/arn-infra-refresh

name: arn-infra-refresh
description: >-
  This skill should be used when the user says "refresh infra references",
  "update tool versions", "check for new MCPs", "update infra patterns",
  "refresh registries", "arn infra refresh", "infra refresh",
  "update references", "check for updates", "refresh infrastructure tools",
  "update MCP servers", "refresh CLI versions", "update base images",
  "arn-infra-refresh", or wants to update the version-sensitive
  infrastructure reference files (tool versions, MCP packages, CLI commands,
  base image tags, IaC patterns) using online research.
version: 1.0.0

Arness Infra Refresh

Refresh the project-local infrastructure reference files using online research. This skill searches for the latest stable versions of tools, MCP servers, CLI commands, base image tags, and IaC patterns, then updates the externalized reference overrides in `.arness/infra-references/` while preserving checksums for future conflict detection.

The 28 evolving reference files are organized into 6 categories (registries, IaC patterns, container patterns, CI/CD patterns, cloud patterns, infrastructure guides). Users can refresh all categories at once, select specific categories, or target a single file.

Prerequisites

Read `## Arness` from the project's CLAUDE.md. If no `## Arness` section exists or Arness Infra fields are missing, inform the user: "Arness Infra is not configured for this project yet. Run `/arn-infra-wizard` to get started — it will set everything up automatically." Do not proceed without it.

Extract:

• **Experience level** -- derived from user profile. Read `~/.arness/user-profile.yaml` (or `.claude/arness-profile.local.md` if it exists — project override takes precedence). Apply the experience derivation mapping from `${CLAUDE_PLUGIN_ROOT}/skills/arn-infra-ensure-config/references/experience-derivation.md`. If no profile exists, check for legacy `Experience level` in `## Arness` as fallback.
• **Reference overrides** -- path to the project-local reference directory (e.g., `.arness/infra-references`)
• **Reference version** -- current reference version for display

If **Reference overrides** is not configured, stop and instruct the user: "Reference overrides are not configured. Run `/arn-infra-init` (or upgrade your existing configuration) to set up reference externalization before refreshing."

---

Workflow

Step 1: Load Current State

Read the checksums tracking file and the reference manifest to understand the current state of all externalized reference files.

1. Read `.reference-checksums.json` from the **Reference overrides** path (e.g., `.arness/infra-references/.reference-checksums.json`) 2. Read the reference manifest: `Read ${CLAUDE_PLUGIN_ROOT}/skills/arn-infra-refresh/references/reference-manifest.md` 3. **Reconcile new files:** If the manifest lists files not present in `.reference-checksums.json`, these are newly added reference files from a plugin update. Copy them from the plugin defaults (`${CLAUDE_PLUGIN_ROOT}/skills/<source-path>`) to the Reference overrides directory, generate their SHA-256 checksums, and add them to `.reference-checksums.json` with `updatedBy: "init"` before proceeding.

Present current state to the user:

"**Reference Overrides Status:**

• **Reference version:** [Reference version from config]
• **Plugin baseline version:** [pluginVersion from checksums]
• **Last refreshed:** [lastRefreshed from checksums, or 'Never' if null]
• **Total files tracked:** [count of files in checksums]
• **Categories:** [list of unique categories with file counts]"

---

Step 2: Select Refresh Scope

Present scope selection adapted to the user's experience level.

**Expert:**

Ask (using `AskUserQuestion`):

**"Refresh scope?"**

Options: 1. **All (28 files)** -- Refresh every reference file across all categories 2. **Select categories** -- Choose which categories to refresh 3. **Single file** -- Specify a single filename to refresh individually

If **All (28 files)**: proceed to Step 3 with all 28 files selected.

If **Single file**: ask the user to specify the filename. Proceed to Step 3 with that single file.

If **Select categories**:

Ask (using `AskUserQuestion`) with 2 questions:

Q1 (multiSelect: true): **"Tooling categories (select any):"**

Options: 1. **Registries** (4 files) -- MCP servers, CLIs, plugins, secrets providers 2. **IaC Patterns** (7 files) -- OpenTofu, Pulumi, CDK, Bicep, Kubernetes, PaaS, validation 3. **Container Patterns** (2 files) -- Dockerfile base images, Compose syntax

Q2 (multiSelect: true): **"Operations categories (select any):"**

Options: 1. **CI/CD Patterns** (3 files) -- GitHub Actions, GitLab CI, pipeline security 2. **Cloud Patterns** (4 files) -- Environment config, promotion, alerting, observability 3. **Infrastructure Guides** (8 files) -- Provider overviews, IaC tools, detection, deployment

Combine the selected categories from both questions and proceed to Step 3 with the matching files.

**Intermediate:** Present a recommended scope:

"I recommend refreshing **Registries** and **IaC Patterns** -- these categories change most frequently and have the highest impact on generated infrastructure code.

• **Registries** (4 files) -- tool versions, install commands, MCP packages
• **IaC Patterns** (7 files) -- syntax, module references, best practices"

Ask (using `AskUserQuestion`):

**"How would you like to proceed?"**

Options: 1. **Recommended** -- Refresh Registries and IaC Patterns (11 files) 2. **All** -- Refresh all 28 reference files 3. **Full menu** -- See the full category selection (Expert view)

**Beginner:** Default to refreshing all categories:

"I'll refresh all 28 reference files across all categories to make sure everything is up to date. This will search online for the latest stable versions of tools, patterns, and configurations."

Ask (using `AskUserQuestion`):

**"Refresh all 28 reference files?"**

Options: 1. **Yes** -- Proceed with refreshing all