Daily Briefing

Daily Briefing

You produce a comprehensive morning briefing by pulling from all available sources and presenting a prioritized view of the day. Output as a Canvas.

The user's input is: $ARGUMENTS

Skill Composition

This skill composes existing tools and patterns:

Read-Only Constraint

This skill MUST NEVER write, create, update, or modify data in any external system. All MCP tool calls must be read-only operations. The one exception is second-brain-checkup, which does write to the local KB files (~/second-brain/*.md) as part of its own job -- that's local state Evan owns, not an external system, and it's second-brain-checkup's explicit responsibility, not something this skill does directly.

Autonomy Principle

This skill runs fully autonomously after invocation. No user prompts during data gathering. If a source fails, log it silently and continue.

Workflow

Step 1: Check current time

Run date in the shell to get the current time and day.

If the user said "tomorrow", use days_ahead=1 for calendar. Otherwise use days_ahead=0.

Step 2: Gather data (all in parallel)

Fire all of these simultaneously:

2a. Calendar events

Server: user-google-workspace

calendar_get_events(days_ahead = 0, max_results = 25)

2b. Unread emails

Server: user-google-workspace

gmail_list_messages(query = "is:unread", max_results = 15)

Also check for Gemini meeting notes from the past few days:

gmail_list_messages(
  query = "(from:gemini-notes OR from:meetings-noreply) after:<3 days ago>",
  max_results = 10
)

2c. Slack activity

Server: <YOUR_SLACK_MCP_SERVER>

search_messages(query = "from:me", sort = "timestamp", limit = 20)

Also check saved items:

search_messages(query = "is:saved", limit = 10)

2d. Jira sprint status

Server: user-atlassian

getMyIssues(
  cloudId = "<YOUR_ATLASSIAN_CLOUD_ID>",
  status = "In Progress,To Do"
)

2e. Cross-repo activity

Run in shell:

python3 .cursor/skills/task-dispatch/scripts/scan_repos.py \
  --repos-dir ~/Desktop/ --days 3

2f. Agent replies

Run in shell:

python3 .cursor/skills/task-dispatch/scripts/dispatch_task.py \
  --check-replies --repos-dir ~/Desktop/

2g. Second Brain KB

Invoke second-brain-checkup Steps 1 and 3-6 (reconcile pending-actions, refresh rolling-context, score quick wins) -- skip its Step 7 chat presentation and AskQuestion call. If it surfaces anything genuinely ambiguous, fold a one-line "needs your input: X" note into this briefing's Priority Actions section instead of stopping to ask -- the morning briefing is meant to run fully autonomously, per this skill's own Autonomy Principle.

Step 3: Classify meetings

For each calendar event, classify its type to determine prep depth:

Step 4: Build Canvas

Read the Canvas skill and follow its instructions to create: ~/.cursor/projects/<YOUR_WORKSPACE_PROJECT_ID>/canvases/daily-briefing.canvas.tsx

Canvas sections:

1. Header -- "Daily Briefing -- [Day], [Date]", current time
2. Schedule -- Table of today's meetings with type classification, time, title, and prep flags. Highlight meetings needing deep prep.
3. Agent Replies -- Unread replies from dispatched tasks across repos. Show repo name, task, and agent's message. Flag as needing response.
4. Priority Actions -- Numbered list of suggested priorities with time estimates. Combine the most urgent items from all sources, including Quick Wins from the Second Brain KB (2g) -- these belong near the top since they build morning momentum.
5. Jira Sprint -- Current sprint items grouped by status (In Progress, To Do). Show ticket key, summary, and priority.
6. Slack Threads -- Notable active threads with context. Group by topic, attribute to channel/person.
7. Emails -- Emails worth noting (not spam/newsletters). Show sender, subject, and whether action is needed.
8. Repo Activity -- Recent git commits and agent sessions across projects. Show what was being worked on.
9. Second Brain -- rolling-context.md's "Right Now" and "Top Open Threads" (from 2g), plus any pending actions reconciled as newly resolved or newly stale. Omit if the KB was empty (per its minimum-evidence gate).
10. Coverage Footer -- Which sources contributed data and which failed.

Canvas rules:

• Import only from cursor/canvas. No npm packages, no fetch calls.
• Use useHostTheme() tokens for all colors. No hardcoded hex.
• No emojis, no gradients, no box-shadows.
• Default-export the top-level component.
• Omit sections with no data.

Step 5: Summarize in chat

Alongside the canvas, provide a brief text summary:

• Number of meetings today (with types breakdown)
• Top 3 priority items
• Any unread agent replies that need attention
• Which sources were unavailable

Fallback Behavior

Always report which sources were unavailable.

What NOT to Do

• Don't ask questions during data gathering. Run fully autonomously.
• Don't write to any external system. Read-only.
• Don't include meeting content from past meetings (that's meeting-prep). Just classify and flag which ones need deep prep.
• Don't generate the full prep for each meeting. Just flag type and depth. The user can invoke /meeting-prep for specific meetings.
• Don't skip the repo scan. The cross-agent awareness is what makes this briefing different from just checking calendar and email.