tasks.yml

name: Task

on:
  workflow_dispatch:
    inputs:
      model:
        description: Model to use (provider/model, e.g. llamacpp/phi-4)
        required: false
        default: ollama_cloud/deepseek-v4-flash
      prompt:
        description: Task for the agent (workflow_dispatch only)
        required: false
        default: ""
      enable_git:
        description: Enable git operations - branch, commit, PR (workflow_dispatch only)
        required: false
        default: "true"
      system_prompt:
        description: Override the direct-prompt system prompt (workflow_dispatch only)
        required: false
        default: ""
  issues:
    types:
      - opened
      - edited
  issue_comment:
    types:
      - created
  pull_request_review_comment:
    types:
      - created

permissions:
  issues: write
  contents: write
  pull-requests: write

jobs:
  infer:
    runs-on: ubuntu-24.04
    timeout-minutes: 25
    steps:
      - uses: actions/create-github-app-token@v3.2.0
        id: app-token
        with:
          client-id: Iv23likOeTBSPtN9mbTj
          private-key: ${{ secrets.OPENTASK_APP_PRIVATE_KEY }}

      - uses: actions/checkout@v7.0.1
        with:
          token: ${{ steps.app-token.outputs.token }}

      - uses: arduino/setup-task@v3.0.0
        with:
          version: 3.x

      - uses: inference-gateway/infer-action@v0.36.0
        with:
          github-token: ${{ steps.app-token.outputs.token }}
          github-app-slug: ${{ steps.app-token.outputs.app-slug }}
          trigger-phrase: "@opentask"
          model: ${{ inputs.model || vars.DEFAULT_MODEL || 'ollama_cloud/deepseek-v4-flash' }}
          llamacpp-api-url: ${{ secrets.LLAMACPP_API_URL }}
          llamacpp-api-key: ${{ secrets.LLAMACPP_API_KEY }}
          direct-prompt: ${{ inputs.prompt }}
          system-prompt-direct: ${{ inputs.system_prompt }}
          enable-git-operations: "${{ inputs.enable_git || 'true' }}"
          debug: "false"
          bash-allow-append: "gh project list( .*)?,gh project field-list( .*)?,gh project item-add( .*)?,gh project item-edit( .*)?,gh issue create( .*)?,gh issue edit( .*)?,gh issue comment( .*)?,gh pr comment( .*)?"
          custom-instructions: |
            When working on a GitHub issue that belongs to a project board, keep the board's status in
            sync as you go. This is best-effort: on any error (no board, no Status field, missing
            Projects permission, network failure) log it and carry on - never abort the task over a
            board update.
            
            Detect membership from the ISSUE side, never by scanning the board - a board can hold
            thousands of items and `gh project item-list` only fetches 30 by default, so scanning
            misses the issue and does not scale:
            
            - `gh issue view <number> --json projectItems` lists the boards this issue is on (and its
              current Status), in one call regardless of board size. Empty list means it is on no
              board - do nothing further.
            - For each board it is on, resolve the project number + id and the Status single-select
              field (with its option ids) via `gh project list --owner <owner> --format json` and
              `gh project field-list <number> --owner <owner> --format json`. No "Status" (or close
              equivalent) field: log a warning and skip that board.
            - Get the item id idempotently with `gh project item-add <number> --owner <owner> --url
              <issue-url> --format json` - it returns the existing item's id in O(1). Do NOT list
              items to find it.
            - Set Status with `gh project item-edit --id <item-id> --project-id <project-id>
              --field-id <field-id> --single-select-option-id <option-id>`: the option closest to
              "In Progress" BEFORE you start changing anything, and the option closest to "Done" when
              your work is complete (body refined, fix pushed, or comment posted).
          plugins: |
            DietrichGebert/ponytail
            ayghri/i-have-adhd
          agents: |
            documentation-agent
          anthropic-api-key: ${{ secrets.ANTHROPIC_API_KEY }}
          openai-api-key: ${{ secrets.OPENAI_API_KEY }}
          google-api-key: ${{ secrets.GOOGLE_API_KEY }}
          deepseek-api-key: ${{ secrets.DEEPSEEK_API_KEY }}
          groq-api-key: ${{ secrets.GROQ_API_KEY }}
          mistral-api-key: ${{ secrets.MISTRAL_API_KEY }}
          cloudflare-api-key: ${{ secrets.CLOUDFLARE_API_KEY }}
          cohere-api-key: ${{ secrets.COHERE_API_KEY }}
          ollama-cloud-api-key: ${{ secrets.OLLAMA_CLOUD_API_KEY }}
          moonshot-api-key: ${{ secrets.MOONSHOT_API_KEY }}
          minimax-api-key: ${{ secrets.MINIMAX_API_KEY }}
          nvidia-api-key: ${{ secrets.NVIDIA_API_KEY }}
          zai-api-key: ${{ secrets.ZAI_API_KEY }}