Skip to content
cccskills
FreePublished on GitHub

daily-rundown

The owner's morning brief — inbound updates awaiting acknowledgment (via the rundown tool), today's meetings and due-or-overdue reminders from the CRM and the real calendar, the emails worth seeing from Gmail, and the Slack threads that need their eyes — each source folded in when connected and stitched into one short digest. Use for "daily rundown", "what's on today", "morning brief", "catch me up".

The English summary is being prepared. Showing the original description.

Works withSlackGmail

Run score Not tested

See how to install

Files included (1)

  • SKILL.md6.3 KB

SKILL.md (original)

Review the instructions the agent will receive before you install.

Daily Rundown

Runtime skill — a playbook the deployed @context agent runs for its owner, invoked in natural language. Not a coding-agent workflow; those live in .agents/skills/.

A focused today brief: one glanceable surface instead of five apps. It folds in the inbound queue, the CRM, the real calendar, the inbox, and Slack — each source only when it's connected — and stitches them into one short digest. Read-only assembly: it pulls and formats, it never files or sends anything.

The whole value is one surface, so stay ruthless about signal. Surface what needs the owner's eyes today, not a mirror of every inbox and channel.

Procedure

Keep it cheap, and fire it once. Each source is a sub-agent call under its own time budget, and they all share one model — so keep the fan-out small and issue it in a single concurrent batch: put the backbone and best-effort tool calls in one message, all at once, so the brief's wall-clock is the slowest source, not the sum. One tight retrieval per source, no exploratory follow-ups, never retry a source. Don't run the sources in separate turns one after another — that serializes their budgets and is the slow path.

The backbone (the two fast local sources below) always runs and carries the brief on its own; the best-effort sources get skipped the moment they're slow, error, or come back empty (see the skip rule). A complete-but-fast brief beats a perfect one that hangs.

  1. Anchor on now, in the owner's zone. Read the current datetime from your context — it carries the owner's timezone, with the zone label shown (e.g. PDT). "Today" is now → end of that local day, and every due/overdue/"today" judgment is made in that zone, not UTC. Render each time with its zone label so nothing is ambiguous. If the owner asks for a different zone ("rundown in GMT"), frame and render the whole brief in that zone instead. Don't auto-detect travel: the configured zone (or the one they ask for) is the source of truth.

Backbone — always run these two (fast, local):

  1. Inbound queue — call rundown. It surfaces updates others marked done that the owner hasn't acknowledged, and the owner's own reminders the hourly sweep filed in once they fell due. (It marks what it shows as briefed — that is the point of a morning brief; un-acknowledged items still resurface tomorrow.) If it returns nothing, drop the section.

  2. Calendar + due work — call query_crm (one retrieval, ask for both). Pass today's local date (read it from your context, in the owner's zone) in the question so the CRM doesn't have to derive it:

    • Meetings whose starts_at is today (the owner's local day).
    • Reminders that are pending and due today, plus anything overdue (due_at < now, still pending).

    If this read times out or errors, note "CRM skipped" in the close and move on — do not call query_crm again. A retry only burns budget and the queue still carries the brief.

Best-effort add-ons — one call each, only if the source is connected; skip on any trouble:

  1. Calendar (calendar connected): query_calendar for today's events; merge with the CRM meetings (dedupe by title + start time; the calendar wins on times).
  2. Inbox (gmail connected): query_gmail for the handful that actually need the owner — unread and important, or addressed to them and awaiting a reply, from today. Ask the sub-agent for that tight selection; never pull the whole inbox.
  3. Slack (slack connected): query_slack for threads/DMs that mention the owner or look like they're waiting on a reply. "Needs your eyes," not an unread dump.
  4. Order each list by its time column, ascending.

Skip rule (add-ons only). If an add-on source returns an error, a skipped / unavailable note, or is plainly slow, drop its section, note it in one line at the end (e.g. "Slack skipped — slow"), and move on. Do not retry it and never let one source hold up the brief. The backbone (queue + CRM) always stands on its own.

