
  aggregate -> summarize;
}

Implementation

1. Read Project Registry

import json
from pathlib import Path

registry_path = Path.home() / ".miller" / "project_registry.json"
if registry_path.exists():
    data = json.loads(registry_path.read_text())
    projects = data.get("projects", {}).values()

2. Calculate Time Range

from datetime import datetime, timedelta

def parse_time_range(arg: str | None) -> datetime:
    """Return start datetime for the range."""
    now = datetime.now()

    if arg is None:
        # Yesterday 00:00
        yesterday = now.date() - timedelta(days=1)
        return datetime.combine(yesterday, datetime.min.time())

    if arg.endswith("d"):
        days = int(arg[:-1])
        start_date = now.date() - timedelta(days=days)
        return datetime.combine(start_date, datetime.min.time())

    # Parse as date: 2025-01-20
    return datetime.strptime(arg, "%Y-%m-%d")

3. Gather Memories

For each project, scan .memories/ for:

Checkpoints (in date directories):

memories_dir = Path(project["path"]) / ".memories"
for date_dir in memories_dir.iterdir():
    if date_dir.is_dir() and date_in_range(date_dir.name):
        for md_file in date_dir.glob("*.md"):
            # Parse YAML frontmatter + content

Completed plan tasks:

plans_dir = memories_dir / "plans"
for plan_file in plans_dir.glob("*.md"):
    content = plan_file.read_text()
    completed = re.findall(r'- \[x\] (.+)', content, re.IGNORECASE)

4. Summarize

Format findings by project, then synthesize into 2-4 paragraph narrative:

• Lead with significant accomplishments
• Group related work
• Use past tense ("Fixed", "Implemented")
• Keep concise - this is for standups, not documentation

Output Example

Yesterday and today I focused on two main areas:

In the Miller project, I designed and implemented a cross-project standup feature. This involved creating a user-level project registry at ~/.miller/ that auto-registers projects on startup, and a /standup skill that aggregates memories across all projects.

In the webapp project, I fixed the authentication token refresh race condition that was causing intermittent 401 errors.

Edge Cases

• Project path gone: Skip silently, project may have been deleted
• No .memories/ directory: Skip, project just doesn't use memories
• No activity found: Report "No development activity found for this period"
• Registry doesn't exist: Report "No projects registered yet. Projects auto-register when opened with Miller."