Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/vanyangyang/cursor-bridge/grok_executegit clone --depth 1 https://github.com/Vanyangyang/cursor-bridgeWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00014 | $0.00701 |
| Opus 5 | $0.00007 | $0.00351 |
| Sonnet 5 | $0.00003 | $0.00140 |
| Haiku 4.5 | $0.00001 | $0.00070 |
Grade A, and why
grok_execute scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Grok Executor Mode Toggle
Use $grok-executor-mode for the task-local role contract and $grok-build-supervisor for Grok session operations. This command is a prompt-level task policy, not a native host collaboration-mode switch.
Interpret the entire trimmed value of "$ARGUMENTS" case-insensitively:
on: bind Grok Executor Mode to the current host task's absolute project directory and immediately ensure its supervised Grok session is ready. Before changing mode state, callgrok_session_inspectwithview: status. If the proxy is not initialized, the project directory is unavailable or ambiguous, another workspace owns active Grok work, or session setup fails, leave the mode off and report only the actionable next step. Otherwise reuse the exact attached visible session for this directory or callgrok_session_openwith that absolutecwd,mode: newwhen no session is attached, the defaultpresentation: windows_terminal, andconfirmation: OPEN_GROK_SESSION. Usemode: resumeonly with an exact verified attached session UUID. If open returnsneeds_workspace_trust, leave executor mode pending and tell the user once that Grok is waiting for their trust decision in the already-open terminal and that/grok_execute ondoes not need to be run again. Do not answer this as a tool permission, simulate input, or pass--trust. After the user confirms, inspect the same session again; the Supervisor reuses that terminal and reportsreadywhen Grok registers it. Commit the task-local mode and workspace binding only after the visible TUI is verified ready. Reply briefly that executor mode is on, the terminal is ready, and the user can provide the task. Render user-facing prose in the language of the user's latest substantive message unless the user explicitly requests another language; do not persist an inferred language. Preserve tool names, states, paths, IDs, and exact permission/form options verbatim. This activation authorizes opening the terminal but must not callgrok_session_promptor send development work. Never ask the user for a separate TUI command. Repeatedonis idempotent and must reuse a trust-pending TUI rather than opening another one.off: explicitly deactivate Grok Executor Mode for the current host task, clear its task-local workspace binding, and confirm briefly. Subsequent ordinary tasks use normal host behavior. Do not automatically cancel a running Grok prompt, disconnect ACP, close the visible TUI, or stop the owned Leader; continue any already-required Supervisor monitoring under its normal contract. Repeatedoffis idempotent.- Empty or any other value: do not change mode state, create a session, or send work. Reply only with the usage
/grok_execute onor/grok_execute off.
Only the most recent exact /grok_execute on or /grok_execute off in this task controls the mode. The workspace selected by a successful on is task-local and must not be written to global proxy settings or silently changed while the mode remains active. Ordinary Grok mentions, task text, direct Skill invocation, partial matches, and every other command must not activate or deactivate it. Activation does not broaden permission for destructive work, publication, external writes, or consequential owner decisions.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 16 lines · 14 tokens per session scan A 079889a75b4b
grok_execute is a command published in the GitHub repository Vanyangyang/cursor-bridge (51 stars, last pushed 3d ago), licensed MIT. It adds 14 tokens to every session and 701 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
add-agent
引导新增一个 Agent 适配器。用法 /add-agent.
check
跑全套本地校验(前端 lint/test/build + Rust check)。.
thirdparty-web-interface-guidelines
Review these files for compliance: $ARGUMENTS.
search
Search X (Twitter) and the web in real time with Grok, then summarize with sources.
result
Show the final output of a completed Grok job.
plan
Research, clarify requirements, and produce an approved implementation plan before writing code (Cursor Plan Mode).