🔍 Claude Code User Documentation Review - 2026-03-15 #21127
Closed
Replies: 1 comment
-
|
This discussion has been marked as outdated by Claude Code User Documentation Review. A newer discussion is available at Discussion #21296. |
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 gaps, assumptions, and blockers that would prevent adoption by non-Copilot users.
Executive Summary
The gh-aw documentation continues to support Claude Code users reasonably well. The Quick Start, engines reference, and auth docs clearly present Claude as a first-class option alongside Copilot. There are no critical blockers preventing Claude Code users from getting started. However, six persistent documentation issues — all identified in previous runs but still unresolved after 18+ days — create minor friction. The overall adoption score remains stable at 7/10 for the 18th consecutive day.
Key Finding: Claude Code users can successfully adopt gh-aw. The friction points are real but manageable, consisting primarily of Copilot-centric architecture diagram labels and a handful of missing mentions of Gemini.
Persona Context
I reviewed this documentation as a developer who:
Question 1: Onboarding Experience
Can a Claude Code user understand and get started with gh-aw?
Yes. The Quick Start guide (
docs/src/content/docs/setup/quick-start.mdx) immediately mentions Claude alongside Copilot in the prerequisites:Step 2 of the Quick Start explicitly says the
add-wizardcommand will prompt users to "Select an AI Engine — Choose between Copilot, Claude, or Codex." The secret nameANTHROPIC_API_KEYis called out clearly, with a link to the auth reference.The landing page (
index.mdx) says "Use GitHub Copilot, Claude by Anthropic or OpenAI Codex" — no Copilot-first assumption in the hero text.Specific Issues Found:
how-they-work.mdxline 26: States "Workflows support GitHub Copilot (default), Claude by Anthropic, and Codex" — Gemini has been a documented fourth engine for weeks but is still absent here. A Claude Code user reading this file gets an incomplete picture of the engine landscape.architecture.mdxlines 177, 248: The Agent Workflow Firewall diagram and MCP Gateway diagram both label the agent container as "Copilot CLI" and "Copilot CLI + MCP client" respectively. Claude Code users may incorrectly assume the firewall only works with Copilot.Recommended Fixes:
how-they-work.mdxline 26 to include Gemini: "GitHub Copilot (default), Claude by Anthropic, Codex, and Gemini"Question 2: Inaccessible Features for Non-Copilot Users
What features or steps don't work without Copilot?
Features That Require Copilot (by design, correctly documented):
assign-to-agentsafe output — explicitly Copilot-only (assign-to-copilot.mdx), clearly labeled, no confusion.github/agents/) — natively supported only with Copilot; thecopilot-custom-agents.mddoc correctly explains "Copilot supports agent files natively, while other engines (Claude, Codex) inject the markdown body as a prompt"/agent agentic-workflowscommand incustom-agent-for-aw.mdx— requires Copilot Chat, but the page includes a "Self-Contained Debugging (Without Copilot)" section usingdebug.mdthat works with any AIFeatures That Work Without Copilot (engine-agnostic):
github:,bash:,edit:,web-fetch:,playwright:,cache-memory:,repo-memory:,agentic-workflows:create-issue,create-pull-request,add-comment, etc.init,compile,add-wizard,run,status,logs,audit,healthMissing Documentation:
creating-workflows.mdxGitHub Web Interface section reportedly still gated: "If you have access to GitHub Copilot" — no equivalent path shown for Claude Code users creating workflows via GitHub web UIQuestion 3: Documentation Gaps and Assumptions
Where does the documentation assume Copilot usage?
Copilot-Centric Language Found In:
architecture.mdx— AWF diagram (line 177): Agent process box labels only "Copilot CLI"; MCP Gateway diagram (line 248): labels "Copilot CLI + MCP client". Both should say "AI Engine" or list alternatives.architecture.mdx— Network firewall configuration example usesengine: copilotas the only illustration. A Claude user may wonder ifnetwork:config works differently for Claude.how-they-work.mdxline 26 — Lists three engines, missing Gemini despite engines.md and auth.mdx both fully documenting Gemini.cli.mdline 214 —secrets bootstrap --enginedocuments options as(copilot, claude, codex), silently omittinggemini.Missing Alternative Instructions:
common-issues.md(only Copilot CLI Not Found and Copilot License Issues)auth.mdxline 104:ANTHROPIC_API_KEYsetup URL points to(platform.claude.com/redacted) — the correct URL for creating API keys ishttps://console.anthropic.com` (this has been flagged for 18+ days)Severity-Categorized Findings
🚫 Critical Blockers (Score: 0/10)
None. Claude Code users can successfully follow the Quick Start and get a workflow running.
