Claude Code TaskCreate Missing? Diagnose the Task Tools Safely

Claude Code TaskCreate Missing? Diagnose the Task Tools Safely

Aident AI

A coral fold and teal clay arc open around a warm oval on a deep indigo field.

Claude Code TaskCreate Missing? Diagnose the Task Tools Safely

If TaskCreate, TaskGet, TaskList, and TaskUpdate are missing from Claude Code, first confirm whether the built-in tools are absent from the session or merely blocked by a permission rule. Current reports show the tools disappearing across fresh sessions, operating systems, models, and even during a running session. In that case, reinstalling Claude Code or editing its private feature cache is unlikely to be a durable fix.

Use the checks below in order. Preserve your existing task data, test a clean session, and keep a plain Markdown task file as a temporary source of truth. As of July 29, 2026, Anthropic's public tools reference still lists all four Task tools as built in and requiring no approval.

Quick Diagnosis

What you observe

Most likely boundary

Safe next step

Tasks remain visible in the UI, but the model cannot call TaskCreate

Tool exposure or remote registration

Check the session tool roster, then reproduce in a fresh session

/permissions shows a bare deny for TaskCreate, another Task tool, or *

Permission configuration

Remove only the unintended deny rule and start a new session

The tools appear in a fresh session but vanish later

Intermittent tool registration

Save the time and session evidence, then resume with a Markdown task file

The tools are absent in safe mode on a current version and another model

Account or service-side behavior is plausible

Stop changing local state and report a minimal reproduction

TaskOutput and TaskStop still appear

Background-process controls are present

Do not mistake them for the task-list tools

TaskOutput and TaskStop manage background work. They do not replace TaskCreate, TaskGet, TaskList, or TaskUpdate.

Prerequisites

Before changing anything:

  • copy the current task list or important handoff state into a normal project file;

  • record the Claude Code version, surface, operating system, model, and approximate failure time;

  • note whether the task list still appears in the UI;

  • avoid deleting ~/.claude, changing private feature flags, or downgrading blindly;

  • keep conversation logs, credentials, local paths, and private repository details out of public issues.

If your workflow depends on durable instructions as well as tasks, review how to audit CLAUDE.md for newer models before changing project guidance.

Step 1: Confirm Which Tools the Session Actually Loaded

Inside Claude Code, run:

/context

Inspect the system tools section for these exact names:

TaskCreate
TaskGet
TaskList
TaskUpdate

Expected result: all four names appear. Anthropic's tools reference says these are built-in tools. If they do not appear, the problem is tool exposure, not simply Claude choosing not to use an available tool.

Also run:

/status

Expected result: you can see which settings sources and managed policies are active. Save this boundary-level evidence without copying private values.

Step 2: Rule Out a Permission Rule That Removes the Tool

Run:

/permissions

Look at the resolved deny rules and the settings file that supplied each rule. Anthropic documents an important distinction: a bare tool-name deny removes that tool from Claude's context entirely. For example, a deny rule named TaskCreate can make the tool disappear rather than showing a permission prompt. A bare * deny can remove every matching tool.

Expected result: no unintended bare deny targets the Task tools. If one exists, edit only the owning settings file, remove the unintended rule, and start a fresh session. Keep organization-managed policy intact and ask the administrator if the rule comes from a managed scope.

Do not add a broad allow rule as a countermeasure. Deny rules take precedence over allow rules, so an extra allow does not override the cause.

Step 3: Check Installation and Settings Health

From a terminal outside the active session, run read-only checks:

claude --version
claude doctor

Inside Claude Code, you can also run:

/doctor

Expected result: the running version is identified, settings files validate, and duplicate installations or configuration problems are reported. Apply a proposed repair only after reading what it will change.

Claude Code normally downloads updates in the background and applies them on the next launch. To ask the owning installation to update immediately, use:

claude update

Then fully close Claude Code, open a fresh terminal, verify claude --version, and start a new session. If Homebrew, WinGet, apk, or an organization manages the installation, keep using that owner instead of installing a second copy.

