Okay, I incorporated the security-related changes from @umrath into the newest version, v1.2.69.
What was changed:
full_access is now false (was true, a leftover from when we needed the Docker socket).
docker_api: true is gone, and the docker CLI is no longer in the image.
capability net_admin is dropped from the AppArmor profile. The app only needs outbound HTTP, not interface or firewall control. net_raw stays.
The ttyd binary is now SHA256-verified at build time; build fails closed if the GitHub release ever serves a different binary.
mbpoll is pinned to v1.5.4 instead of cloning a moving main.
FONT_SIZE read from /data/options.json is sanitized before being passed to ttyd, just in case.
If you were using Claude Code to spawn or manage docker containers, this is not possible anymore without explicitly allowing it: you'd need to add docker_api: true back in your config and reinstall the docker CLI in a local override.
To pull it: Settings → Apps → App Store → ⋮ → Reload, then Update on the Claude Code app page. Restart after.
Full credit to @umrath, this is a direct port from his fork with his permission. Let me know if anything breaks.
Thank you for the effort to make it happen!
I will switch to your version now and mark (close? need to figure out what's best ...) my fork accordingly.
I am running into an issue with updating claude code, it cannot write the update.
Here is the auto update failure:
[WARN] Claude Code update failed/timed out (exit 243, see /tmp/claude-update.log). Last lines:
| npm error dest: '/usr/local/lib/node_modules/@anthropic-ai/.claude-code-1devilah'
| npm error }
| npm error
| npm error The operation was rejected by your operating system.
| npm error It is likely you do not have the permissions to access this file as the current user
| npm error
| npm error If you believe this might be a permissions issue, please double-check the
| npm error permissions of the file and its containing directories, or try running
| npm error the command again as root/Administrator.
| npm error A complete log of this run can be found in: /root/.npm/_logs/2026-05-21T09_34_41_875Z-debug-0.log
Here I am running the command directly inside the claude code app terminal.
npm install -g @anthropic-ai/claude-code@latest
npm error code EACCES
npm error syscall rename
npm error path /usr/local/lib/node_modules/@anthropic-ai/claude-code
npm error dest /usr/local/lib/node_modules/@anthropic-ai/.claude-code-1devilah
npm error errno -13
npm error Error: EACCES: permission denied, rename '/usr/local/lib/node_modules/@anthropic-ai/claude-code' -> '/usr/local/lib/node_modules/@anthropic-ai/.claude-code-1devilah'
npm error at async Object.rename (node:internal/fs/promises:784:10)
npm error at async moveFile (/usr/lib/node_modules/npm/node_modules/@npmcli/fs/lib/move-file.js:30:5)
npm error at async Promise.allSettled (index 0)
npm error at async [reifyPackages] (/usr/lib/node_modules/npm/node_modules/@npmcli/arborist/lib/arborist/reify.js:325:11)
npm error at async Arborist.reify (/usr/lib/node_modules/npm/node_modules/@npmcli/arborist/lib/arborist/reify.js:142:5)
npm error at async Install.exec (/usr/lib/node_modules/npm/lib/commands/install.js:150:5)
npm error at async Npm.exec (/usr/lib/node_modules/npm/lib/npm.js:207:9)
npm error at async module.exports (/usr/lib/node_modules/npm/lib/cli/entry.js:74:5) {
npm error errno: -13,
npm error code: 'EACCES',
npm error syscall: 'rename',
npm error path: '/usr/local/lib/node_modules/@anthropic-ai/claude-code',
npm error dest: '/usr/local/lib/node_modules/@anthropic-ai/.claude-code-1devilah'
npm error }
npm error
npm error The operation was rejected by your operating system.
npm error It is likely you do not have the permissions to access this file as the current user
npm error
npm error If you believe this might be a permissions issue, please double-check the
npm error permissions of the file and its containing directories, or try running
npm error the command again as root/Administrator.
npm error A complete log of this run can be found in: /root/.npm/_logs/2026-05-21T09_30_20_267Z-debug-0.log
@nstrelow thanks for the detailed log dumps in post #23, those were exactly what I needed.
You were right that this is not an ownership issue. The app runs as root, so the EACCES is coming from the AppArmor profile, not from Unix permissions. The profile grants r and ix on /usr/local/** but no w, and npm install -g ...@latest updates the package by renaming the existing directory inside /usr/local/lib/node_modules, which the kernel then denies. npm reports errno -13, exits 243, and the boot wrapper logs it as "failed/timed out", which is a misleading message: the actual cause is a hard permission denial, not the 300s update timeout that was fixed in 1.2.67.
Tracked as issue #13 and shipped in 1.2.70. The short version: I moved the npm global prefix off the read-only /usr tree to /data/npm-global. /data is the app's persistent state directory, the AppArmor profile already grants it rwk, and Supervisor keeps it across restarts and updates. On first boot of 1.2.70 the script seeds /data/npm-global from a staging copy baked into the image, and from then on both auto_update_claude at boot and a manual npm install -g @anthropic-ai/claude-code@latest from the app terminal land their writes there. /usr stays read-only, in line with the 1.2.69 hardening.
To pick it up: Settings -> Apps -> App Store -> three-dot menu -> Reload, then Update on the Claude Code page. After the update, auto_update_claude should work with Protection mode left on.
claude-code:/homeassistant# which claude
/opt/npm-global/bin/claude
claude-code:/homeassistant# claude
bash: /opt/npm-global/bin/claude: Permission denied
I think I found the source of the problem. It was introduced with the security changes in the permissions. I have now deployed a new version that works on my end. Let me know if that fixes it for you guys. 1.2.72
Just as a precaution that old apparmor configs are lying around I suggest to uninstall and reinstall completely from scratch. That has fixed it for me.
@sproft Thanks for your work to improve security and continue development on this very useful ha app. I make extensive use of node-red for automations and would love for claude to be able to access the node-red config, from what I understand this would need all_addon_configs:rw adding to the map. Is this something you would consider adding, or have an option of enabling?
Heyo, another user was requesting this feature and i can refer you to this Issue:
I decided against enabling this by default as some apps could be holding access_keys that you might not want to give to claude. The Issue includes instructions on how to enable this manually if you want.
If more users request this, I might look into how to enable this on a per app basis, but for now I will refer you to the manual solution. Let me know if it works for you.
continue get the error "command not found: claude"
the log:
[INFO] Seeding /data/npm-global from /opt/npm-global (first boot or fresh install)
cp: cannot create hard link '/data/npm-global/./lib/node_modules/@anthropic-ai/claude-code/bin/claude.exe' to '/data/npm-global/./lib/node_modules/@anthropic-ai/claude-code/node_modules/@anthropic-ai/claude-code-linux-x64-musl/claude': Permission denied
Hey @blademckain , I have just updated the App, so hopefully that fixes your Issue, because I was unable to recreate on my side. Let me know if that fixes it.
I get this when I try to authorize. I did copy the whole 2 lines. I didn't have a problem with the original fork of this add on. I just moved to this one because you are updating it and he was not. Looking froward to using it with Opus 4.8.
Thanks for doing this.
Thanks!
