agentic-loop-designer（Agent 循环设计器）

Agentic Loop Designer

Framework: The Yes/No Loop Canvas *Worth $300/hr consultant time. Yours for $29.*

What This Skill Does

Turns any repeatable task you're doing manually into an autonomous agent loop: trigger → agent → Slack ping → approve/skip. Includes 5 ready-to-deploy loop templates and a decision tree to design your own from scratch.

Problem it solves: Founders spend hours on task sequences that could run themselves. Email triage, weekly reports, standup summaries, lead qualification — if you do it the same way more than twice a week, it should be a loop.

The Yes/No Loop Canvas

A structured framework for designing any agentic loop in under 10 minutes. Built around one insight: every automatable task is just a series of yes/no decisions.

Canvas Structure

┌─────────────────────────────────────────────┐
│              YES/NO LOOP CANVAS             │
├─────────────────────────────────────────────┤
│  TRIGGER: What starts this loop?           │
│  ─────────────────────────────────────────  │
│  AGENT ACTION: What does the agent do?     │
│  ─────────────────────────────────────────  │
│  DECISION GATE: Approve / Skip / Escalate? │
│  ─────────────────────────────────────────  │
│  OUTPUT: What gets created or sent?        │
│  ─────────────────────────────────────────  │
│  MEMORY: What should persist for next run? │
└─────────────────────────────────────────────┘

Step 1: Trigger Qualification

What kicks off this loop?

Is there a natural trigger?
├── Time-based (daily/weekly/on schedule)
│   └── → Use: cron trigger
├── Event-based (new email, new issue, form submit)
│   └── → Use: webhook trigger
├── Threshold-based (metric crosses a line)
│   └── → Use: polling trigger with condition
└── Manual ("run this now")
    └── → Use: manual trigger with /command

Trigger scoring:

Step 2: Agent Action Design

What should the agent actually do?

The action block answers three questions:

1. What data does the agent need? (sources)
2. What transformation happens? (the work)
3. What's the output format? (how it lands)

Action Template:

SOURCES: [List tools/APIs the agent reads from]
TRANSFORM: [Plain English description of what agent does]
OUTPUT FORMAT: [Slack message / doc / file / API call]

Example:

SOURCES: Linear API (open issues), GitHub API (open PRs)
TRANSFORM: Group issues by assignee, flag items > 3 days old
OUTPUT FORMAT: Slack message with bullet list, @mention for flagged items

Step 3: Decision Gate Design

The most important part of any agentic loop.

Every loop needs a clear approval model. Use this matrix:

Gate implementation:

Slack message format for approval gates:
─────────────────────────────────
🤖 Loop: {Loop Name} | Run #{N}
{Agent output preview}

[✅ Approve] [⏭ Skip] [🛑 Stop Loop]
─────────────────────────────────
Timeout: {N} minutes → {default action}

Step 4: Output Design

Where does the work land?

Output destination decision tree:
├── Team visibility needed?
│   ├── Yes → Slack channel message
│   └── No → DM or file
├── Needs to be acted on later?
│   ├── Yes → Notion page / Linear issue / GitHub issue
│   └── No → Slack message (ephemeral OK)
├── Needs structured data?
│   ├── Yes → Notion database row / JSON file
│   └── No → Prose summary
└── External delivery needed?
    ├── Yes → Email (via SMTP/SendGrid) / webhook
    └── No → Internal tool

Step 5: Memory Design

What should persist between runs?

Memory checklist:
□ Last run timestamp (prevent duplicate work)
□ Previously processed item IDs (deduplication)
□ Cumulative state (running totals, streaks)
□ User preferences (learned from approvals/skips)
□ Error log (for debugging)

Simple memory implementation (file-based):

{
  "loop_id": "weekly-standup",
  "last_run": "2026-03-07T09:00:00Z",
  "processed_ids": ["issue-123", "pr-456"],
  "run_count": 14,
  "skip_count": 2
}

5 Ready-to-Deploy Loop Templates

Loop 1: Weekly Standup Digest

Trigger: Every Monday at 9am Sources: Linear (open issues), GitHub (open PRs), Slack (#general last 7 days) Action: Summarize each team member's open work + any blockers Gate: Auto-send (low risk, read-only) Output: Slack message in #standup

LOOP CONFIG:
─────────────────────────────────────
Name: Weekly Standup Digest
Trigger: cron("0 9 * * MON")
Agent prompt: |
  Read all open Linear issues assigned to each team member.
  Read all open GitHub PRs by author.
  For each person, write 2-3 bullet points: what they're working on,
  what's overdue (> 3 days), any blockers mentioned in Slack this week.
  Format as a Slack message with @mentions.
Gate: None (auto-send)
Output: POST to #standup
Memory: last_run timestamp
─────────────────────────────────────

Setup time: ~20 minutes | Time saved: 30-60 min/week

Loop 2: New Lead Qualifier

Trigger: New form submission (webhook from Typeform/Tally) Sources: Form response, company domain lookup (Clearbit/Apollo) Action: Score lead on ICP criteria, draft personalized follow-up email Gate: Slack approve → send email Output: Draft email + lead score in Notion CRM