An update is a useful controlled test, not proof of the cause. Reports reproduced the missing Task tools on several current versions and, in one case, on the same binary before and after the tools disappeared.

Step 4: Reproduce in a Safe, Fresh Session

Start a new session with customizations disabled:

claude --safe-mode

Then run /context again and inspect the system tools.

Expected result: the built-in Task tools appear. Safe mode disables CLAUDE.md, skills, plugins, hooks, MCP servers, custom commands, and custom agents for that session. Authentication, built-in tools, permissions, and managed policy still apply.

This comparison separates custom project surfaces from the built-in tool boundary:

  • if the tools return in safe mode, re-enable customizations one at a time and use Claude Code's setup plugin safely to avoid stacking unrelated changes;

  • if they remain absent, the cause is outside ordinary project customizations;

  • if organization-managed policy is active, ask an administrator to reproduce with the same account policy.

Do not use a clean configuration as an excuse to delete the original. A temporary comparison is enough.

Step 5: Avoid the GrowthBook Cache Workaround

Community issue threads propose disabling feature traffic or editing cached GrowthBook values. Do not make that the first fix. The workaround is undocumented, changes internal feature evaluation, and one reporter later found that the original form disabled unrelated plugin channels. A later two-variable version is still a community workaround tied to private implementation details.

Use it only if Anthropic support gives you account-specific instructions and you understand the rollback. Otherwise, preserve local state and wait for the remote condition to clear. The open reports include task tools returning and later vanishing in the same session, which local cache editing cannot safely explain on its own.

Step 6: Keep Working With a Plain Task File

If the native tools remain absent, use a project file that humans and agents can inspect without private client state:

# TASKS

## In progress

- [ ] Describe the current task and its verification step.

## Blocked

- [ ] Record the blocker, owner, and next check.

## Done

- [x] Preserve completed work with the evidence that verified it

Ask Claude to read this file at the start of the session and edit only the relevant checkbox or note. Commit it only if the team wants it in version control; otherwise keep it in an ignored local handoff file.

Expected result: the work queue remains explicit and reviewable even without the native task panel. This is a fallback, not a claim that the missing tools are repaired.

Step 7: Report a Minimal Reproduction

If the tools are still absent after a fresh current-version session, /permissions shows no relevant deny, and safe mode reproduces the problem, report it through /feedback or the matching Anthropic issue.

Include:

  • Claude Code version and installation owner;

  • operating system and surface, such as CLI, desktop, or IDE extension;

  • model and account plan, without account identifiers;

  • whether tasks remain visible in the UI;

  • whether all four Task tools are missing from /context;

  • whether the issue reproduces in a new safe-mode session;

  • whether the tools vanished mid-session and the approximate UTC time.

Do not include credentials, complete transcripts, private paths, repository contents, or a full diagnostics archive. A short comparison across two sessions is more useful than a list of destructive repairs.

Why This Order Works

The sequence follows the same structure that makes Aident's Ollama networking guide useful: name the exact symptom, test one boundary at a time, show the expected result, and explain why the check matters.

The tool roster proves whether the capability exists. Permissions catch a documented local way to remove it. Doctor and version checks validate the installation without reinstalling. Safe mode separates project customizations from built-in tools. A normal task file preserves progress while a remote registration problem is unresolved.

Verify One External Action With Aident Loadout

After the session is stable, verify one read-only external Action independently of Claude Code's native task panel. Install or update Aident Loadout with this prompt:

Follow https://aident.ai/SETUP.md

Then run:

aident account auth status
aident vault vault --action status
aident capabilities search \
  --targetEnv staging \
  --queries '["find one connected read-only GitHub Action that searches issues"]' \
  --types '["action"]'

Inspect the selected schema before any execution:

aident capabilities get --name "<canonical-action-name>" --targetEnv staging

Success means Aident is authenticated, Vault can identify the connected integration, and one current Action contract is inspectable without pasting a provider credential into the Claude Code session. It does not restore TaskCreate; it verifies that an external tool boundary remains usable.

Use Aident Loadout to inspect one read-only Action. Measure success as one live schema inspected, zero provider secrets pasted, and zero repository changes.

