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 skills/bigu1/grok-plugin-codex/grok-cli-runtimenpx skills add bigu1/grok-plugin-codex --skill grok-cli-runtimegit clone --depth 1 https://github.com/bigu1/grok-plugin-codexWhat 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 | $0.00019 | $0.01159 |
| Opus 5 | $0.00010 | $0.00580 |
| Sonnet 5 | $0.00004 | $0.00232 |
| Haiku 4.5 | $0.00002 | $0.00116 |
Grade A, and why
grok-cli-runtime 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 yesterday.
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.
This is a copy
89% identical to grok-cli-runtime — 4 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Grok Runtime
Use through the Grok MCP tools. If MCP is unavailable, call the companion directly with node plugins/grok/scripts/grok-companion.mjs <command> ....
Recommended Grok CLI version: ≥ 1.0.5 (minimum still 0.2.118; flags are capability-gated).
Workspace
Installed Codex plugin MCP servers start from the plugin cache. Pass cwd with the active project
directory on every Grok MCP call so Grok inspects, edits, and stores artifacts in the intended
workspace rather than the cached plugin directory. Direct companion calls can use --cwd <path>.
Concurrency
- Multiple companion jobs may run at once. There is no global single-agent lock.
- Prefer
background: trueor--backgroundwhen a Codex turn is launching more than one Grok job. - Each MCP call should make exactly one companion invocation.
- Parallelism = multiple background companion jobs, not a serialized queue.
- When several jobs are running, always pass job ids to
status/result/cancel.
Control flags (most write/plan commands)
MCP input keys map to companion flags:
| MCP property | Companion flag |
|---|---|
sandbox |
--sandbox |
planMode |
--plan |
permissionMode |
--permission-mode |
agent |
--agent |
noSubagents |
--no-subagents |
memory / noMemory |
--memory / --no-memory |
allow / deny |
--allow / --deny (repeatable) |
disableWebSearch |
--disable-web-search |
forkSession |
--fork-session |
maxTurns |
--max-turns |
CLI posture
- Prefer denylist (
--disallowed-tools) over tools allowlist (Grok session-create bugs). - Media: no yolo / no tools allowlist.
--dry-run/--validate-only/ babysitlist: read-only (no yolo).- Write-capable default for rescue/design/execute/babysit add|check|remove.
Depth notes
grok_execute_planwithlatest=trueresolves newest.grok-designs/*.md.- Design/workflow/plan/document jobs harvest copies into
.grok-designs//.grok-workflows//.grok-plans//.grok-docs/. - Review
postPending=true: skips empty findings; empty/oversize diffs fail closed and save findings under.grok-reviews/. - Plan results prefer harvested
plan.mdbody over narration. - Stop-gate uses sandbox
read-only+ denylist (no yolo).
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.
- yesterday First seen · 102 lines · 19 tokens per session scan A 2f7331474918
grok-cli-runtime is a skill published in the GitHub repository bigu1/grok-plugin-codex (0 stars, last pushed 8d ago), licensed Apache-2.0. It adds 19 tokens to every session and 1,159 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. It is 89% identical to grok-cli-runtime, differing in 4 lines, and is treated as a copy.
Other skills, from other repositories
codex-delegate
Delegate coding, review, diagnosis, or follow-up work to the local Codex CLI from Grok Build. Use when the user asks to command Codex, hand work to Codex, run /codex, /codex:rescue, /codex:review, or says 让 Codex 干、叫 Codex 审查、 指挥 Codex.
codex-cli-runtime
Internal helper contract for calling the codex-companion runtime from Grok Build.
gpt-5-4-prompting
Internal guidance for composing Codex and GPT-5.4 prompts for coding, review, diagnosis, and research tasks inside the Grok Codex plugin.
codex-result-handling
Internal guidance for presenting Codex helper output back to the user.
goal-verifier
Independently verify a Grok Build goal is truly complete. Checks GOAL.md done conditions, runs tests, and returns PASS or REJECT with evidence. Use before updategoal(completed: true), when asked to "verify goal", "check if done", or as a reviewer sub-agent at the end of a goal run.
goal-completion-check
Pre-flight check before calling updategoal(completed: true). Ensures verifier ran, tests passed, and GOAL.md checklist is complete. Use when implementer believes work is done, or when asked "can I mark goal complete?".