> For the complete documentation index, see [llms.txt](https://docs.paradime.io/app-help/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.paradime.io/app-help/guides-new/programmable-agents/github-issues-backlog-agent.md). # GitHub Issues Backlog Agent Automate your analytics engineering backlog with a native DinoAI agent that reads every open GitHub issue labelled **Agent Ready**, implements the required dbt™ models with tests and YAML docs, opens a PR per issue and posts the link to Slack. No custom Python, no SDK glue, no orchestrator. The agent is a single YAML file in your repo, triggered on a schedule by Bolt. {% hint style="info" icon="compass" %} **Before You Start** **Paradime** * Your Paradime API endpoint, API key, and API secret. — You can generate under [Workspce Settings → API](/app-help/developers/generate-api-keys-legacy.md). Make sure to have `DinoAI agent API` capabilities. Requires Admin access to generate API keys. **GitHub** * Write access to the repository you want to review PRs on * Ability to add repository secrets and create GitHub Actions workflows **Integrations** The following must already be connected in Paradime: * [**Slack**](/app-help/integrations/slack.md) — the agent posts reviews to `#pr-reviews` via `post_slack_message` **Labelling** * Issues you want automated must be **open** and carry the label **`agent-ready`**. {% endhint %} ### What You'll Build By the end of this guide you'll have: * A `github-backlog-agent` DinoAI agent YAML committed under `.dinoai/agents/` * A Bolt schedule that triggers the agent on a cron cadence using the **Run Paradime DinoAI Agent** command — no scripts to maintain #### What the Agent Does Per Issue Once triggered, the agent works through the backlog and, for each open **Agent Ready** issue: ``` 1. Reads the issue 2. Identifies source tables and implied metrics 3. Explores the repo to understand existing models and sources 4. Drafts staging (and mart) dbt™ models with tests and YAML docs 5. Branches + commits + pushes (gh--…) 6. Opens a PR (Closes #) and posts "✅ Done: " to Slack ``` The agent never invents a column that isn't in the source. When an issue is ambiguous it surfaces explicit **Open questions** in the PR description rather than guessing. #### Architecture Overview ```mermaid flowchart LR A[Bolt schedule\nRun Paradime DinoAI Agent] B[Paradime API\ntriggerDinoaiAgentRun] C[Agent pod\ngithub-backlog-agent] D[GitHub\nissues + PRs] E[Slack\nstatus updates] A --> B --> C C -->|run_terminal_command| D C --> E ``` ### How It Works Bolt runs a single `paradime dinoai` command on your chosen cadence. That spins up one agent pod with a task message telling it to process the **Agent Ready** backlog. The agent list issues, create branch, implement each issue, push, and open a PR - then posts to Slack. Bolt blocks until the agent reaches a terminal state and inherits its success or failure, and the agent's messages stream into the Bolt run logs. {% stepper %} {% step %} #### Create the Agent YAML Create the following file in your repository at `.dinoai/agents/github-backlog-agent.yml`. This defines the agent's role, goal, tools, and Slack output channel. {% code title=".dinoai/agents/github-backlog-agent.yml" lineNumbers="true" %} ```yaml name: github-backlog-agent version: 1 role: > Analytics Engineer turning GitHub issues into well-scoped dbt model PRs, using the gh CLI. goal: > For each open GitHub issue labelled "agent-ready" — or the specific issue number in the trigger message: read it, implement the dbt models it needs (staging, plus a mart where appropriate) with tests and YAML docs on a new branch, open a PR that closes the issue, and post the PR link to Slack. One branch and one PR per issue. Skip any issue that already has an open PR. backstory: > You never invent a column that isn't in the source. When an issue is ambiguous, you flag it under "Open questions" in the PR instead of guessing. You follow the project's dinorules conventions and implement everything yourself — no delegation. tools: mode: allowlist list: - read_file - write_file - replace_in_file - search_files_and_directories - ripgrep_search - run_sql_query - run_terminal_command - post_slack_message slack: channel: "#analytics-eng" ``` {% endcode %} {% endstep %} {% step %} #### Commit the Agent to Your Repo Commit the YAML to the git branch your Bolt schedule will run against (e.g. `main`): ```bash git add .dinoai/agents/github-backlog-agent.yml git commit -m "Add github-backlog-agent" git push ``` The agent now appears in the **Agent Name** dropdown when you configure a Bolt command - the list is populated from `.dinoai/agents/*.yml` on the configured branch. {% endstep %} {% step %} #### Add Paradime Credentials to the Bolt Schedule The `paradime dinoai` command authenticates with your Paradime API credentials. Store them on the schedule (or at the workspace level under **Workspace Settings → Environment Variables** to share across all schedules): | Variable | Value | | ----------------------- | ------------------------------ | | `PARADIME_API_ENDPOINT` | Your Paradime GraphQL endpoint | | `PARADIME_API_KEY` | Your Paradime API key | | `PARADIME_API_SECRET` | Your Paradime API secret | {% hint style="info" %} Make sure the API key has **DinoAI Agent** capabilities enabled (Workspace Settings → API). These are the only secrets you manage. {% endhint %} {% endstep %} {% step %} #### Create the Bolt Schedule Go to **Bolt → Schedules** and click **New Schedule**. Name it something like `GitHub Backlog Agent`, then add a command using the **Run Paradime DinoAI Agent** command type.