Sources

Refresh this guide when Anthropic documents the Task-tool rollout or feature gate, closes the open July reports with a confirmed cause, or changes the commands used to inspect built-in tools and permissions.

Home

Home

Home

Integrations

Integrations

Integrations

Vault

Vault

Vault

Audit

Audit

Audit

Arana Grande

Arana Grande

Arana Grande

Free

Free

Free

30-day audit summary

30-day audit summary

30-day audit summary

Daily action-call volume and the latest receipts from the Loadout audit trail.

Daily action-call volume and the latest receipts from the Loadout audit trail.

Daily action-call volume and the latest receipts from the Loadout audit trail.

View Audit

View Audit

View Audit

Loadout usage

Loadout usage

Loadout usage

617 action calls in the last 30 days

617 action calls in the last 30 days

617 action calls in the last 30 days

May 19 - Jun 17

May 19 - Jun 17

May 19 - Jun 17

10 active days

10 active days

10 active days

Less

Less

Less

More

More

More

Recent activity

Recent activity

Recent activity

Latest action-call receipts from connected agents

Latest action-call receipts from connected agents

Latest action-call receipts from connected agents

Apr 23, 09:23 AM

Apr 23, 09:23 AM

Apr 23, 09:23 AM

Shopify

Shopify

Shopify

Creates Or Updates An Asset For A Theme

Creates Or Updates An Asset For A Theme

Creates Or Updates An Asset For A Theme

Success

Success

Success

Apr 23, 09:21 AM

Apr 23, 09:21 AM

Apr 23, 09:21 AM

Shopify

Shopify

Shopify

Update Products Param Product Id

Update Products Param Product Id

Update Products Param Product Id

Success

Success

Success

Apr 23, 08:53 AM

Apr 23, 08:53 AM

Apr 23, 08:53 AM

Shopify

Shopify

Shopify

Update Products Param Product Id

Update Products Param Product Id

Update Products Param Product Id

Failed

Failed

Failed

Apr 22, 22:13 PM

Apr 22, 22:13 PM

Apr 22, 22:13 PM

Shopify

Shopify

Shopify

Create Product Image

Create Product Image

Create Product Image

Success

Success

Success

Apr 22, 22:12 PM

Apr 22, 22:12 PM

Apr 22, 22:12 PM

Shopify

Shopify

Shopify

Create Product Image

Create Product Image

Create Product Image

Success

Success

Success

Connected integration coverage

Connected integration coverage

Connected integration coverage

162

162

162

of 753 accessible connected

of 753 accessible connected

of 753 accessible connected

Callable actions

Callable actions

Callable actions

1,126

1,126

1,126

Vault credentials

Vault credentials

Vault credentials

8

8

8

Explore what's possible

Explore what's possible

Explore what's possible

See all Integrations

See all Integrations

See all Integrations

Google Ads

Google Ads

Google Ads

All available Goolge Ads tools via...

All available Goolge Ads tools via...

All available Goolge Ads tools via...

X (twitter)

X (twitter)

X (twitter)

All available X tools via...

All available X tools via...

All available X tools via...

Github

Github

Github

All available Github tools via...

All available Github tools via...

All available Github tools via...

Notion

Notion

Notion

All available Notion tools via...

All available Notion tools via...

All available Notion tools via...

Slack

Slack

Slack

All available Slack tools via...

All available Slack tools via...

All available Slack tools via...

Firecrawl

Firecrawl

Firecrawl

All available Firecrawl tools via...

All available Firecrawl tools via...

All available Firecrawl tools via...

753 integrations are available for loadouts.

753 integrations are available for loadouts.

753 integrations are available for loadouts.

Plug your entire stack into your AI agents.

Plug your entire stack into your AI agents.

Plug your entire stack into your AI agents.

Skip the integration headache. Plug 750+ tools into Claude Code, Codex, and OpenClaw in one go, and let your agents execute today.

Skip the integration headache. Plug 750+ tools into Claude Code, Codex, and OpenClaw in one go, and let your agents execute today.