gryph: Skill for Claude Code

.claude/skills/gryph-policy-authoring/SKILL.md

gryph-policy-authoring is a skill for Claude Code from safedep/gryph. It costs 200 tokens per session (1,825 once invoked), scanned B, original, Apache-2.0.

A guide for writing Gryph security policies, which control what an AI coding agent may read, write, or run. It helps draft rules that allow, block, warn about, or require approval for actions.

In plain words
What is it for?
Use it to create and validate a policy draft, then give a human the command needed to install it. The skill does not install policies itself.
Why use it?
It provides a reviewable way to define agent permissions without changing the active security settings automatically.

Skill for Claude Code

Written for Claude Code: installed under .claude/.

This is safedep/gryph's own configuration. It tells Claude Code how to work on gryph itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything gryph configures →

Needs its repository: it runs a file that does not travel with it, so clone the repository first. The line is gryph policy validate --file ./gryph-policy/<name>.yaml.

Reuse

Borrowing it

Nothing to install: this file belongs to safedep/gryph. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/safedep/gryph/main/.claude/skills/gryph-policy-authoring/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/safedep/gryph

Made for: Claude Code.

Wrote 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.

agentmods badge for gryph-policy-authoring

README.md
[![agentmods](https://agentmods.dev/badge/skills/safedep/gryph/gryph-policy-authoring/github.svg)](https://agentmods.dev/skills/safedep/gryph/gryph-policy-authoring)
Your own site
<a href="https://agentmods.dev/skills/safedep/gryph/gryph-policy-authoring"><img src="https://agentmods.dev/badge/skills/safedep/gryph/gryph-policy-authoring/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.

agentmods 80×15 button for gryph-policy-authoring

Your own site · 80×15
<a href="https://agentmods.dev/skills/safedep/gryph/gryph-policy-authoring"><img src="https://agentmods.dev/badge/skills/safedep/gryph/gryph-policy-authoring.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 200 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,825 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector warn 7 Sept 2026
SkillSpector: 2 findings, up to high

These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →

  • high Tool Misuse · line 84
    Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
    Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
  • high Tool Misuse · line 84
    Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).
    Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
How audits are shown
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5.1 $0.00200 $0.01825
Opus 5 $0.00100 $0.00912
Sonnet 5 $0.00040 $0.00365
Haiku 4.5 $0.00020 $0.00183

Measured 11d ago against content hash 1621b8b3c0ce, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade B, and why

gryph-policy-authoring scanned grade B with 1 finding 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.

Recursive force deletemediumDestructive command

rm -rf with a variable or a broad path is one typo away from removing the wrong tree.

gryph policy test --file ./gryph-policy/<name>.yaml --action command_exec --command "rm -rf /"

Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.

.claude/skills/gryph-policy-authoring/SKILL.md · 142 lines

How it starts

The opening of the file, as written. The whole thing — 142 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Authoring Gryph Policies

Gryph mediates what an AI coding agent reads, writes, and runs. A policy is an ordered list of rules. Each rule matches an action and returns a decision. This skill helps a human draft those rules, with an agent's help, then hands the human a command to install them.

You draft in a workspace directory. You never install. You never write into Gryph's config directory. The human reviews the draft and runs the install command you give at the end. This keeps the human in control of what becomes active policy.

The two hard rules

  1. Draft only in a workspace directory you create, for example ./gryph-policy/. Never write a policy file into Gryph's config directory or its policies/ directory. Gryph blocks an agent write there by design, and a direct write bypasses the human review.
  2. Never install. Do not run gryph policy install, gryph policy edit, or gryph policy init <name> (a bare name writes into policies/). These change live policy. Your job ends at a validated draft plus an install command for the human.

The gryph binary is the source of truth

The installed gryph binary is authoritative and version-matched. Read the schema and the existing policy from it. Do not author fields from memory.

  • gryph policy schema prints the exact rule fields, match criteria, and decision set. This is the single source of truth for the schema. Check it before you write a field.
  • gryph policy builtin prints Gryph's own built-in rules. Read them as working examples to imitate, and to see what already fires so your rule does not collide with them.
  • gryph policy list shows every active source and its rule count.

The security policy guide at https://github.com/safedep/gryph/blob/main/docs/security-policy.md is a concept guide. It explains match criteria, scope, CEL conditions, decisions, messages, receipts, and has worked examples. Read the section you need for concepts. When the doc and the binary disagree, trust the binary and tell the user.

Read the full file on GitHub · 142 lines

Changes

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.

  1. 11d ago First seen · 142 lines · 200 tokens per session scan B 1621b8b3c0ce

Subscribe to this mod's changes

gryph-policy-authoring is a skill published in the GitHub repository safedep/gryph (161 stars, last pushed 11d ago), licensed Apache-2.0. It adds 200 tokens to every session and 1,825 once invoked, about $0.0010 per session on Opus 5. A static security scan graded it B with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

spec-kitty-charter-doctrine

Run charter interview, generation, context, and sync workflows for project governance in Spec Kitty 3.x. Access doctrine artifacts programmatically via DoctrineService. Resolve agent profiles. Load action-scoped governance context iteratively, not all at once. Triggers: "interview for charter", "generate charter"…

Priivacy-ai/spec-kitty · 135 tokens

mongodb-query-optimizer

Help with MongoDB query optimization and indexing. Use only when the user asks for optimization or performance: "How do I optimize this query?", "How do I index this?", "Why is this query slow?", "Can you fix my slow queries?", "What are the slow queries on my cluster?", etc. Do not invoke for general MongoDB query…

fcakyon/claude-codex-settings · 98 tokens

polar-local-environment

This skill should be used when setting up or managing Polar local development environment with Docker.

fcakyon/claude-codex-settings · 22 tokens

neo4j-modeling-skill

Design, review, and refactor Neo4j graph data models. Use when choosing node labels vs relationship types vs properties, migrating relational/document schemas to graph, detecting anti-patterns (generic labels, supernodes, missing constraints), designing intermediate nodes for n-ary relationships, enforcing schema with…

neo4j-contrib/neo4j-skills · 152 tokens

lov-article-creator

A skill that turns source material or a draft into a complete, branded WeChat public-account article package.

lovstudio/skills · 75 tokens

lov-repo2docs

Turn any folder of source material — a code repository, a pile of articles, a mixed knowledge dump with images — into a professional, polished Fumadocs (Next.js) documentation website, then deploy it to https://{product-id}.example.com/docs. Works by reading the folder one unit at a time and incrementally growing and…

lovstudio/skills · 186 tokens