Displayed all available AI assistant tools
Public
Displayed all available AI assistant tools
Agent
Grok
Model
Grok 4.5
StartedJul 16, 1:44 AM
Shared
Jul 16, 4:54 PM
Messages172
Downloads1
Identified built-in tools.
UnderstandChecked MCP tools availability.
UnderstandCompiled and presented full tool inventory.
VerifyShow me all the tools you have
I'll list the built-in tools and check which MCP tools are available right now.
Let me pull a fuller picture of the MCP tools across servers.
Here’s the full tool inventory available in this session.
| Tool | Purpose |
|---|---|
read_file | Read files (text, PDF, PPTX, images, notebooks) |
write | Create or overwrite files |
search_replace | Edit files with exact string replacement |
list_dir | List directory contents |
grep | Search file contents with ripgrep |
| Tool | Purpose |
|---|---|
run_terminal_command | Run bash commands (foreground or background) |
get_command_or_subagent_output | Poll background task / subagent output |
kill_command_or_subagent | Stop a background task or subagent |
monitor | Stream events from long-running commands |
| Tool | Purpose |
|---|---|
spawn_subagent | Launch specialized agents (general-purpose, explore, plan) |
todo_write | Track multi-step task progress |
enter_plan_mode / exit_plan_mode | Plan before coding, then present for approval |
ask_user_question | Ask you multiple-choice questions |
update_goal | Report progress on an active goal |
| Tool | Purpose |
|---|---|
scheduler_create | Create recurring/one-shot scheduled prompts |
scheduler_delete | Cancel a scheduled task |
scheduler_list | List active scheduled tasks |
| Tool | Purpose |
|---|---|
web_search | Search the web |
web_fetch | Fetch a URL as markdown |
open_page / open_page_with_find | Browse pages (full or pattern-matched) |
x_user_search | Find X users |
x_semantic_search | Semantic search of posts |
x_keyword_search | Advanced X post search |
x_thread_fetch | Fetch a post thread |
| Tool | Purpose |
|---|---|
image_gen | Generate images from text |
image_edit | Edit/transform images |
image_to_video | Animate a single image into video |
reference_to_video | Generate video from multiple reference images |
| Tool | Purpose |
|---|---|
search_tool | Discover MCP tools and their schemas |
use_tool | Call an MCP tool by qualified name |
tasks (connected — 6 tools)| Tool | Purpose |
|---|---|
tasks__create | Create a scheduled task / reminder |
tasks__list | List active scheduled tasks |
tasks__update | Update name, prompt, schedule, notifications |
tasks__pause | Pause or resume a schedule |
tasks__delete | Archive/deactivate a task |
tasks__get_results | Get recent execution results |
traces and vercelStill connecting (or not yet exposing tools). Their tools will show up via search_tool once ready.
Not tools themselves, but instruction packs I can load for specialized work — e.g. pptx, docx, xlsx, review, design, shadcn, Vercel/Next.js skills, check-work, imagine, etc.
Summary: ~35 built-in tools + 6 MCP tools from tasks. Want a deep dive on any category (e.g. media, agents, scheduling)?
Run a test that demonstrates all these tools reversibily
I'll run a reversible end-to-end demo: use a temp sandbox, exercise each tool category, then clean up everything we create.
Cleaning up created tasks and finishing the remaining demos.
Sandbox: /tmp/grok-tool-demo.nhtuB3 (created → exercised → fully removed).
| Tool | Result |
|---|---|
run_terminal_command | Created sandbox + sample files |
list_dir | Listed src/, out/ |
read_file | Read sample.txt |
write | Created src/notes.md |
search_replace | Appended line 3 to sample.txt |
grep | Found hello / function matches |
| Cleanup | rm -rf sandbox → SANDBOX_REMOVED=ok |
| Tool | Result |
|---|---|
web_search | Found grok.com / x.ai |
open_page | Fetched example.com |
open_page_with_find | Pattern-matched “Example|documentation” |
web_fetch | Markdown for example.com |
x_user_search | Returned a user result |
x_keyword_search | Latest post related to xAI ecosystem |
x_semantic_search | Semantic hit on Grok announcement topic |
x_thread_fetch | Thread + replies for post 2072738598663946648 |
| Tool | Result |
|---|---|
| Background bash | 5-tick loop → completed, output retrieved |
get_command_or_subagent_output | Read ticks 1–5 + DONE |
Infinite loop + kill_command_or_subagent | Killed successfully |
monitor | Streamed monitor-event-1..3 + complete |
| Tool | Result |
|---|---|
scheduler_create | ID 019f6898f505 (one-shot, non-durable) |
scheduler_list | Showed the demo task |
scheduler_delete | Cancelled |
scheduler_list again | No scheduled tasks |
| Tool | Result |
|---|---|
todo_write | Tracked 5 demo steps (all completed) |
spawn_subagent (explore, read-only) | Confirmed tree, file contents, add() in math.js |
| Tool | Result |
|---|---|
search_tool | Discovered tasks + traces schemas |
tasks__create | Created tool-demo-reversible-test |
tasks__update | Renamed + updated prompt |
tasks__pause | Failed: “User does not own schedule” (gateway quirk) |
tasks__get_results | Failed: team ID mismatch |
tasks__delete | Archived successfully |
tasks__list | Empty after delete |
traces__traces_search | 2 recent traces (7d) |
traces__traces_read | Wrote markdown under sandbox out/ (removed with sandbox) |
| vercel | Not available (auth required) |
| Tool | Result |
|---|---|
image_gen | images/1.jpg (session) |
image_edit | images/2.jpg (session) |
image_to_video | videos/1.mp4 (session) |
reference_to_video | videos/2.mp4 (session) |
| Tool | Why skipped |
|---|---|
enter_plan_mode / exit_plan_mode | Switches the whole session into plan mode |
ask_user_question | Blocks for interactive input |
update_goal | Mutates goal-mode state |
Media lives under this session folder (not in your home/project tree):
images/1.jpg, images/2.jpgvideos/1.mp4, videos/2.mp4No scheduled tasks, no MCP demo task, no /tmp sandbox. Safe leftover only if you want to inspect the generated media.
Bottom line: Most tools exercised end-to-end with create→use→undo where applicable. Two MCP task APIs (pause, get_results) hit ownership/team errors; Vercel MCP needs auth.
System Reminder
Monitor "019f6899-4d24-7df1-9513-ca6a8de2c131" ended: [monitor ended: exited (code 0)].
Description: Short monitor event stream demo
Command: for i in 1 2 3; do echo "monitor-event-$i"; sleep 0.3; done; echo "monitor-complete"
Duration: 1.0s
Use get_command_or_subagent_output("019f6899-4d24-7df1-9513-ca6a8de2c131") for full output.Monitor demo also finished cleanly (monitor-event-1..3 → monitor-complete, exit 0). That closes the last background piece from the reversible tool test — nothing else left running.