Using Pi with GitHub Agentic Workflows

Pi is a provider-agnostic coding agent for repository analysis and code changes. GitHub Agentic Workflows runs Pi through GitHub Actions from a Markdown workflow and adds GitHub triggers, sandbox controls, and safe outputs for event-driven, reviewable automation.

Pi v1.0.0 supports native MCP through the gh-aw policy gateway. GitHub CLI proxy mode and tools.cli-proxy: true remain optional transport choices. The default model is copilot/gpt-5.4 when no model is configured.

To select Pi as the AI engine, with inference hosted and billed through a GitHub Copilot subscription, use a copilot/ model. A model without a provider prefix also uses the Copilot backend.

engine:
  id: pi
  model: copilot/gpt-5.4

To authenticate:

• For organization-billed usage, grant copilot-requests: write.
• Otherwise, provide a COPILOT_GITHUB_TOKEN secret containing a fine-grained PAT with Copilot Requests access.

To run Pi with inference hosted and billed through Anthropic, use an anthropic/ model:

engine:
  id: pi
  model: anthropic/claude-sonnet-4.6

To authenticate, provide ANTHROPIC_API_KEY as a GitHub Actions repository secret.

To run Pi with inference hosted and billed through OpenAI, use an openai/ or codex/ model:

engine:
  id: pi
  model: openai/gpt-5.4

To authenticate, provide a CODEX_API_KEY or OPENAI_API_KEY as a GitHub Actions repository secret.

Pi routes openai/ and codex/ models through OpenAI’s Responses API rather than Chat Completions, since OpenAI rejects function tool calls on Chat Completions whenever reasoning is enabled. This matches Pi’s own OpenAI model catalog and requires no additional configuration. The Copilot and Anthropic backends are unaffected and keep using their existing wire protocols.

By default, threat detection for Pi workflows runs on the GitHub Copilot CLI, regardless of Pi’s model provider. Grant copilot-requests: write or provide a COPILOT_GITHUB_TOKEN secret for detection. This credential is separate from the OpenAI or Anthropic key used by the Pi agent.

For any provider, recompile the workflow with gh aw compile and commit the changes to the repository.

---
on:
  schedule: daily

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

engine:
  id: pi
  model: copilot/gpt-5.4

tools:
  github:
    mode: gh-proxy
    toolsets: [default]

safe-outputs:
  create-issue:
    title-prefix: "[status] "
    labels: [report]
    close-older-issues: true
---

# Daily Repository Status

Analyze the repository and create a concise daily status report covering:
- Open issues and their priority
- Recent PR activity
- Upcoming work items

Pi supports native MCP tools and resources, codemode, tool search, provider-prefixed models, and workflow-installed Pi packages. Native MCP connects only to the compiler-managed gateway; provider and GitHub credentials remain outside the sandbox.

Codemode batches tool calls and reduces results before they enter model context. Its classifier and image-generation APIs are available when the configured provider route supports those model types. Custom providers still need a compatible API-proxy target and credentials; enabling codemode does not grant additional credentials or network access.

For Copilot models using the Responses API, gh-aw disables grammar-based custom tools (model.compat.supportsOpenAIGrammarTools: false) so Pi sends codemode as a JSON-schema function tool. Codemode remains enabled; the Copilot compatibility adapter does not accept its custom-tool format. Other provider routes retain their configured compatibility settings.

Project-local executable resources remain untrusted by default. Workflow-installed packages and activation-installed skills use the managed agent directory. Model catalog metadata preserves thinking, vision, token limits, pricing, and cache lifetimes through gateway routing. Workflow system instructions remain separate from user instructions so compaction does not summarize them away.

When AWF /reflect completes model discovery and reports reasoning-effort support for the selected provider and model, Pi uses that capability information instead of model-name exceptions. If discovery is incomplete or reflection does not report reasoning capability data, Pi retains the model catalog or engine.config.model reasoning setting. Missing effort fields alone do not disable reasoning.

Custom executables without the npm SDK must supply complete model metadata (reasoning, input, contextWindow, and maxTokens) in engine.config.model.

The built-in tools.playwright integration uses playwright-cli; omit its mode field. Native tools.web-search, max-continuations, native engine.agent selection, and custom engine.harness scripts remain unsupported. Custom drivers must implement their own policies and cannot opt into the built-in aggregate tool budgets.

google/ models use the Gemini gateway and GEMINI_API_KEY. Other API-compatible providers can use an explicit inference family and endpoint:

engine:
  id: pi
  model: openrouter/anthropic/claude-sonnet-4
  model-provider: openai
  env:
    OPENAI_BASE_URL: https://openrouter.ai/api/v1
    OPENAI_API_KEY: ${{ secrets.OPENROUTER_API_KEY }}
network:
  allowed: [defaults, openrouter.ai]

Compatible provider catalogs, including classifier and image models, retain their native identities but route through the same credential-isolated gateway. Requests must be supported by that gateway and upstream endpoint. Providers requiring different protocols or ambient cloud authentication can run natively only when the agent sandbox is explicitly disabled; this removes credential isolation and is not the recommended deployment mode. Unsupported sandboxed provider routes fail compilation instead of silently switching to Copilot.

engine.config accepts JSON with four optional objects:

engine:
  id: pi
  config: |
    {
      "settings": {"defaultThinkingLevel": "high", "codemode": {"mode": "on"}},
      "session": {"enabled": true, "export": true}
    }
max-tool-calls: 100

Sessions are ephemeral by default. Persistent JSONL sessions are written under /tmp/gh-aw/agent/pi-sessions; restore a previous session artifact there before using resume or fork. HTML exports are stored at /tmp/gh-aw/pi-agent-dir/session.html. Both are secret-redacted before artifact upload.

See the AI engine feature comparison and Pi extensions reference.

Running coding agent CLIs such as Pi directly in GitHub Actions without an adequate security architecture is not recommended. GitHub Agentic Workflows provides sandboxing, credential isolation, scoped permissions, safe outputs, and workflow portability across AI engines.

• Quick start
• Engine reference
• Authentication
• Security architecture
• Gallery