> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gentic.chat/llms.txt
> Use this file to discover all available pages before exploring further.

# Create, Assign, and Manage AI Coding Issues in Gentic

> Issues are the unit of work in Gentic. Assign an issue to an AI agent, and the agent writes code and opens a pull request on your behalf.

An issue represents a discrete coding task — a feature, bug fix, piece of feedback, or idea — that you hand off to an AI agent. Once you move an issue from **Draft** to **Todo**, Gentic queues an agent run: the agent clones your repository, follows the instructions in your prompt, commits changes, and opens a pull request. Every issue tracks its own status, agent conversation, and PR link in one place.

## Create an Issue

<Steps>
  <Step title="Open the new issue form">
    From **Home**, click **New issue** in the top-right corner.
  </Step>

  <Step title="Select a project">
    Choose the project (GitHub repository) the agent should work in. If you have no projects yet, [create one in Settings](/web/projects) first.
  </Step>

  <Step title="Fill in the issue fields">
    Complete the **Title**, **Type**, **Agent**, **Prompt**, and **Status** fields (see [Field reference](#field-reference) below).
  </Step>

  <Step title="Submit">
    Click **Create issue**. If you set the status to **Todo**, the agent run is queued immediately.
  </Step>
</Steps>

<Tip>
  Write a detailed prompt with explicit acceptance criteria — for example, "Add a `/health` endpoint that returns `{ status: 'ok' }` with HTTP 200. Include a Jest test." The more context you give the agent, the fewer revision cycles you need.
</Tip>

## Field Reference

<ParamField body="project" type="string" required>
  The project whose GitHub repository the agent will clone and work in. You cannot change the project after the issue is created.
</ParamField>

<ParamField body="title" type="string" required>
  A short description of the work to be done, shown on the Home issues list and on the GitHub pull request. Maximum 160 characters.
</ParamField>

<ParamField body="type" type="feature | bug | feedback | idea">
  Categorises the issue so you can filter and visually distinguish different kinds of work at a glance.

  | Value      | When to use                             |
  | ---------- | --------------------------------------- |
  | `feature`  | New functionality or enhancements       |
  | `bug`      | Defects and regressions to fix          |
  | `feedback` | Changes driven by user or team feedback |
  | `idea`     | Exploratory or speculative work         |
</ParamField>

<ParamField body="agent" type="Claude Code | Codex">
  The AI agent that will execute the issue. Choose **Claude Code** (Anthropic) or **Codex** (OpenAI) depending on your preference or the nature of the task. You can change the agent when editing an issue.
</ParamField>

<ParamField body="prompt" type="text">
  Detailed instructions for the agent. Include background context, step-by-step requirements, acceptance criteria, links to relevant files, and any constraints. The prompt is passed directly to the agent at the start of each run.
</ParamField>

<ParamField body="status" type="enum">
  The workflow status of the issue. Setting the initial status to **Todo** queues the agent immediately on creation. See the [Issue Statuses reference](/web/issue-statuses) for the full list and what each status means.
</ParamField>

## Issue Workflow

### Triggering an Agent Run

Moving an issue's status from **Draft** to **Todo** is the action that queues the agent run. You can do this at creation time by setting the initial status to **Todo**, or later by updating the status on the issue detail page.

### Sending Messages While the Agent Runs

While an agent run is active (status: **In progress** or **Waiting for input**), a chat interface is available on the issue detail page. Type a message and send it to communicate with the agent mid-run — for example, to clarify a requirement or unblock a decision.

### Re-queuing a Completed or Cancelled Issue

If an issue has reached **Completed** or **Cancelled** status and you want the agent to do more work, send a new message from the chat interface. Gentic re-queues the agent run and the agent resumes with the full prior conversation context intact.

### Tracking the Pull Request

Once the agent opens a pull request on GitHub, the PR URL appears directly on the issue detail page. Click it to review the agent's changes on GitHub.

## Edit an Issue

Open the issue detail page and click **Edit**. You can update the **Title**, **Prompt**, **Agent**, and **Type** fields. The **Project** field cannot be changed after creation.

## Reset the Agent

Resetting an agent clears the current agent session and re-queues the issue as if it were starting fresh. Use this when a run is stuck or you want the agent to restart without any prior conversation history. To reset, open the issue and click **Reset agent**.

<Warning>
  Resetting the agent is irreversible — the existing session and conversation history for that run are permanently cleared.
</Warning>

## Delete an Issue

To permanently remove an issue, open the issue detail page, click **Delete**, and confirm. Deletion redirects you back to the Home issues list.