Add Agent Canvas critic configuration docs

[xingyaoww]

Summary

• Add an Agent Canvas guide for configuring the experimental critic feature.
• Document when to leave Critic API Key empty, when to use the OpenHands Provider LLM Key, and how iterative refinement works.
• Link the new guide from the Agent Canvas navigation and settings overview.

Validation

• python3 -m json.tool docs.json
• git diff --check
• Verified referenced internal docs pages exist locally

Note: mintlify broken-links could not run locally because Mintlify CLI does not support the current Node.js 25.9.0 runtime.

[mintlify]

Preview deployment for your docs. Learn more about Mintlify Previews.

Project	Status	Preview	Updated (UTC)
all-hands-ai	🟢 Ready	View Preview	Jun 8, 2026, 5:40 PM

💡 Tip: Enable Workflows to automatically generate PRs for you.

[all-hands-bot]

PR Review: Add Agent Canvas critic configuration docs

Overall: This PR is well-structured and follows the documentation style guide. The new critic guide is clear, comprehensive, and appropriately covers prerequisites, configuration steps, advanced options, and troubleshooting.

Minor Suggestions (non-blocking):

1. Line 17 - Prefer wording: The word "Prefer" in "Prefer the OpenHands LLM provider" sounds optional. Consider "Use the OpenHands LLM provider" to better reflect that it provides automatic critic handling without additional setup.

2. Line 95 - Ambiguous term: "Use the OpenHands agent path" could be clearer as "Use an OpenHands agent conversation" or "Use the OpenHands agent type" to avoid confusion.

3. Optional enhancement: Consider adding a brief note about what a "success likelihood score" looks like (e.g., range 0-1) in the introduction.

What's Working Well:

• Clear structure following the style guide
• Proper use of Mintlify components (Warning, Note)
• All internal links use absolute paths
• Numbered steps for procedures, bullets for troubleshooting
• Title Case for headers
• All referenced internal docs exist
• docs.json syntax is valid
• Helpful update to customize-and-settings.mdx description

Verification:

• docs.json is valid JSON
• All linked pages exist
• Navigation entry properly added to docs.json

LGTM - Approve for merge with optional suggestions.

[all-hands-bot]

Consider "Use the OpenHands LLM provider" instead of "Prefer" to better convey automatic critic handling.

[all-hands-bot]

Consider rephrasing to "Use an OpenHands agent conversation" or "Use the OpenHands agent type" instead of "agent path".

[all-hands-bot]

✅ Review complete.

This review was performed through OpenHands Cloud Automation. You can log in and view the conversation here.

[all-hands-bot]

PR Review: Add Agent Canvas critic configuration docs

Taste Rating: 🟢 Good taste

This PR adds a well-structured documentation page for the experimental Agent Canvas critic feature. The changes follow the existing docs style consistently.

Verification:

• ✅ docs.json: Valid JSON, navigation entry properly added
• ✅ openhands/usage/agent-canvas/customize-and-settings.mdx: Updated description and Related Guides link
• ✅ New critic.mdx: Comprehensive coverage of prerequisites, configuration, advanced options, and troubleshooting
• ✅ All linked internal docs pages verified to exist (customize-and-settings.mdx, llm-profiles.mdx)
• ✅ Proper use of Mintlify components (Warning, Note)
• ✅ Consistent formatting: Title Case headers, numbered steps for procedures, bullets for troubleshooting

Notes:

The previous review (by all-hands-bot) raised good suggestions about wording consistency. Those are optional improvements worth considering but not blocking.

Risk Assessment:

• [Overall PR] ⚠️ Risk Assessment: 🟢 LOW
• This is purely documentation changes with no code, no dependencies, and no security implications
• The changes cannot break existing functionality

VERDICT: ✅ Worth merging — Documentation is complete, well-organized, and follows project conventions.

---

This review was generated by an AI agent (OpenHands) on behalf of the user through OpenHands Automation. View conversation

[all-hands-bot]

✅ Review complete.

This review was performed through OpenHands Cloud Automation. You can log in and view the conversation here.

[all-hands-bot]

✅ Review complete.

This review was performed through OpenHands Cloud Automation. You can log in and view the conversation.

[all-hands-bot]

PR Review: Add Agent Canvas critic configuration docs (Updated)

Taste Rating: 🟢 Good taste

New Changes Since Last Review (commit 3e1a43f)

The new commit adds:

• ✅ Two screenshot images showing the critic settings and results UI
• ✅ A new "View Critic Results" section describing the compact and expanded critic result views

These additions are appropriate and enhance the documentation.

Previous Review Feedback (Still Open)

Two minor suggestions from the prior review remain unaddressed:

1. Line 17 - Consider "Use the OpenHands LLM provider" instead of "Prefer" to better convey automatic critic handling
2. Line 107 - Consider rephrasing "Use the OpenHands agent path" to "Use an OpenHands agent conversation" for clarity

These are optional improvements and not blocking.

Verification:

• ✅ docs.json: Valid JSON, navigation entry properly added
• ✅ critic.mdx: Comprehensive, well-structured content with proper Mintlify components
• ✅ Screenshots added: agent-canvas-critic-settings.png (107KB), agent-canvas-critic-result.png (40KB)
• ✅ New "View Critic Results" section with descriptive text and image
• ✅ All internal links verified to exist
• ✅ Consistent formatting throughout

Risk Assessment:

• [Overall PR] ⚠️ Risk Assessment: 🟢 LOW
• Pure documentation changes with no code, dependencies, or security implications

VERDICT: ✅ Worth merging — Documentation is complete and follows project conventions. The optional suggestions from the prior review are minor style improvements.

---

This review was generated by an AI agent (OpenHands) on behalf of the user through OpenHands Automation. View conversation

[all-hands-bot]

PR Review: Add Agent Canvas critic configuration docs

🟢 Good taste - Well-structured documentation with clear organization.

Summary

This PR adds comprehensive documentation for the experimental critic feature in Agent Canvas. The content is well-organized with clear sections covering prerequisites, configuration, iterative refinement, results viewing, advanced options, and troubleshooting.

What's Working Well

• Clear structure: Prerequisites → Enable → Configure → Troubleshoot flow is logical
• Appropriate use of Mintlify components: Warning for experimental status, Note for clarifications
• Valid links: All internal references verified (llm-profiles.mdx, openhands-llms.mdx, sdk/guides/critic.mdx)
• Good navigation integration: docs.json updated correctly, link added to customize-and-settings.mdx
• docs.json valid: JSON syntax is correct

Minor Suggestions (non-blocking)

1. Line 17 - "Prefer the OpenHands LLM provider" could be clearer as "Use the OpenHands LLM provider" to indicate this is the recommended path rather than an optional preference.

2. Line 95 - "Use the OpenHands agent path" may be ambiguous. Consider "Use an OpenHands agent conversation" for clarity.

3. Optional enhancement - A brief note about the score range (e.g., 0-1) in the introduction could help set expectations.

Verification

• docs.json is valid JSON with correct navigation structure
• All internal documentation links resolve
• Image references are properly formatted
• Customization settings description update is accurate

LGTM - Ready to merge.

---

This review was generated by an AI agent (OpenHands) on behalf of the user through OpenHands Automation.