Obstacle 1: ANTHROPIC_API_KEY Setup URL May Be Wrong
Impact: Claude users directed to potentially incorrect URL to create their API key
Current State:
auth.mdxline 104 reads:Quality of Examples
Copilot Examples: Comprehensive — covers all patterns (daily ops, chat ops, issue ops, PR events, multi-repo, etc.)
Claude Examples: Good — 37 real-world workflows covering a wide variety of use cases. Growing steadily (up from 33 on Feb 26, 35 yesterday). Claude users can find relevant examples.
Codex Examples: Adequate — 14 workflows available. Limited but covers major patterns.
Gemini Examples: None — Gemini was added as a supported engine but has zero example workflows in the repository.
Recommended Actions
Priority 1: Critical Documentation Fixes
auth.mdxline 104: Change(platform.claude.com/redacted) tohttps://console.anthropic.com/api-keys`how-they-work.mdxengine list — Line 26 omits Gemini; update to "GitHub Copilot (default), Claude by Anthropic, Codex, and Gemini"Priority 2: Major Improvements
architecture.mdxlines 177 and 248: Replace "Copilot CLI" and "Copilot CLI + MCP client" with engine-agnostic labels like "AI Engine (Copilot/Claude/Codex/Gemini)"secrets bootstrap --enginedocs —cli.mdline 214: Expand(copilot, claude, codex)to includegeminicreating-workflows.mdx: Provide Claude Code / generic alternative to the Copilot-gated web interface sectionPriority 3: Nice-to-Have Enhancements
.github/workflows/to validate Gemini support and provide a referencecommon-issues.md— e.g., "ANTHROPIC_API_KEY not set", "Claude rate limits", "Model not found"architecture.mdxalongside or instead of the Copilot-only snippetPositive Findings
What Works Well for Claude Code Users
add-wizardwizard guides users through Claude engine selection andANTHROPIC_API_KEYsetup interactivelyengines.mdhas comprehensive coverage of all four engines with version pinning examples for eachauth.mdxcovers all four engines with equal detail (modulo the URL issue)assign-to-agentcopilot-custom-agents.mdcorrectly explains Claude behavior: "other engines (Claude, Codex) inject the markdown body as a prompt"custom-agent-for-aw.mdxincludes "Self-Contained Debugging (Without Copilot)" section usingdebug.mdinit,compile,run,status,logs,audit,health) are engine-agnosticConclusion
Can Claude Code Users Successfully Adopt gh-aw?
Answer: Yes, with minor friction
The gh-aw documentation has made deliberate choices to include Claude Code users: the Quick Start names Claude explicitly, the engines reference is comprehensive, authentication is well-documented (modulo one URL), and 37 Claude example workflows exist in the repository. The core adoption path is clear and unobstructed.
The remaining friction is cosmetic and persistent: architecture diagrams still show "Copilot CLI" labels, Gemini is inconsistently mentioned across docs, one authentication URL is potentially wrong, and one web interface creation path is Copilot-gated. None of these are blockers — they're rough edges that reduce confidence and add minor confusion.
Overall Assessment Score: 7/10
Trend
This score has been stable at 7/10 for 18 consecutive days. The same six issues have been reported in every daily run without resolution. This suggests either deliberate product decisions (Copilot-centric diagrams as architectural truth) or low documentation priority for these specific gaps.
Next Steps
The most impactful single change would be fixing the
ANTHROPIC_API_KEYsetup URL inauth.mdx— a one-line edit that removes a real potential confusion for new Claude users. The second most impactful change is updating the architecture diagram labels to be engine-agnostic, as these appear early in the conceptual documentation and set an incorrect framing.Appendix: Files Reviewed
README.mddocs/src/content/docs/index.mdxdocs/src/content/docs/setup/quick-start.mdxdocs/src/content/docs/introduction/how-they-work.mdxdocs/src/content/docs/introduction/architecture.mdxdocs/src/content/docs/reference/tools.mddocs/src/content/docs/reference/engines.mddocs/src/content/docs/reference/auth.mdxdocs/src/content/docs/setup/cli.mddocs/src/content/docs/reference/custom-agent-for-aw.mdxdocs/src/content/docs/reference/copilot-custom-agents.mddocs/src/content/docs/reference/assign-to-copilot.mdx.github/workflows/*.md(sampled engine distribution across 172 files)Report Generated: §23120834763
Workflow: claude-code-user-docs-review
Engine Used: claude (eating our own dog food 🐕)
Days at Score 7/10: 18
Beta Was this translation helpful? Give feedback.
All reactions