**Option 1: From the Bolt UI** Click **Add command** → **Run Paradime DinoAI Agent**, then fill in: * **Agent Name** — pick `github-backlog-agent` from the dropdown. * **Task** — the agent's opening message. For a full backlog sweep: {% code overflow="wrap" %} ``` Find every open GitHub issue labelled agent-ready. For each one: read it, implement the required dbt models with tests and YAML docs on a new branch, open a PR with Closes #. ``` {% endcode %} **Save Command**, then set the trigger type (cron, on-merge, on-completion) and deploy. **Option 2 Schedules as Code** If you manage Bolt schedules with YAML, add the same command directly. The UI just writes a `paradime dinoai` invocation under `commands:`: ```yaml schedules: - name: github backlog agent slug: github-backlog-agent-l3m4n5 description: "Sweep open Agent Ready GitHub issues into dbt PRs" owner_email: data-team@acme.io environment: production git_branch: main commands: - >- paradime dinoai --agent="github-backlog-agent" --message="Find every open GitHub issue labelled agent-ready. For each one: read it, implement the required dbt models with tests and YAML docs on a new branch, open a PR with Closes #." schedule: "0 9 * * 1-5" timezone: UTC ``` | UI field | CLI flag | Notes | | ------------- | -------------------- | ---------------------------------------------------------------------------- | | Agent Name | `--agent=` | Must match an existing `.dinoai/agents/.yml` on the schedule's branch. | | Task | `--message=""` | The agent's opening user message. Use a single quoted string. | | {% endstep %} | | | {% step %} #### Choose a Schedule Frequency Pick the cadence that matches how often your team labels new issues as **Agent Ready**: | Cadence | Cron expression | When it runs | | ------------------------- | --------------- | ---------------------------- | | Every day at 9 AM | `0 9 * * *` | Monday–Sunday, 9:00 AM | | Weekdays at 9 AM | `0 9 * * 1-5` | Monday–Friday, 9:00 AM | | Once a week (Monday 9 AM) | `0 9 * * 1` | Every Monday, 9:00 AM | | Twice a week (Mon & Thu) | `0 9 * * 1,4` | Monday and Thursday, 9:00 AM | | Every 6 hours | `0 */6 * * *` | 12 AM, 6 AM, 12 PM, 6 PM | {% hint style="info" %} For most teams, **weekdays at 9 AM** (`0 9 * * 1-5`) is a good default. If there are no Agent Ready issues, the agent reports an empty backlog and the run completes cleanly. {% endhint %} {% endstep %} {% endstepper %} ### File Structure The only file you author lives in your repo: ``` your-repo/ ├── dbt_project.yml └── .dinoai/ └── agents/ └── github-backlog-agent.yml ``` Everything else — triggering, polling, notifications — is handled natively by Bolt and the agent runtime. ### Related Docs * [**Programmable Agents — Quick Start** — getting started with DinoAI agents](/app-help/products/dino-ai/programmable-agents/quick-start.md) * [**Programmable Agents — YAML Configuration** — full reference for agent config options](/app-help/products/dino-ai/programmable-agents/yaml-configuration.md) * [**Programmable Agents — Tools Reference** — all available tools including `read_file`, `get_linear_issue`, and `post_slack_message`](/app-help/products/dino-ai/programmable-agents/tools-reference.md) * [**Running from Bolt** — trigger agents from a Bolt schedule (UI and YAML)\`](/app-help/products/bolt.md) * [**Slack Integration** — connecting Slack to Paradime](/app-help/integrations/slack.md) * [**Paradime API & Credentials** — where to find your API endpoint, key, and secret](/app-help/developers/generate-api-keys-legacy.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.paradime.io/app-help/guides-new/programmable-agents/github-issues-backlog-agent.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.