Skip to content
SkillModel-led2,751 tokens in use

Grok Agents

Drive the headless Grok CLI from another agent as a peer, a builder or a fresh reviewer from a different model family.

680 stars on the repository MIT Licensephuryn/product-engineering · agent-teams/skills/grok-agents
npx skills add phuryn/product-engineering
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing Grok Agents costs

38tokens idle
2,751tokens in use
2,964tokens, up to

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Up to adds the files its instructions link. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter8/15
  • Structure19/25
  • Examples9/15
  • Output15/15
  • Size15/15
  • Resources15/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 43
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 44

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance100
  • Community44
  • Maturity4
  • Momentum40

Grok Agents handles requests like “ask Grok” or “have Grok build or review this”. It documents the exact grok flags for three modes: a read-only peer that argues a decision across turns, a builder that works in its own worktree, and a reviewer with memory and compatibility sources switched off.

The flags were read from grok —help and run on Grok CLI 1.0.70 on 2026-11-02, and the skill says the CLI changes often, so check the help first.

What You Get

  • Command recipes for peer, builder and reviewer sessions
  • Session handling with —session-id and —resume, and a warning against relying on -c
  • The reasons —always-approve is needed for reads and what an allowlist makes it safe for
  • Notes on sign-in with device auth and on running out of credit
  • A compact partner discipline that holds even when the folder is copied alone

Common Use Cases

  • Getting a different model family’s critique of a draft
  • Having Grok build a stage in a separate worktree
  • Running a fresh review with Grok memory and instruction sources switched off
  • Signing in to the CLI from a machine without a browser

What Makes It Unique

The recipes come from runs that went wrong: a call without approval that exited 0 after a sentence of preamble, a denylist that still allowed file creation, and an open stdin that waited forever. The reviewer recipe uses an allowlist of read tools as the write boundary and turns off 13 compatibility cells, measured with grok inspect. Short output is named as the tell of a round that could not read.

Under The Hood

The skill is instructions and shell recipes, with no scripts of its own. The agent runs the grok command-line tool as a subprocess in your repository, with the brief supplied as a file and the answer written to a file. A builder session can edit files and run shell commands with every tool approved, so the skill puts it in its own worktree and keeps deploys and secrets with the session agent. The recipes disable web search and MCP for peers and reviewers. The skill also tells the agent to run grok login, grok usage and grok sessions commands.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers