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 naplesblue/grokbridge --skill grok-build-delegategit clone --depth 1 https://github.com/naplesblue/grokbridgeWrote 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/naplesblue/grokbridge/grok-build-delegate)<a href="https://agentmods.dev/skills/naplesblue/grokbridge/grok-build-delegate"><img src="https://agentmods.dev/badge/skills/naplesblue/grokbridge/grok-build-delegate/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/naplesblue/grokbridge/grok-build-delegate"><img src="https://agentmods.dev/badge/skills/naplesblue/grokbridge/grok-build-delegate.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 | $0.00127 | $0.02268 |
| Opus 5.5 | $0.00051 | $0.00907 |
| Sonnet 5.5 | $0.00025 | $0.00454 |
| Haiku 4.5 | $0.00013 | $0.00227 |
Grade A, and why
grok-build-delegate 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 2mo 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 — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Grok Build delegate
Bridge to the local, subscription-authenticated Grok Build CLI (xAI). Use it to get an independent second opinion, an adversarial review, or a delegated investigation without leaving the Codex or Claude Code session. This is a separate review/task channel, not a model swap for your own work.
Delegation policy
Delegate when the user explicitly asks for Grok, Grok Build, grok-4.5, a second opinion, or an external/adversarial review. Also delegate selectively when all of these hold:
- the task has a clear scope and acceptance condition;
- Grok can work independently without secrets or missing product judgment;
- the host agent can verify the result with a diff, tests, or concrete evidence.
Good implicit candidates include reviewing a diff, investigating a focused failure, implementing an already-written plan, and mechanical or repetitive edits. Keep fuzzy product work, architecture, UX/copy judgment, security-sensitive work, and final acceptance in the host agent. Do not delegate merely to avoid doing ordinary work.
Routing
- Setup or health check →
grok_build_setup. Run it first when anything looks misconfigured; it reports CLI availability, auth, model, fallback, and state storage. - Quick question or second opinion →
grok_build_prompt. Foreground, strictly read-only. - Current web/X research → read-only
grok_build_taskwithwrite: false,background: true, andtimeout_ms: 180000or longer. Explicitly ask Grok to useweb_search/web_fetch, return direct source URLs, and distinguish official or first-party posts from community commentary. Pollgrok_build_status, then read the terminal record withgrok_build_result. - Code review →
grok_build_review. Passscope(working-tree,branch, orrepo) andbase_refforbranch/repo; setadversarial: trueto challenge assumptions. - Investigation or a longer task →
grok_build_task. For long runs setbackground: true, then pollgrok_build_statusand readgrok_build_resultwhen it finishes. Usegrok_build_cancelto stop a running job. - Bounded implementation → write-enabled
grok_build_task. Leave shell disabled when Grok can implement with native file tools alone. Addallow_bash: ["test"]only for one focused test or type-check command, orallow_bash: ["render"]only for one short render command. Do not ask Grok to run the full acceptance suite; the host agent performs final verification. - Image generation or editing →
grok_build_image. It always usesgrok-4.5. Passreference_imagesas absolute local paths for edits and for any named real person. - Video generation →
grok_build_video. It always usesgrok-4.5. Pass an absolutesource_imagewhen available; otherwise Grok creates a first frame, then animates it. Duration must be 6 or 10 seconds, and resolution must be480por720p. - Speech / voiceover / TTS generation →
grok_build_tts. This is a direct call to xAI/v1/ttsusing the local Grok Build OIDC subscription login from~/.grok/auth.json; it does not need or useXAI_API_KEY. Pass the spokentext, an ISO language code (zh,en, etc.), and optionallyvoice_id(eveis the default;ara,leo,rex,sal, or a custom voice ID can be used). If supplied,output_pathmust be an absolute.mp3path insidecwd.
What ships with it
1 file 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.
- 2mo ago First seen · 135 lines · 127 tokens per session scan A 40f44d168d70
grok-build-delegate is a skill published in the GitHub repository naplesblue/grokbridge (3 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 127 tokens to every session and 2,268 once invoked, about $0.0005 per session on Opus 5.5. 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.
grok-build-delegate compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| grok-build-delegatenaplesblue/grokbridge | Skill | 3 | 3mo ago | Apache-2.0 | A | 127 | Codex |
| imagegenopenai/codex | Skill | 133k | yesterday | Apache-2.0 | A | 113 | Codex |
| playwright-climicrosoft/playwright | Skill | 99k | yesterday | Apache-2.0 | A | 19 | Claude Code |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | 4d ago | MIT | A | 200 | Claude Code, Codex |
| systematic-debuggingobra/superpowers | Skill | 313k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 95 | Claude Code |
| next-partial-prefetching-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 103 | Claude Code |