Guide · Agent skill

How to use the AgentRC Policy Helper agent skill

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

The acreadiness-policy skill is an agent skill that helps you create, choose and apply a policy for AgentRC, a readiness-scoring tool. A policy is a small JSON file that turns off checks you don't need, raises the importance of checks you care about, and sets a minimum pass rate. This guide explains what agent skills are in practice, how to install this one, how to ask for what you need, and how to use policies in a single repo or across an organisation.

AgentRC Policy Helpergithub/awesome-copilot · MIT Download the skill

Quick start

1Install the skill. Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder, or upload it under Settings → Capabilities → Skills in the Claude apps.
2Check the prerequisites. Make sure Node.js with npx is installed. AgentRC is fetched on demand when the skill runs.
3Pick a starting point. Ask the agent to recommend one of the built-in policies: strict, ai-only or repo-health-only.
4Create or adjust your policy. Use the new sub-command or ask in plain language. Decide what to disable, which impacts to raise and what pass-rate threshold to use.
5Reference the policy. Add the policy to the policies array in agentrc.config.json so it stays in effect for the repository.
6Apply the policy and review the report. Run apply to generate a readiness report that reflects the policy, then review any gaps that remain visible.
7Add a CI gate if needed. Add the npx command with --fail-level to your pipeline to fail builds below your chosen maturity level.

What an agent skill is and what this one does

An agent skill is a folder containing a SKILL.md file with instructions that an AI agent loads when your request matches the skill's description. This skill is defined in a single file, SKILL.md.

Once installed, the agent can help you work with AgentRC policies. It can walk you through choosing settings, suggest built-in starter policies (strict, ai-only and repo-health-only), and run AgentRC with your policy so the readiness report reflects it. It can also show how to gate CI builds on a maturity level.

Show policies currently in effect
Scaffold a new policy file with sensible defaults
Apply a policy by running AgentRC with it
Explain the trade-offs of disabling specific checks

Install the acreadiness-policy skill

Download the ZIP for the skill from this page. Then install it using one of these two methods.

For Claude Code, unzip the file into ~/.claude/skills/ to make it available in all your projects, or into a project's .claude/skills/ folder to share it with that project only.

In the Claude apps, upload the ZIP under Settings → Capabilities → Skills.

Personal install: ~/.claude/skills/
Project install: .claude/skills/ inside your repository
Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Understand what a policy file contains

A policy is a JSON file with three optional sections: criteria, extras and thresholds. The criteria section can disable checks and override their impact or level. The extras section can disable extra checks, which do not affect the score. The thresholds section sets the pass rate.

Impact sets each check's weight: critical is 5, high is 4, medium is 3, low is 2 and info is 0. The score is 1 minus deductions divided by the maximum possible weight. Grades run from A at 0.9 or above, through B at 0.8, C at 0.7 and D at 0.6, to F below 0.6.

criteria.disable: a list of check names to turn off
criteria.override: change impact or level, or retitle a check
extras.disable: a list of extra checks to turn off
thresholds.passRate: a number such as 0.9

{"name":"my-policy","criteria":{"disable":["env-example","observability","dependabot"],"override":{"readme":{"impact":"high","level":2},"lint-config":{"title":"Linter required"}}},"extras":{"disable":["pre-commit"]},"thresholds":{"passRate":0.9}}

Ask for what you need in plain language

You do not need to remember the command names. The skill loads when your request involves policies, strict mode, custom scoring, disabling checks, org standards or CI gating. For example, you can ask it to recommend a starting policy for a static site, or to set up a strict baseline for a frontend repo.

You can also call the skill directly with its slash-command forms: show lists the active policies, new followed by a name scaffolds a policy file, and apply followed by a path or package runs AgentRC with a policy.

"Which built-in policy suits a team that wants strict scoring?"
"Turn off observability checks, because this is a static site."
"Create a policy called strict-frontend."
"Set up a CI step that fails below maturity level 3."

/acreadiness-policy show
/acreadiness-policy new strict-frontend
/acreadiness-policy apply ./org-baseline.json

Chain an org baseline with team overrides

Most organisations get the best results from a shared baseline policy plus per-team overrides. Pass the policies to AgentRC as a comma-separated list, and the skill can run this for you with apply. Reference the policy from agentrc.config.json in the policies array to keep it in effect for the repository.

The skill's operating rules favour overriding impact over disabling a check. Disabling hides a gap completely, while overriding keeps it visible in the report.

Keep one org-wide baseline file, such as org-baseline.json
Add a per-team file, such as team-frontend.json, that overrides only what differs
Reference policies in agentrc.config.json with a policies array

npx -y github:microsoft/agentrc readiness --json --policy ./org-baseline.json,./team-frontend.json

Gate CI builds on a maturity level

To make a build fail when a repository falls below a set maturity level, add an AgentRC step to your CI pipeline and pass --fail-level. The example below uses the strict policy and fails when the level is below 3.

Only use the gate with settings that reflect real standards. Do not quietly disable important checks just to get a green build, because that hides real gaps from reviewers.

Use --policy to pick the policy the gate uses
Use --fail-level to set the minimum maturity level

- run: npx -y github:microsoft/agentrc readiness --policy ./policies/strict.json --fail-level 3

Safety notes before you run it

This skill runs scripts and uses the network. It runs AgentRC through npx, which fetches the AgentRC tool from github:microsoft/agentrc on demand. Review what the skill will run before you approve it, and check that you trust the source.

Policies only disable or re-weight existing criteria. JSON policies cannot add new criteria. For new detection logic, the source points to AgentRC's TypeScript plugin system, described in docs/dev/plugins.md.

Requires Node.js with npx
Fetches the AgentRC tool on demand from microsoft/agentrc
To re-render reports, the skill hands off to the assess skill or the ai-readiness-reporter agent

What people use it for

Standardise readiness checks across several repositories with one shared baseline policy.
Skip observability checks for a static site that has no runtime to observe.
Give a team a strict policy with raised impact on must-have files such as the README and CODEOWNERS.
Add a CI step that fails when a repository falls below a set maturity level.

Frequently asked questions

How do I install the acreadiness-policy skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or into a project's .claude/skills/ folder for one repository. In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.

What do I need to run it?

You need Node.js with npx, because the skill runs AgentRC through npx github:microsoft/agentrc, which is fetched on demand. The skill also uses the assess skill or the ai-readiness-reporter agent to re-render reports.

Is the acreadiness-policy skill free to use?

The skill is published under the MIT licence. The source does not describe any pricing, so check the licence file in the repository for the terms.

Can a policy add my own custom checks?

No. JSON policies can only disable checks, override their impact or level, and set thresholds. New detection logic requires writing a TypeScript plugin, which the AgentRC documentation covers in docs/dev/plugins.md.

Why is my policy not taking effect?

Check that the policy file is referenced in the policies array of agentrc.config.json. Use the show sub-command to list the policies currently in effect, and confirm the path is correct relative to the file.

Should I disable checks or lower their impact?

Prefer overriding impact over disabling. A disabled check disappears from the report, while an overridden check still shows the gap. Only disable checks that are truly irrelevant to your stack.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Use this skill if you are an engineering lead or maintainer who needs consistent readiness standards across repositories, or a CI gate that fails below a set maturity level. Skip it if you only want a one-off readiness report with default settings, since no policy is needed for that, or if you need custom detection logic, which requires writing TypeScript plugins instead.

Download the skill