[claude-code-user-docs-review] Claude Code User Documentation Review - 2026-04-05 #24738
Closed
Replies: 1 comment
-
|
This discussion has been marked as outdated by Claude Code User Documentation Review. A newer discussion is available at Discussion #24891. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
This is a daily automated review of the gh-aw documentation from the perspective of a developer who uses Claude Code as their primary AI assistant and does not use GitHub Copilot. The goal is to identify blockers, gaps, and assumptions that prevent successful adoption by non-Copilot users.
Persona Context
Reviewing as a developer who:
Files reviewed:
README.md,docs/src/content/docs/setup/quick-start.mdx,docs/src/content/docs/introduction/how-they-work.mdx,docs/src/content/docs/reference/tools.md,docs/src/content/docs/setup/cli.md,docs/src/content/docs/reference/auth.mdx,docs/src/content/docs/reference/engines.mdExecutive Summary
Claude Code users can successfully adopt gh-aw. The core documentation — Quick Start, Authentication reference, Engines reference, and CLI reference — all treat Claude as a first-class engine alongside Copilot. The interactive wizard (
gh aw add-wizard) prompts for engine selection and handles Claude API key setup. No critical blockers exist for non-Copilot users.Key Finding: The documentation is well-structured for multi-engine use. The main friction points are cosmetic: a few persistent stale URLs, Gemini being inconsistently listed across pages, and the Quick Start ordering that lists Copilot's token first. None prevent adoption.
Trend: Score remains 8/10. One new minor issue found (inconsistent Gemini mention in
how-they-work.mdx). Claude engine workflow examples grew from 35 → 37 since the last run — a positive signal.Question 1: Onboarding Experience
Can a Claude Code user understand and get started with gh-aw?
Yes, with minor friction. The Quick Start prerequisites (
quick-start.mdx, line 29) explicitly list:The
add-wizardstep (Step 2) correctly states that it will "Select an AI Engine — Choose between Copilot, Claude, or Codex" and documents all three secrets (COPILOT_GITHUB_TOKEN,ANTHROPIC_API_KEY,OPENAI_API_KEY).Specific Issues Found:
auth.mdxandengines.mdbut NOT in prerequisites oradd-wizardstep description (quick-start.mdxline 29 and 68). This inconsistency means a Gemini user would be confused, though Claude users are unaffected.Recommended Fixes:
GitHub Copilot, Anthropic Claude, OpenAI Codex, or Google GeminiQuestion 2: Inaccessible Features for Non-Copilot Users
What features or steps don't work without Copilot?
Features That Require Copilot exclusively:
max-continuations(autopilot mode with multiple consecutive runs) —engines.mdfeature table, Copilot-onlyengine.agent(custom agent files in.github/agents/) — Copilot-only, documented inengines.mdFeatures That Work Without Copilot (Engine-Agnostic):
edit,github,bash,web-fetch,playwright,cache-memory,repo-memory,qmd,agentic-workflowsmcp-servers:)max-turns(Claude-only per feature table, but documented clearly)gh aw compile,gh aw run,gh aw trial,gh aw logs,gh aw audit— all engine-agnostic CLI toolsMissing Documentation:
engines.mdshowsmax-turnsis Claude-only but doesn't explain why this is valuable or how to use it.engine.agentCopilot custom agent feature has a dedicated reference page — there's no equivalent Claude-specific configuration guide.Question 3: Documentation Gaps and Assumptions
Where does the documentation assume Copilot usage?
