[FAQ] Add: billing model, spend controls, and loop cost FAQ entries

[chrizbo]

Adds three new entries to the Costs & Usage section of the FAQ, addressing enterprise questions raised in github/agentic-workflows#495:

1. Are Actions minutes charged in addition to AI costs? — clarifies the dual billing nature (Actions minutes + AI inference are separate charges).
2. How do retries and agent loops affect costs? — explains that gh-aw has no automatic retries, and documents max-turns/max-continuations as the cost levers for reasoning depth.
3. How do I control spend and set budgets? — covers spend controls at the provider level (Anthropic Console, OpenAI platform, GitHub Actions spending limits) and per-repo tracking via dedicated API keys.

These are new entries — no existing FAQ entries covered these topics.

Source issue: github/agentic-workflows#495

Generated by Feedback Question Answerer for issue #495 · ● 988.8K · ◷

[github-actions]

📰 DEVELOPING STORY: Smoke Copilot ARM64 reports was cancelled. Our correspondents are investigating the incident...

[github-actions]

❌ Smoke Temporary ID encountered failures. Check the logs for details.

[github-actions]

❌ Smoke Multi PR failed to create multiple PRs. Check the logs.

[github-actions]

✅ smoke-ci: safeoutputs CLI comment + comment-memory run (25085329040)

Generated by Smoke CI for issue #29023 · ◷

[github-actions]

Comment Memory

CI lights the path
Green checks bloom at dawn
Quiet bots still sing

Note

This comment is managed by comment memory.

It stores persistent context for this thread in the code block at the top of this comment.
Edit only the text inside the backtick fences; workflow metadata and the footer are regenerated automatically.

Learn more about comment memory

Generated by Smoke CI for issue #29023 · ◷

[Copilot]

Pull request overview

Adds new Costs & Usage FAQ entries to clarify billing, loop/retry cost controls, and spend/budget management guidance.

Changes:

• Documented that GitHub Actions billing and AI inference billing are separate cost components.
• Added guidance on controlling loop depth via max-turns and max-continuations.
• Added provider-level spend/budget control recommendations and per-repo tracking guidance.

Show a summary per file

File	Description
docs/src/content/docs/reference/faq.md	Adds three new Costs & Usage FAQ Q&As covering Actions minutes vs inference, loop/retry cost levers, and spend/budget controls.

Copilot's findings

Tip

Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Comments suppressed due to low confidence (3)

docs/src/content/docs/reference/faq.md:559

• “gh-aw has no automatic retry mechanism” is inaccurate as written: the Copilot harness includes automatic retry/backoff for certain partial-execution failures, which can increase both inference usage and wall-clock time within a single workflow run. Suggest clarifying the scope (e.g., no automatic workflow re-run on failure) and noting that engine drivers may retry transient errors inside the run.

gh-aw has no automatic retry mechanism — each workflow trigger produces exactly one run. However, you can control reasoning depth and autopilot continuation, which directly affects how many tokens and how much wall-clock time (Actions minutes) a run consumes:

docs/src/content/docs/reference/faq.md:562

• max-continuations is described here as “multiple consecutive triggered runs,” which reads like multiple separate workflow triggers. Elsewhere (e.g., engines reference) it’s documented as multiple consecutive runs/continuations in Copilot autopilot mode. Recommend aligning wording to avoid implying new workflow runs are triggered.

- `max-turns` (Claude only) — limits the number of AI chat iterations per run
- `max-continuations` (Copilot only) — enables autopilot mode with multiple consecutive triggered runs

docs/src/content/docs/reference/faq.md:577

• This bullet groups Gemini together with Anthropic/OpenAI spend controls (“Anthropic Console or OpenAI platform”), but Gemini keys come from Google AI Studio (per auth docs) and budgeting/quotas are managed in Google’s console/AI Studio rather than Anthropic/OpenAI. Suggest splitting the provider bullets (Claude → Anthropic Console, Codex → OpenAI platform, Gemini → Google AI Studio/Google Cloud) so readers aren’t sent to the wrong place.

- **Actions minutes**: Set an org spending limit in GitHub Billing settings.
- **Claude / Codex / Gemini**: Configure spend limits in the Anthropic Console or OpenAI platform. These apply at the API key or project level.
- **Copilot**: Usage is quota-based (premium requests per month) rather than dollar-metered, so the natural cap is the plan's monthly request quota.

• Files reviewed: 1/1 changed files
• Comments generated: 1

[Copilot]

The answer implies every run always consumes billable GitHub Actions minutes. That’s not true when using self-hosted runners (they don’t consume Actions minutes, though they still use compute). Consider rephrasing to clarify this applies to GitHub-hosted runners / GitHub-hosted minutes billing, and optionally mention self-hosted runner cost is out-of-band.

This issue also appears in the following locations of the same file:

• line 558
• line 560
• line 575