ultraplan / ultrareview commands not visible despite eligibility
Preflight Checklist
- [x] I have searched existing issues and this hasn't been reported yet
- [x] This is a single bug report (please file separate reports for different bugs)
- [x] I am using the latest version of Claude Code
What's Wrong?
Environment
- Claude Code version: 2.1.111
- Install: npm global
- Platform: macOS (Darwin 25.3.0)
- Account:
- Model: Opus 4.7 (1M context)
Expected
/ultraplan and /ultrareview should appear as invokable slash commands in the CLI, per docs (they're listed under both the web platform and CI/CD sections of the Claude Code
docs map).
Actual
Typing /ultraplan or /ultrareview does not resolve to a command. Neither appears in /help output. claude ultraplan / claude ultrareview from the shell also not recognized.
Verified
- claude --version → 2.1.111
- CLAUDE_CODE_MAX_OUTPUT_TOKENS=64000, API auth configured
- Onboarding completed, multiple projects tracked
- Other slash commands and plugins (superpowers, codex-peer-review, deep-plan, etc.) work fine
- Prior session memory (2026-04-16) confirms these are documented CLI features, not plugins requiring install
Ask
Clarify whether ultraplan / ultrareview require a flag, feature gate, or specific entitlement that isn't advertised — or if docs are ahead of the release.
What Should Happen?
It should show ultraplan and ultrareview as I AM USING CLAUDE MAX
Error Messages/Logs
Steps to Reproduce
C
Claude Model
Opus
Is this a regression?
No, this never worked
Last Working Version
_No response_
Claude Code Version
2.1.111
Platform
Anthropic API
Operating System
macOS
Terminal/Shell
Terminal.app (macOS)
Additional Information
_No response_
Showing cached comments. Read the full discussion on GitHub ↗
14 Comments
Found 2 possible duplicate issues:
This issue will be automatically closed as a duplicate in 3 days.
🤖 Generated with Claude Code
+1, same issue on an independent Max 20x account. Adding a Linux data point and some stronger evidence that should help rule out local causes.
Please don't auto-close this as a duplicate of #49314 (
/buddy, unrelated command) or #16846 (/teleport, different feature and older version). This is a distinct, unresolved bug.Environment
2.1.112(also reproduced on2.1.111)/home/<user>/.local/share/claude/versions/2.1.112), auto-update channel:latestANTHROPIC_API_KEYset). Relevant fields in~/.claude/.credentials.json:subscriptionType: "max"rateLimitTier: "default_claude_max_20x"user:sessions:claude_codeSymptoms
/and filteringultrashows nothing./ultraplanand/ultrarevieware not registered in the command palette at session start./tasksis registered and functional, so the cloud surface itself is initialised in the session — onlyultraplan/ultrarevieware missing.Ruling out local causes
I ran Claude Code with a completely isolated profile and a brand-new OAuth login to eliminate every local variable:
That environment has:
settings.json,NODE_OPTIONS/CLAUDE_CODE_*overrides,Result is identical — commands still absent. This rules out user configuration, plugins, hooks, MCP servers, custom env vars, or stale credentials as the cause.
Support path already exhausted
Opened a ticket with support@anthropic.com (conversation IDs
215473947107196,215473947183070,215473947232591). Support confirmed that the plan is Max 20x, OAuth is correct, and the configuration is fine, but the replies are automated and loop through "tryclaude updateand restart" — which does not help because the CLI is already on the latest version and command registration is driven by the server response at session start, not local state.Likely cause
Given that:
rateLimitTier: "default_claude_max_20x"),2.1.92(where/ultraplanlanded),this appears to be a server-side feature-flag / entitlement not applied to some Max 20x accounts, not a client-side bug in the CLI.
Ask
Could someone from the team:
/ultraplan//ultrarevieware gated by a server-side flag that can be stuck off for some qualifying Max 20x accounts.duplicatelabel (the linked issues are different commands/features).Happy to provide a
/doctordump or anonymisedcredentials.jsonfields if that helps triage.+1 same symptoms. Claude Code v2.1.112, macOS, at an eligible subscription tier.
+1 facing the same issues. I'm on Claude Code v2.1.112, macOS, at an eligible subscription tier as well.
+1, same issue here.
Environment
2.1.114Observed
/ultraplanis not available/ultrareviewis not availableclaude --help | grep -i ultrareturns nothing/helpalso does not list/ultraplanor/ultrareviewSo the problem is still reproducible on
2.1.114, not just2.1.111.+1 same v2.1.114, Max plan, extra usage enabled, /login fresh, still Unknown command.
Linux with 2.1.114 on a max x20 plan and I can't access this command either.
I also tried logging in again.
I am guessing we are gated because we are in the AU region. I get you have to roll things out slowly but can you at least give us a heads up and not radio silence.
Imagine if you had a tool that could be used to communicate these types of things, you know like a large language model....
@wiltaylor not in AU region
Same issue here in Canada. Windows. Claude v2.1.116. Max20 plan.
FWIW, I had the exact same issue. I checked my
claude/settings.json, saw that I hadDISABLE_TELEMETRYset to1, removed that line, and now the slash commands appear when I launch the cli.Good spot – if I remove
DISABLE_TELEMETRYthe command does show up. Although I’m not sure that’s a desirable trade-off.Update: the visibility of
/ultraplanand/ultrareviewis controlled by a server-side dynamic config, not by anything local.The relevant config appears to be named
tengu_ultraplan_config. When itsenabledis nottruefor a given user, the CLI never registers the command at all —/, the keyword path, and/helpall miss it./ultrareviewfollows the same pattern; its policy-blocked state surfaces under the telemetry eventtengu_ultraplan_create_failedwithreason: "policy_blocked", but that only fires if launch was attempted — affected users here never get that far.Side note on
~/.claude/statsig/: on 2.1.11x this directory is no longer the authoritative cache. The SDK keeps evaluations in memory per process and does not recreate the directory on fresh startup. Mine was stuck atreceivedAt2026-01-31 withsource: "NetworkNotModified"for 80+ days; deleting it changes nothing, and the #28777 workaround no longer applies on these versions.Ask
tengu_ultraplan_config(and theultrareviewanalogue) and report whetherenabledistrue.false/nullon qualifying plans, either enable it or share the targeting criteria — right now there is no way to tell from the user side whether we're hitting a rollout bucket, a stuck flag, or an intentional gate.Removing
DISABLE_TELEMETRY=1did the trick for me...Closing for now — inactive for too long. Please open a new issue if this is still relevant.