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 skills add Sunwood-ai-labs/delegate-to-grok-skill --skill skillgit clone --depth 1 https://github.com/Sunwood-ai-labs/delegate-to-grok-skillWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sunwood-ai-labs/delegate-to-grok-skill/skill)<a href="https://agentmods.dev/skills/sunwood-ai-labs/delegate-to-grok-skill/skill"><img src="https://agentmods.dev/badge/skills/sunwood-ai-labs/delegate-to-grok-skill/skill/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sunwood-ai-labs/delegate-to-grok-skill/skill"><img src="https://agentmods.dev/badge/skills/sunwood-ai-labs/delegate-to-grok-skill/skill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00074 | $0.01245 |
| Opus 5 | $0.00037 | $0.00622 |
| Sonnet 5 | $0.00015 | $0.00249 |
| Haiku 4.5 | $0.00007 | $0.00125 |
Grade A, and why
delegate-to-grok 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 11d 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.
How it starts
The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Delegate to Grok Build
Use local grok as a secondary coding and research agent. Codex owns scope, safety, verification, and the final report.
Invocation contract
- Invoke every headless task through
scripts/invoke-grok.ps1. Never bypass it withgrok -p, including on retry. - Name
-WorkingDirectory. The default approved root is exactly that directory; use-AllowedRootonly for an explicitly confirmed parent root. - The wrapper runs a fresh
grok modelspreflight before the task and after it. It fails closed unless both explicitly show a logged-in session and a model list. - The wrapper is single-flight: wait up to 30 seconds; treat
GROK_BUSYas ownership by an existing run. Never start a second Grok or OAuth process. - Read-only is the default:
--permission-mode default,--sandbox read-only,--no-subagents,--no-plan, and--no-memory.-AllowWrites,-AllowSubagents, and-AllowMemoryeach require a specific user request. - Do not use
--always-approve,bypassPermissions,dontAsk, persistent debug logs, or a command that exposes prompts or auth files. Prompts always use a temporary--prompt-file. - Never read or expose
%USERPROFILE%\.grok\auth.json. The wrapper copies it only to an isolated temporary home and commits it back only after the run and postflight succeed. - Put all task text in one
-Promptvalue. This safely supports punctuation, quotes, commas, multiline text, and Japanese.
Authentication and OAuth recovery
Run grok models in a fresh process. A valid result must explicitly say the user is logged in and list available models. Do not infer authentication merely from an existing auth.json.
If it explicitly says authentication is missing or expired:
- Ensure no wrapper invocation or OAuth process is active. If
GROK_BUSY, wait instead of starting another login. - Start exactly one
grok login --oauth; the wrapper never logs in or out. - Read the browser-control skill before browser control. Choose the newest authorization tab created by this flow whose URL carries the active
redirect_uri=http://127.0.0.1:<port>/callback. - Inspect the visible page before clicking Allow. Hand password, one-time-code, CAPTCHA, and other secret steps to the user.
- Verify with a fresh
grok models. Browser success alone is insufficient.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 11d ago First seen · 75 lines · 74 tokens per session scan A 36d3f0ae6ba7
delegate-to-grok is a skill published in the GitHub repository Sunwood-ai-labs/delegate-to-grok-skill (2 stars, last pushed 1mo ago), licensed MIT. It adds 74 tokens to every session and 1,245 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
grok-subagent
Delegate bounded coding, investigation, review, implementation, and real-time X/Reddit/web research tasks from Codex to the locally authenticated Grok Build CLI. Use when the user asks Codex to consult Grok, use Grok as a subagent, compare independent model conclusions, spend SuperGrok quota on useful project work…
grok-prompting
Internal guidance for composing Grok Build prompts for coding, review, diagnosis, and research tasks inside the Grok Claude Code plugin.
grok-bridge
Delegate any substantial local task to Grok Build as one durable logical job, in a repository or a non-repository workspace, and retrieve its genuine final result across turn, token, context, and caller-session boundaries. Use when the user asks for Grok/Grok Build, an independent builder/tester/reviewer/researcher…
nexfade
Create one-time NexFade encrypted links for attached files or sensitive notes, check link status, and revoke links.
grok-prompting
Internal guidance for composing clear Grok task prompts from Codex rescue handoffs.
readiness-scorer
Calculate and explain a "Grok Readiness Score" (0-100) for a project based on concrete signals: quality of AGENTS.md / project rules, quantity and validity of skills, test/CI presence, git hygiene, ignore configuration, and verification culture. Use when asked to "score", "readiness score", "grok score", "how agent…