feat: dropping perms before running claude (#509)

Co-authored-by: DevCats <christofer@coder.com>
Co-authored-by: Atif Ali <atif@coder.com>

Author: 3 people

registry/coder/modules/claude-code/README.md

@@ -13,7 +13,7 @@ Run the [Claude Code](https://docs.anthropic.com/en/docs/agents-and-tools/claude
 ```tf
 module "claude-code" {
   source         = "registry.coder.com/coder/claude-code/coder"
-  version        = "3.3.1"
+  version        = "3.3.2"
   agent_id       = coder_agent.example.id
   workdir        = "/home/coder/project"
   claude_api_key = "xxxx-xxxxx-xxxx"
@@ -51,7 +51,7 @@ module "claude-code" {
   boundary_log_level               = "WARN"
   boundary_additional_allowed_urls = ["GET *google.com"]
   boundary_proxy_port              = "8087"
-  version                          = "3.3.1"
+  version                          = "3.3.2"
 }
 ```
 
@@ -70,7 +70,7 @@ data "coder_parameter" "ai_prompt" {
 
 module "claude-code" {
   source   = "registry.coder.com/coder/claude-code/coder"
-  version  = "3.3.1"
+  version  = "3.3.2"
   agent_id = coder_agent.example.id
   workdir  = "/home/coder/project"
 
@@ -106,7 +106,7 @@ Run and configure Claude Code as a standalone CLI in your workspace.
 ```tf
 module "claude-code" {
   source              = "registry.coder.com/coder/claude-code/coder"
-  version             = "3.3.1"
+  version             = "3.3.2"
   agent_id            = coder_agent.example.id
   workdir             = "/home/coder"
   install_claude_code = true
@@ -129,7 +129,7 @@ variable "claude_code_oauth_token" {
 
 module "claude-code" {
   source                  = "registry.coder.com/coder/claude-code/coder"
-  version                 = "3.3.1"
+  version                 = "3.3.2"
   agent_id                = coder_agent.example.id
   workdir                 = "/home/coder/project"
   claude_code_oauth_token = var.claude_code_oauth_token
@@ -202,7 +202,7 @@ resource "coder_env" "bedrock_api_key" {
 
 module "claude-code" {
   source   = "registry.coder.com/coder/claude-code/coder"
-  version  = "3.3.1"
+  version  = "3.3.2"
   agent_id = coder_agent.example.id
   workdir  = "/home/coder/project"
   model    = "global.anthropic.claude-sonnet-4-5-20250929-v1:0"
@@ -259,7 +259,7 @@ resource "coder_env" "google_application_credentials" {
 
 module "claude-code" {
   source   = "registry.coder.com/coder/claude-code/coder"
-  version  = "3.3.1"
+  version  = "3.3.2"
   agent_id = coder_agent.example.id
   workdir  = "/home/coder/project"
   model    = "claude-sonnet-4@20250514"

registry/coder/modules/claude-code/scripts/start.sh

@@ -166,18 +166,9 @@ function start_agentapi() {
       BOUNDARY_ARGS+=(--pprof-port ${ARG_BOUNDARY_PPROF_PORT})
     fi
 
-    # Remove --dangerously-skip-permissions from ARGS when using boundary (it doesn't work with elevated permissions)
-    # Create a new array without the dangerous permissions flag
-    CLAUDE_ARGS=()
-    for arg in "${ARGS[@]}"; do
-      if [ "$arg" != "--dangerously-skip-permissions" ]; then
-        CLAUDE_ARGS+=("$arg")
-      fi
-    done
-
     agentapi server --allowed-hosts="*" --type claude --term-width 67 --term-height 1190 -- \
       sudo -E env PATH=$PATH setpriv --inh-caps=+net_admin --ambient-caps=+net_admin --bounding-set=+net_admin boundary "${BOUNDARY_ARGS[@]}" -- \
-      claude "${CLAUDE_ARGS[@]}"
+      claude "${ARGS[@]}"
   else
     agentapi server --type claude --term-width 67 --term-height 1190 -- claude "${ARGS[@]}"
   fi