• feat: convert soft_limit to limit #22048
• feat: remove agent workspaces limit #21998 👈 (View in Graphite)
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.

Documentation Check

Updates Needed

• docs/ai-coder/ai-governance.md - Update section "Agent Workspace Build Limits" (lines 131-145) to clarify that limits are advisory only and do not block workspace creation

  • ✅ Already accurate: Line 143 explicitly states "Coder Tasks and other AI features continue to function normally even if the limit is breached"
  • ✅ No blocking language: Documentation correctly describes warning-only enforcement (line 144)

• docs/ai-coder/ai-governance.md - Update the warning message reference (line 144) to match the new warning text

  • ✅ Not applicable: Documentation doesn't quote the specific warning text, only describes the behavior

Analysis

This PR changes managed agent limits from hard enforcement to soft enforcement (advisory warnings only). After reviewing the current documentation:

✅ Documentation is already accurate:

• docs/ai-coder/ai-governance.md (lines 139-145) correctly states that features "continue to function normally even if the limit is breached"
• No mention of blocking workspace creation - only warnings
• Accurately describes the enforcement model

Minor consideration (optional enhancement):

• docs/ai-coder/tasks.md uses the phrase "To scale beyond the 1,000 build limit" which could potentially be clarified to emphasize that exceeding the limit triggers warnings (not blocking), but this is already implied by the ai-governance.md documentation that states features continue to function normally

Conclusion: No documentation changes are required. The existing documentation already accurately reflects the advisory-only nature of managed agent limits.

Automated review via Coder Tasks