Copilot-as-Default Assumption:
how-they-work.mdxline 26: "Workflows support GitHub Copilot (default), Claude by Anthropic, and Codex." The "(default)" label is accurate but not explained. A Claude user wondering "what happens if I use a workflow withoutengine:specified?" could be confused until they findengines.mdwhich explains the default more explicitly.cli.mdinitcommand: runs in "interactive mode for engine selection" — good, no Copilot assumption.Copilot-Centric Language Found In:
how-they-work.mdxline 26 — Copilot listed first with "(default)", Gemini omitted entirely from this introductory pageauth.mdxlines 17–18 — Copilot listed first in the engine-secret mapping table (ordering preference, not a functional issue)engines.mdCopilot Custom Configuration section — has a dedicated subsection with a reference page; Claude has no equivalentMissing Alternative Instructions:
engines.mdandauth.mdxbut no dedicated onboarding path.engine:field use Copilot by default — discoverable inengines.mdbut not in Quick Start or How They Work.Severity-Categorized Findings
🚫 Critical Blockers — Score: 0/10
None found. Claude engine is fully supported with clear authentication documentation, CLI support, and workflow examples.
Obstacle 1: README gives no indication that non-Copilot engines are supported
Impact: First-time discovery friction — developers finding the project via GitHub may assume Copilot is required before reading the docs.
Current State:
README.mdOverview section describes gh-aw generically without mentioning any AI engine. The only AI reference is an implicit one through the Guardrails section.Why It's Problematic: A Claude Code user evaluating whether to invest time in this tool gets no signal from the main project page that Claude is supported. They must click through to documentation to find this out.
Suggested Fix: Add to README Overview: "Supports multiple AI coding agents including GitHub Copilot (default), Claude by Anthropic, OpenAI Codex, and Google Gemini — bring your own API key."
Affected Files:
README.md(lines 33–35)💡 Minor Confusion Points — Score: 6/10
auth.mdxline 103 —ANTHROPIC_API_KEYsetup URL is(platform.claude.com/redacted) (a docs page) rather than the direct API key creation page (https://console.anthropic.com/settings/keys`). Minor UX friction./gh-aw/guides/web-search/) — based on previous reviews, only showsengine: copilotexample, no Claude MCP web search example. (Not re-read today but unresolved from previous runs.)quick-start.mdxline 29 and 68 — Gemini missing from Quick Start prerequisites andadd-wizardstep description, despite being documented inauth.mdxandengines.md.engines.mdline 18 — Claude entry links tohttps://www.anthropic.com/index/claudewhich is potentially a stale URL (Anthropic has reorganized their site).architecture.mdx— AWF firewall diagram labels the agent component as "Copilot CLI" instead of a generic "AI Agent" label, implying the architecture is Copilot-specific.how-they-work.mdxline 26 — AI Engines section lists "GitHub Copilot (default), Claude by Anthropic, and Codex" — omits Gemini despite it being a supported 4th engine inengines.mdandauth.mdx.Engine Comparison Analysis
Available Engines
engine:valuecopilot(default)COPILOT_GITHUB_TOKENclaudeANTHROPIC_API_KEYcodexOPENAI_API_KEYgeminiGEMINI_API_KEYDocumentation Quality by Engine
Rating Scale: ⭐⭐⭐⭐⭐ Excellent · ⭐⭐⭐⭐ Good · ⭐⭐⭐ Fair · ⭐⭐ Poor · ⭐ Missing/Minimal
Tool Availability Analysis
Engine-Agnostic Tools (work with any engine):
edit— file editinggithub— GitHub API operations (all toolsets)bash— shell command executionweb-fetch— web content fetchingplaywright— browser automationcache-memory— cross-run persistent memoryrepo-memory— repository-specific memoryqmd— documentation search (experimental)agentic-workflows— workflow introspectionEngine-Specific Notes:
web-search— Codex: opt-in via explicit declaration; Copilot/Claude/Gemini: requires third-party MCP serverengine.agent— Copilot only (custom agent files)max-turns— Claude onlymax-continuations— Copilot onlyUnclear/Undocumented:
tools.mdmentions "Claude: 60s, Codex: 120s" but doesn't state Copilot's or Gemini's defaults.Authentication Requirements
Current Documentation Coverage
ANTHROPIC_API_KEYdocumented with setup steps; URL slightly suboptimal (docs page vs. direct key creation page)OPENAI_API_KEYdocumented with setup steps and direct URLGEMINI_API_KEYdocumented with Google AI Studio URLSecret Names Reference
COPILOT_GITHUB_TOKENANTHROPIC_API_KEYOPENAI_API_KEYGEMINI_API_KEYExample Workflow Analysis
Workflow Count by Engine
Quality Assessment
Claude Examples (37 workflows): Solid representation across diverse use cases including
typist.md,static-analysis-report.md, shared MCP components, and semantic refactoring. These provide practical reference for Claude-specific configurations.Copilot Examples (89 workflows): The largest set. Many use default engine (no explicit
engine:field) which defaults to Copilot.Codex Examples (17 workflows): Adequate for reference.
