diff --git a/apps/x/packages/core/src/application/assistant/instructions.ts b/apps/x/packages/core/src/application/assistant/instructions.ts
index a29b225f..0e9a8c5d 100644
--- a/apps/x/packages/core/src/application/assistant/instructions.ts
+++ b/apps/x/packages/core/src/application/assistant/instructions.ts
@@ -37,6 +37,8 @@ Rowboat is an agentic assistant for everyday work - emails, meetings, projects,
 
 **Slack:** When users ask about Slack messages, want to send messages to teammates, check channel conversations, or find someone on Slack, load the \`slack\` skill. You can send messages, view channel history, search conversations, and find users. Always show message drafts to the user before sending.
 
+**Code with Agents:** When users ask you to write code, build a project, create a script, fix a bug, or do any software development task, load the \`code-with-agents\` skill first. It provides guidance for delegating coding work to Claude Code or Codex via acpx.
+
 ## Learning About the User (save-to-memory)
 
 Use the \`save-to-memory\` tool to note things worth remembering about the user. This builds a persistent profile that helps you serve them better over time. Call it proactively — don't ask permission.
diff --git a/apps/x/packages/core/src/application/assistant/skills/code-with-agents/skill.ts b/apps/x/packages/core/src/application/assistant/skills/code-with-agents/skill.ts
new file mode 100644
index 00000000..e4b24e54
--- /dev/null
+++ b/apps/x/packages/core/src/application/assistant/skills/code-with-agents/skill.ts
@@ -0,0 +1,79 @@
+export const skill = String.raw`
+# Code with Agents Skill
+
+Use this skill when the user asks you to write code, build a project, create scripts, fix bugs, or do any software development task that should be delegated to a coding agent (Claude Code or Codex).
+
+## Prerequisites
+
+The user must have one of the following installed on their machine:
+- **Claude Code** — https://claude.ai/code
+- **Codex** — https://codex.openai.com
+
+These are external tools that you cannot install for the user.
+
+## Workflow
+
+### Step 1: Gather requirements
+
+Before running anything, confirm the following with the user:
+
+1. **Working directory** — Ask which folder the code should be written in, unless the user has already specified it. Example: "Which folder should I work in?"
+2. **Agent choice** — Ask whether to use **Claude Code** or **Codex**. Mention that the chosen agent must already be installed on their machine.
+
+### Step 2: Confirm execution plan
+
+Once you know the folder and agent, tell the user:
+
+> I'll use [Claude Code / Codex] to [description of the task] in \`[folder]\`. All permission requests from the coding agent (file writes, command execution, etc.) will be automatically approved. Shall I proceed?
+
+Wait for the user to confirm before executing.
+
+### Step 3: Execute with acpx
+
+Use the \`executeCommand\` tool to run the coding agent via acpx. The command format is:
+
+**For Claude Code:**
+` + "`" + `
+npx acpx@latest --approve-all --cwd <folder> claude exec "<prompt>"
+` + "`" + `
+
+**For Codex:**
+` + "`" + `
+npx acpx@latest --approve-all --cwd <folder> codex exec "<prompt>"
+` + "`" + `
+
+### Critical: flag order
+
+The \`--approve-all\` and \`--cwd\` flags are **global flags** and MUST come **before** the agent name (\`claude\` or \`codex\`). This is the correct order:
+
+` + "`" + `
+npx acpx@latest [global flags] <agent> exec "<prompt>"
+` + "`" + `
+
+**Correct:**
+` + "`" + `
+npx acpx@latest --approve-all --cwd ~/projects/myapp claude exec "fix the bug"
+` + "`" + `
+
+**Wrong (will fail):**
+` + "`" + `
+npx acpx@latest claude --approve-all exec "fix the bug"
+` + "`" + `
+
+### Writing good prompts
+
+When constructing the prompt for the coding agent:
+- Be specific and detailed about what to build or fix
+- Include file names, function signatures, and expected behavior
+- Mention any constraints (language, framework, style)
+- If the user gave you a short request, expand it into a clear, actionable prompt for the agent
+
+### Step 4: Report results
+
+After the command finishes:
+- Summarize what the agent did
+- If it failed, explain why and suggest fixes
+- If the exit code is 5, it means permissions were denied — this should not happen with \`--approve-all\`, but if it does, let the user know
+`;
+
+export default skill;
diff --git a/apps/x/packages/core/src/application/assistant/skills/index.ts b/apps/x/packages/core/src/application/assistant/skills/index.ts
index 44774d6e..0ade5a62 100644
--- a/apps/x/packages/core/src/application/assistant/skills/index.ts
+++ b/apps/x/packages/core/src/application/assistant/skills/index.ts
@@ -12,6 +12,7 @@ import backgroundAgentsSkill from "./background-agents/skill.js";
 import createPresentationsSkill from "./create-presentations/skill.js";
 
 import appNavigationSkill from "./app-navigation/skill.js";
+import codeWithAgentsSkill from "./code-with-agents/skill.js";
 
 const CURRENT_DIR = path.dirname(fileURLToPath(import.meta.url));
 const CATALOG_PREFIX = "src/application/assistant/skills";
@@ -96,6 +97,12 @@ const definitions: SkillDefinition[] = [
     summary: "Navigate the app UI - open notes, switch views, filter/search the knowledge base, and manage saved views.",
     content: appNavigationSkill,
   },
+  {
+    id: "code-with-agents",
+    title: "Code with Agents",
+    summary: "Write code, build projects, create scripts, or fix bugs by delegating to Claude Code or Codex via acpx.",
+    content: codeWithAgentsSkill,
+  },
 ];
 
 const skillEntries = definitions.map((definition) => ({