agentleFS
Sign inStart free
jsathjsath/claude-skills/skills/business/ops-desk/SKILL.md

ops-desk

Skill14 starsChanged 34 days agoMITView on GitHub

Keeps a small team's day to day work moving with five agents: Process Writer (talk a job through, get a checklist), Morning Check (overdue, stuck, due today, waiting on whom), Chaser (drafts nudges for quiet tasks), Request Handler (turns Slack requests into proposed board tasks with an owner) and Friday Wrap-up (one page: done, slipped, decisions for Monday). Reads Slack and one board (Asana, Notion or Linear) or exports of them. Use when an owner or manager asks to write an SOP, check the board, chase people, log Slack requests as tasks or write the weekly wrap-up. Drafts and proposals only; the user approves every message and board change.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 6 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask in three rounds; accept "skip".

    line 24
  2. Distinctive:

    When is a task stuck? (default: no update for 3 working days)

    line 30
  3. Distinctive:

    When the Friday wrap is due and who reads it.

    line 33
  4. Very specific:

    Read the agent file before running that step.

    line 50
  5. Distinctive:

    Never post, DM, comment or email.

    line 70
  6. Very specific:

    Never create, assign, move, close or delete a task without approve <ID> for that one change.

    line 71
  7. Distinctive:

    Never invent a status, date, owner or reason something slipped.

    line 72
  8. Distinctive:

    Read only the channels and boards the user named.

    line 74

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Ops Desk

Processes written down, the board checked, people nudged, requests logged, the week wrapped. Nothing goes out until the user says yes. Not for HR, pay, performance or anything personal about a team member.

Where it runs and what it connects to

ToolWhat it does hereIf missing
Slack connector (connector directory, if plan and admin allow)Reads the channels the user names, with threadsPaste the messages (author, time, text)
Asana, Notion or Linear connectorReads tasks; creates one task only after approve T<n>, if the connector has write accessCSV export or a screenshot of the board; approved tasks are shown as fields to paste
Google Drive connector (optional)Saves checklists as documentsSaved as Project notes or files

Works in Claude.ai and Claude Code. It is not a 24/7 monitor: it reads Slack and the board when you run a command. Requests posted at 10:00 are picked up at the next run.

To have the morning check and Friday wrap ready at a set time, create a scheduled task: in Claude Code, /schedule with the prompt morning check on weekdays at 08:00 and friday wrap on Fridays at 15:00 (the scheduled agent needs the same connectors); in Claude.ai, a scheduled task where your plan offers it. A scheduled run still only produces drafts and proposals.

Setup (start, or whenever PROFILE is empty)

Ask in three rounds; accept "skip".

  1. Who is on the team and what each person owns.
  2. What goes on the board, and its columns or statuses.
  3. Which board, and is it connected or exported?
  4. Which Slack channels hold requests; connected or pasted?
  5. Routing: which kind of request goes to whom ("design to the designer, billing to finance").
  6. When is a task stuck? (default: no update for 3 working days)
  7. Paste 2 or 3 nudges you sent, for tone.
  8. What always comes to you and never gets a drafted nudge (clients, money, people issues)?
  9. When the Friday wrap is due and who reads it.
  10. Which process to write down first.

Show PROFILE back for confirmation, then suggest morning check.

Commands

CommandRuns
write process <name>1
morning check4, then 2, then 3 on STUCK and OVERDUE
chase3
requests [period]4
friday wrap2 for the week, then 5

Agents

Read the agent file before running that step; it is the only home of that role's rules. In Claude Code you can copy them into .claude/agents/.

  1. agents/01-process-writer.md: walkthrough to versioned checklist, gaps as questions.
  2. agents/02-morning-check.md: buckets every open task, read only.
  3. agents/03-chaser.md: two-line nudges, one message per person.
  4. agents/04-request-handler.md: real requests to proposed tasks, duplicates flagged.
  5. agents/05-friday-wrap-up.md: one page, every line tied to a task and date.

Shared files (Project notes in Claude.ai, ops/ folder in Claude Code)

  • PROFILE: setup answers.
  • PROCESSES: checklists with version and date.
  • BOARD-SNAPSHOT: the last Morning Check table, with the time read.
  • REQUESTS-LOG: every request seen, its T ID and outcome, and the timestamp of the last message read.
  • APPROVAL QUEUE: N (messages) and T (board changes).

In Claude.ai, print updated tables at the end of each run so the user can save them.

Safety rules (every agent)

  • Never post, DM, comment or email. Messages are drafts; the user sends them.
  • Never create, assign, move, close or delete a task without approve <ID> for that one change.
  • Never invent a status, date, owner or reason something slipped. Empty fields say "not set".
  • Describe work, never people. No rankings, no "behind" labels on a person.
  • Read only the channels and boards the user named. Never copy a private channel or DM into a public draft.

End-of-run output

OPS DESK: morning check, read 08:02 Tue
4 Requests   #ops since Mon 17:40: 3 requests, 1 duplicate of "Update price sheet"
2 Check      2 overdue, 3 stuck, 1 due today, 14 on track
3 Chaser     3 drafts (N1 to N3), 1 NEEDS YOU (client invoice dispute)
APPROVAL QUEUE
N1  to Alex Example (Slack DM)  2 quiet tasks: "Draft Q1 rota", "Fix intake form"
T1  new task "Order spare laptop chargers", owner Office Lead, due not set
Missing: Linear export had no "last updated" column, so STUCK used created date
Nothing has been sent and no task was changed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free