Gemini Examples (0 workflows): No production examples yet — Gemini appears documentation-complete but example-starved.
Recommended Actions
Priority 1: Quick Wins (Low Effort, High Impact)
platform.claude.com/docs/en/get-startedtohttps://console.anthropic.com/settings/keysindocs/src/content/docs/reference/auth.mdxline 103. Day 7 without a fix.docs/src/content/docs/setup/quick-start.mdxline 29: add "or Google Gemini" to the AI Account bullet.docs/src/content/docs/introduction/how-they-work.mdxline 26: add ", and Gemini by Google" after Codex.anthropic.com/index/claudestill resolves correctly; if not, update to current URL.Priority 2: Meaningful Improvements
Priority 3: Nice-to-Have Enhancements
how-they-work.mdxor as a standalone guide: tradeoffs between Copilot (GitHub-native), Claude (powerful reasoning, pay-per-use), Codex (OpenAI ecosystem), Gemini.max-turnsis Claude-only but has no dedicated explanation of when/why to use it.Positive Findings
What works well for Claude Code users:
add-wizardis engine-neutral — Interactive setup correctly prompts for engine selection and handles all API keyscompile,run,trial,logs,auditall work regardless of engine--engine claudeflag available on key CLI commands —gh aw new,gh aw add,gh aw validate,gh aw initall support engine selectionsecrets bootstrapsupports all engines —gh aw secrets bootstrap --engine claudeworksANTHROPIC_BASE_URLenv var andapi-targetfield documented for Claude, enabling corporate proxy/custom endpoint useConclusion
Can Claude Code Users Successfully Adopt gh-aw?
Answer: Yes, with minor friction.
The documentation presents Claude as a genuine first-class engine. The Quick Start wizard, authentication reference, and engines reference all handle Claude cleanly. A developer with an Anthropic API key can follow the Quick Start guide, select Claude when prompted, and have a working agentic workflow in under 10 minutes.
The friction points are real but minor: a slightly wrong URL for getting an API key, Gemini being inconsistently listed across pages, and the README not signaling multi-engine support to first-time visitors. None of these would block a motivated Claude user.
Overall Assessment Score: 8/10
Trend vs. Previous Run (2026-04-03)
how-they-work.mdxGemini omission)Next Steps
The 5 issues persisting since day 7 (especially the ANTHROPIC_API_KEY URL and Gemini in Quick Start) are small fixes that could be batched into a single documentation PR. The architecture diagram label change and README improvement are slightly larger but straightforward. No urgent action required — the documentation is serving Claude users adequately as-is.
Appendix: Files Reviewed
Complete List of Documentation Files Analyzed
README.mddocs/src/content/docs/setup/quick-start.mdxdocs/src/content/docs/introduction/how-they-work.mdxdocs/src/content/docs/reference/tools.mddocs/src/content/docs/setup/cli.mddocs/src/content/docs/reference/auth.mdxdocs/src/content/docs/reference/engines.mdReport Generated: §24002231502
Workflow: claude-code-user-docs-review
Engine Used: claude (eating our own dog food)
Beta Was this translation helpful? Give feedback.
All reactions