dogfood/coder/main.tf (2)

41-42: Guard against whitespace-only prompts when toggling AI features

Today, any non-empty string (including whitespace) flips local.has_ai_prompt to true and enables the Claude-related resources. Trim the value before checking emptiness to avoid accidental enablement.

-  has_ai_prompt  = data.coder_parameter.ai_prompt.value != ""
+  has_ai_prompt  = trimspace(data.coder_parameter.ai_prompt.value) != ""

---

252-258: Optional: add prompt-length validation to prevent oversized env vars

The prompt is surfaced as an environment variable (CODER_MCP_CLAUDE_TASK_PROMPT). Extremely large values can exceed shell/process limits or bloat plans. If the provider supports string validation for coder_parameter, consider enforcing a reasonable max length; otherwise, we can trim on assignment.

Option A (preferred, if supported by the provider): add a validation block.

 data "coder_parameter" "ai_prompt" {
   type        = "string"
   name        = "AI Prompt"
   default     = ""
   description = "Prompt for Claude Code"
+  validation {
+    # Keep prompts to <= 4000 chars to avoid oversized env vars.
+    # If regex validation for strings is supported, enforce it here.
+    # regex = "^(?s).{0,4000}$"
+    # message = "Prompt must be 4000 characters or fewer."
+  }
   mutable     = true // You may wish to change the prompt, so mutability here is desirable.
 }

Option B (fallback): trim and hard-cap when setting the env value (silent truncate).

 resource "coder_env" "claude_task_prompt" {
   count    = local.has_ai_prompt ? data.coder_workspace.me.start_count : 0
   agent_id = coder_agent.dev.id
   name     = "CODER_MCP_CLAUDE_TASK_PROMPT"
-  value    = data.coder_parameter.ai_prompt.value
+  value    = substr(trimspace(data.coder_parameter.ai_prompt.value), 0, 4000)
 }

Can you confirm whether the coder provider supports string regex validation on parameters? If so, I’ll post a precise regex and message.

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

💡 Knowledge Base configuration:

• MCP integration is disabled by default for public repositories
• Jira integration is disabled by default for public repositories
• Linear integration is disabled by default for public repositories

You can enable these sources in your CodeRabbit configuration.

dogfood/coder/main.tf (1)

257-257: Change looks good: enabling mutability for AI prompt is consistent with intended UX

Allowing mid-workspace edits to the AI task prompt is desirable and aligns with how related parameters here are already marked mutable.