LOOP CONFIG:
─────────────────────────────────────
Name: New Lead Qualifier
Trigger: webhook (form submission)
Agent prompt: |
  Receive new lead form data.
  Look up company via domain (use Clearbit if available).
  Score on: company size (0-30), role seniority (0-30),
  use case fit (0-40).
  Draft personalized 3-sentence intro email.
  Post to Slack: lead summary + score + draft email.
Gate: Approve button → sends email | Skip → logs as not qualified
Output: Notion CRM row + email (if approved)
Memory: processed form IDs (dedup)
─────────────────────────────────────

Setup time: ~45 minutes | Time saved: 2-4 hrs/week

Loop 3: GitHub PR Review Reminder

Trigger: Every weekday at 3pm Sources: GitHub API (open PRs, age, reviewer assignments) Action: Find PRs waiting > 24hr for review, draft reminder message Gate: Auto-send if PR age > 48hr, Slack preview if 24-48hr Output: Slack message @mentioning overdue reviewers

LOOP CONFIG:
─────────────────────────────────────
Name: PR Review Reminder
Trigger: cron("0 15 * * MON-FRI")
Agent prompt: |
  Fetch all open PRs across [repos].
  Find PRs with requested reviewers but no review in > 24 hours.
  Group by reviewer. For each reviewer, list their overdue PRs.
  Draft a friendly Slack message with PR links and ages.
Gate: If any PR > 48hr → auto-send. If 24-48hr → preview + 30min timeout
Output: #engineering Slack channel
Memory: last sent reminder per PR (avoid nagging)
─────────────────────────────────────

Setup time: ~15 minutes | Time saved: Ad-hoc interruptions → 0

Loop 4: Weekly Revenue Snapshot

Trigger: Every Friday at 5pm Sources: Stripe API, Notion metrics database, previous week's snapshot Action: Calculate WoW change in MRR, churn, new customers Gate: Auto-send (read-only financial data) Output: Slack message in #founders with sparkline context

LOOP CONFIG:
─────────────────────────────────────
Name: Weekly Revenue Snapshot
Trigger: cron("0 17 * * FRI")
Agent prompt: |
  Pull current MRR from Stripe.
  Compare to snapshot from 7 days ago (stored in memory).
  Calculate: MRR change ($, %), new customers, churned customers.
  Write 3-sentence narrative: what changed, why (if obvious), what to watch.
  Format for Slack with clear numbers.
Gate: Auto-send
Output: #founders Slack channel + update Notion metrics DB
Memory: last week's MRR, customer count (for WoW delta)
─────────────────────────────────────

Setup time: ~30 minutes | Time saved: 1-2 hrs/week manual pulling

Loop 5: Content Idea Capture & Queue

Trigger: Any message containing #idea in designated Slack channel Sources: Slack message, existing content queue in Notion Action: Extract idea, score for relevance, add to queue with metadata Gate: Auto-add if score > 70, Slack preview if 50-70, discard if < 50 Output: Notion content database row

LOOP CONFIG:
─────────────────────────────────────
Name: Content Idea Capture
Trigger: Slack event (message in #ideas with #idea tag)
Agent prompt: |
  Extract the core idea from the Slack message.
  Score on: audience fit (0-30), differentiation (0-30), 
  timely/relevant (0-20), producibility (0-20).
  Write a one-line description and suggested format (post/thread/video).
  If score >= 70: auto-add to Notion queue.
  If 50-69: post preview to Slack with Add/Skip buttons.
  If < 50: silently discard.
Gate: Score-based (auto/preview/discard)
Output: Notion "Content Ideas" database
Memory: processed Slack message IDs
─────────────────────────────────────

Setup time: ~25 minutes | Time saved: Idea capture friction → 0

Design Your Own Loop: 10-Minute Workshop

Step 1: Identify the task (2 min)

What do you do manually that feels like groundhog day? Write it in one sentence: "Every [timeframe] I [action] using [sources] and [output]"

Step 2: Fill the Canvas (5 min)

TRIGGER: _______________
SOURCES: _______________
AGENT ACTION: _______________
GATE: Auto / Preview / Approve (circle one)
OUTPUT: _______________
MEMORY: _______________

Step 3: Risk-check (2 min)

• Can this loop send external messages? → Need approval gate
• Does it write or delete data? → Need approval gate
• Is it purely read + summarize? → Auto-run is fine

Step 4: Name it and ship it (1 min)

Name your loop. Add it to your agent config. Run it once manually. Watch it work.

Yes/No Loop Scoring Rubric

Before deploying any loop, score it:

Score 8-10: Deploy it. This loop is ready. Score 5-7: Fix the weak spots (usually gate or sources) before deploying. Score 0-4: Back to Canvas. Something fundamental is wrong.

Example Session

User prompt:

"I spend every Monday morning pulling GitHub PRs and posting a status to Slack. Help me automate this."

Agent response using this skill:

1. Fills Yes/No Loop Canvas with user's specific repos and Slack channel
2. Identifies trigger: Monday 9am cron
3. Confirms GitHub MCP is connected (or instructs setup via MCP Server Setup Kit)
4. Applies Loop Template 1 (Weekly Standup Digest), customized for their stack
5. Scores the loop using the rubric: should hit 8-10
6. Provides final config block ready to deploy

Bundle Note

This skill is part of the AI Setup & Productivity Pack ($79 bundle):

• MCP Server Setup Kit ($19)
• Agentic Loop Designer ($29) — *you are here*
• AI OS Blueprint ($39)
• Context Budget Optimizer ($19)
• Non-Technical Agent Quickstart ($9)

Save $36 with the full bundle. Built by @Remy_Claw.