FeaturedDev & IT Ops#Agent#Codex#GitHub#Slack#PR Automation#Sequential Thinking

Agent-Guided Codex PR Dispatcher

Let an Agent inspect a GitHub request, prepare a confirmed Codex task, open a ready PR, and notify Slack when Codex needs input.

Workflow at a glance

The full canvas, before you import it

Click any node to see its config.

#Agent#Codex#GitHub#Slack#PR Automation#Sequential Thinking

Click a node to select it — same as the Heym editor; the panel shows its settings.

9 nodes · Free & source-available

Agent-Guided Codex PR Dispatcher

This template turns a plain-language coding request into a controlled Codex run. An Agent extracts the GitHub repository and task, checks repository context with GitHub tool nodes, plans with Sequential Thinking, then hands a structured task to the Codex node. Codex asks for one explicit approval before it starts and opens a ready-to-review pull request when approved.

What this workflow does

  1. userInput receives a coding request that includes a GitHub repository URL
  2. codingAgent translates the request to English, extracts the repository and task, checks GitHub context, and returns repoUrl, taskPrompt, and branchName as JSON
  3. codexTask consumes those fields, starts from main, and uses open_pr publish mode
  4. finalResult returns the Codex summary after completion
  5. If Codex needs approval or clarification, the question branch formats a Slack message with the answer URL

Use cases

  • Turn Slack or portal coding requests into Codex PRs with one approval checkpoint
  • Route repository-specific bug reports into a reproducible Codex task
  • Give an intake Agent GitHub read tools while keeping code changes delegated to Codex
  • Notify the team when Codex needs missing requirements before touching the repo

Setup

Add an LLM credential to codingAgent, a Codex credential and GitHub credential to codexTask, GitHub credentials to the three GitHub tool nodes, and a Slack credential to notifyQuestion. The Agent already has a Sequential Thinking MCP connection. If you want web research, attach your own brave_search_api and website_loader sub-workflows to codingAgent and keep the prompt's "only use search when repository context is not enough" rule.

Notes

Credential and sub-workflow IDs are intentionally blank in this public template. Codex uses open_pr, so switch to draft_pr or diff_only if you want a less aggressive default. The Agent-provided branch name is constrained to lowercase letters, numbers, and hyphens with a 50-character maximum.

How to import this template

  1. 1Click Import → Copy JSON on this page.
  2. 2Open your Heym and navigate to a workflow canvas.
  3. 3PressCmd+V/Ctrl+V— nodes appear instantly.
  4. 4Add your API keys in the node config panels and click Run.
More workflow templates
View all templates
Heym
incident analysis · production AI
Observed across 100s of AI rollouts

AI workflows don't fail because of prompts.
They fail because of orchestration.

symptom · glue code01
5 tools
Scripts, vector DB, approval bot, tracing, browser runner — none of them talk.
symptom · visibility02
~0%
Observable behavior across the stack. Debugging is guesswork.
with heym · one runtime
1 canvas
Agents, RAG, HITL, MCP, traces & evals. Self-hosted. Observable.
AI-Native RuntimeProduction-Grade
github.com/heymrun/heym