Content Packs¶

Content packs define the perspective, tone, and heuristics used during a review. They're plain YAML — human-editable, version-controllable, and easy to share.

Built-in packs¶

Pack	Perspective	Best for
`staff-core`	Staff Engineer	General architecture and code reviews
`oncall-future-you`	On-call engineer, 3am	Operational readiness, failure mode review
`mentor-mode`	Experienced mentor	Learning, teaching, growth-focused feedback
`solutions-architect`	Solutions Architect	System design, entity modeling, build vs buy
`idp-readiness`	Platform Engineering	IDP maturity, platform vs process decisions

# See all available packs with descriptions
greybeard packs

# Use a specific pack
git diff main | greybeard analyze --pack oncall-future-you

Installing community packs¶

Anyone can publish a greybeard pack repo on GitHub. Install packs from any public repo:

# Install all packs from a repo's packs/ directory
greybeard pack install github:owner/repo

# Install a single file
greybeard pack install github:owner/repo/packs/my-pack.yaml

# Install from a raw URL
greybeard pack install https://example.com/my-pack.yaml

Installed packs are cached at ~/.greybeard/packs/ and available by name just like built-ins:

git diff main | greybeard analyze --pack my-pack-name

Managing installed packs¶

# List installed remote packs
greybeard pack list

# Re-download (refresh) a source
greybeard pack install github:owner/repo --force

# Remove a source
greybeard pack remove owner__repo

Using a local custom pack¶

Create a .yaml file and pass it directly — no install needed:

cat design-doc.md | greybeard analyze --pack ./my-team.yaml

Writing a pack¶

Packs are YAML files with the following structure:

name: my-pack
description: Short description shown in greybeard packs output

perspective: |
  Who is speaking. Be specific — this shapes the entire tone.
  Example: "A Staff Engineer on a team that owns 20+ services and
  has been paged about this exact type of failure before."

tone: calm and direct, not condescending

focus_areas:
  - operational impact
  - ownership clarity
  - long-term maintenance cost

heuristics:
  - "Would I be okay getting paged about this at 3am?"
  - "Who is the last person to understand this, and what happens when they leave?"
  - "Is this solving today's problem or a hypothetical future one?"

example_questions:
  - "Who owns the runbook for when this fails?"
  - "What does a 3am failure look like?"
  - "Is the rollback story clear and tested?"

communication_style: |
  Optional. How to frame concerns — collaborative, teaching, direct, etc.

See the built-in packs in packs/ for complete examples.

Tips for writing good packs¶

Be specific about perspective. "A Staff Engineer" is vague. "A Staff Engineer who has been paged about caching failures at 3am and spent a week diagnosing a silent data corruption bug" is much better — the model will reason from that specificity.
Heuristics are the soul of the pack. They're the questions your perspective would instinctively ask. Make them memorable and concrete.
Tone matters more than you think. The difference between "skeptical but fair" and "blunt and impatient" meaningfully changes the output.
Focus areas filter the noise. Without them, the model might surface concerns that aren't relevant to your perspective.

Publishing a pack¶

Create a public GitHub repo (e.g. your-handle/greybeard-packs)
Add a packs/ directory with .yaml files
Anyone can install it with:

greybeard pack install github:your-handle/greybeard-packs

Feel free to open an issue on the greybeard repo to get your pack listed.