De-dup — one item, one line. The queue and the other sources overlap; prefer the queue copy and don't repeat:

  • A reminder the sweep already filed into the queue (it shows under Awaiting you) is the same follow-up query_crm returns as due/overdue. Show it once, under Awaiting you — don't also list it under Overdue / Today.
  • A Slack @-mention that a teammate already left via the queue shouldn't reappear under Slack. Prefer the queue copy.

Format

Lead in this order, dropping any empty section. Render every clock time with its zone label (e.g. 9:30 AM PDT), in the owner's local zone (or the one they asked for):

  1. Awaiting you — the rundown items, one line each (who / what).
  2. Overdue — past-due pending reminders not already shown above, with how late.
  3. Today — meetings (with start time) and reminders due today, time-ordered.
  4. Inbox — the emails worth seeing (sender · subject · the one-line why), from gmail when connected.
  5. Slack — threads/DMs needing a look (channel or DM · who · the ask), from slack when connected.

Close with a one-line tally, e.g. 2 awaiting · 1 overdue · 4 today · 3 inbox · 2 slack.

Edge cases

  • Quiet day: if every section is empty, say so in one line ("Nothing waiting on you and nothing on the calendar today.") — don't pad it.
  • A disconnected source just vanishes. No gmail / slack / calendar → drop its section silently; never error or apologize for it.
  • Cite the sources you used (the queue via rundown, meetings/reminders via crm, events via calendar, mail via gmail, threads via slack — each when connected). Report empties honestly; never invent items.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

Run this repo's eval suite (python -m evals), diagnose every failing case, and fix in scope (agent instructions, or the eval case when its assertion was wrong) until all cases pass. Use whenever the user wants to run the evals, check for regressions, or fix a red eval suite ("run the evals", "why is a case failing", "make the eval suite green"). Targets the committed cases in evals/cases.py; for ad-hoc probing of one agent's live behavior use improve-agent instead.

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

User-driven loop to change the context agent (or any agent in this repo) — add a source/tool/MCP server, add a capability (knowledge base, memory, sub-agent, scheduled task), refine its instructions, or fix a specific known bug, verifying each change against the live container. Use whenever the user names a concrete change ("add a Google Drive source", "make the daily rundown include the calendar", "the agent keeps doing Y, fix it"). For autonomous hardening with no specific change in mind use improve-agent.

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

Autonomously harden the context agent (or any agent in this repo) — derive probes from the agent's own instructions, run them against the live container, judge the responses and tool calls, and edit the agent until it reliably does what it says it does. Use when the user wants to test, harden, stress-test, or improve an agent's reliability WITHOUT naming a specific change ("make sure context is solid", "harden this agent", "probe it for failure modes"). For a user-directed change use extend-agent; to run the committed eval suite use eval-and-improve.

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

Sweep the knowledge base for what needs attention — stubs with no real content, specs whose status is stale, decisions that were made but never written down, and pages missing from the index — and return a short "what needs attention" list. Read-only, so it reports rather than fixing anything itself. Use for "what in my knowledge base needs attention", "review my knowledge base", "what's stale or unfiled", "knowledge base health check", "tidy up my knowledge".

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

Brief the owner before a meeting or conversation with a person, company, or project — sweep what we already know (crm + knowledge, plus slack, calendar, and gmail when connected), anchor to any upcoming meeting with them, and for people or orgs we don't know, widen to a web search for public background. Use for "prep me for my meeting with X", "brief me on X before the call", "what do we know about X".

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

Process the day into filed context — sweep today's activity (the inbound queue, CRM meetings and reminders, plus calendar, slack, and gmail when connected), extract what's worth keeping, and file it into the CRM and knowledge base. Use for "process today", "log today", "file today", "end-of-day wrap".

The English summary is being prepared. Showing the original description.

agno-agi/context13Updated Jul 10, 2026

See all skills by agno-agi

Report a problem with this skill