本文へ移動
cccskills
無料GitHub で公開

plugin-settings

Per-project plugin configuration via .local.md files — covers the .claude/plugin-name.local.md pattern for storing user-configurable settings with YAML frontmatter and markdown body. Use when implementing plugin settings, reading YAML frontmatter from hooks, creating configuration-driven behavior, managing agent state files, or adding per-project plugin configuration. Covers file structure, parsing techniques, common patterns (temporarily active hooks, agent state management, configuration-driven behavior), security considerations, and best practices.

インストール方法を見る

含まれるファイル(3)

  • SKILL.md7.5 KB
  • references/parsing-techniques.md8.1 KB
  • references/real-world-examples.md6.3 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Plugin Settings Pattern

Plugins store user-configurable settings in .claude/plugin-name.local.md files within the project directory. YAML frontmatter provides structured configuration; the markdown body carries prompts or additional context.

When to Use This Pattern

flowchart TD
    Start(["Plugin needs user-configurable behavior"]) --> Q1{What kind of state?}
    Q1 -->|"Per-project configuration<br>(validation level, enabled/disabled, limits)"| Settings["Use .claude/plugin-name.local.md<br>YAML frontmatter for config values"]
    Q1 -->|"Agent coordination state<br>(task assignment, session tracking)"| AgentState["Use .claude/plugin-name.local.md<br>Frontmatter for metadata<br>Body for task description"]
    Q1 -->|"Hook activation control<br>(enable/disable without editing hooks.json)"| HookControl["Use .claude/plugin-name.local.md<br>enabled: true/false field"]
    Q1 -->|"Loop iteration state<br>(iteration counter, completion promise)"| LoopState["Use .claude/plugin-name.local.md<br>Frontmatter tracks loop state<br>Body is the prompt to loop"]
    Settings --> Location
    AgentState --> Location
    HookControl --> Location
    LoopState --> Location
    Location["File location — .claude/plugin-name.local.md<br>Lifecycle — user-managed, gitignored<br>Structure — YAML frontmatter + markdown body<br>Consumers — hooks, skills, agents"]

File Structure

---
enabled: true
setting1: value1
numeric_setting: 42
list_setting: ["item1", "item2"]
---

# Additional Context

Markdown body for prompts, task descriptions,
additional instructions, or documentation.

Reading Settings

From Hooks (Bash)

The standard pattern for reading settings in hooks follows three steps — check existence, parse frontmatter, extract fields.

#!/bin/bash
set -euo pipefail

STATE_FILE=".claude/my-plugin.local.md"

# Quick exit if not configured
if [[ ! -f "$STATE_FILE" ]]; then
  exit 0
fi

# Parse YAML frontmatter (between --- markers)
FRONTMATTER=$(sed -n '/^---$/,/^---$/{ /^---$/d; p; }' "$STATE_FILE")

# Extract fields
ENABLED=$(echo "$FRONTMATTER" | grep '^enabled:' | sed 's/enabled: *//' | sed 's/^"\(.*\)"$/\1/')

if [[ "$ENABLED" != "true" ]]; then
  exit 0
fi

From Skills and Agents

Skills and agents read settings with the Read tool, then parse YAML frontmatter inline.

Check for plugin settings at `.claude/my-plugin.local.md`.
If present, parse YAML frontmatter and adapt behavior according to:
- enabled field controls whether the plugin is active
- mode field selects processing behavior (strict, standard, lenient)

Extracting the Markdown Body

Content after the closing --- marker is the body.

BODY=$(awk '/^---$/{i++; next} i>=2' "$FILE")

Common Implementation Patterns

Pattern 1 — Temporarily Active Hooks

Control hook activation via settings file instead of editing hooks.json (which requires restart).

STATE_FILE=".claude/security-scan.local.md"

if [[ ! -f "$STATE_FILE" ]]; then
  exit 0
fi

FRONTMATTER=$(sed -n '/^---$/,/^---$/{ /^---$/d; p; }' "$STATE_FILE")
ENABLED=$(echo "$FRONTMATTER" | grep '^enabled:' | sed 's/enabled: *//')

if [[ "$ENABLED" != "true" ]]; then
  exit 0
fi

# Run hook logic only when enabled

Pattern 2 — Agent State Management

Store agent-specific state and task assignments.

---
agent_name: auth-agent
task_number: 3.5
pr_number: 1234
coordinator_session: team-leader
enabled: true
dependencies: ["Task 3.4"]
---

# Task Assignment

Implement JWT authentication for the API.

Hooks read this to coordinate agents.

AGENT_NAME=$(echo "$FRONTMATTER" | grep '^agent_name:' | sed 's/agent_name: *//')
COORDINATOR=$(echo "$FRONTMATTER" | grep '^coordinator_session:' | sed 's/coordinator_session: *//')

tmux send-keys -t "$COORDINATOR" "Agent $AGENT_NAME completed task" Enter

Pattern 3 — Configuration-Driven Behavior

Settings drive validation strictness, file size limits, and allowed extensions.

LEVEL=$(echo "$FRONTMATTER" | grep '^validation_level:' | sed 's/validation_level: *//')

case "$LEVEL" in
  strict)   ;; # Maximum validation
  standard) ;; # Balanced validation
  lenient)  ;; # Minimal validation
esac

Creating Settings Files

flowchart TD
    Start(["Add settings to a plugin"]) --> S1["Design settings schema<br>— which fields, types, defaults"]
    S1 --> S2["Create template in plugin README<br>— document the .local.md format"]
    S2 --> S3["Add .gitignore entry<br>— .claude/*.local.md"]
    S3 --> S4["Implement parsing in hooks/skills<br>— use quick-exit pattern"]
    S4 --> S5["Provide sensible defaults<br>— when settings file is absent"]
    S5 --> S6["Document restart requirement<br>— changes need session restart"]

Best Practices

File naming — use .claude/plugin-name.local.md format. The .local suffix signals user-local, gitignored files.

Gitignore — add to project .gitignore:

.claude/*.local.md
.claude/*.local.json

Defaults — provide sensible defaults when the settings file does not exist.

if [[ ! -f "$STATE_FILE" ]]; then
  ENABLED=true
  MODE=standard
else
  # Parse from file
fi

Validation — validate field values before use.

if ! [[ "$MAX" =~ ^[0-9]+$ ]] || [[ $MAX -lt 1 ]] || [[ $MAX -gt 100 ]]; then
  echo "Invalid max_value in settings (must be 1-100)" >&2
  MAX=10
fi

Atomic updates — use temp file + atomic move to prevent corruption.

TEMP_FILE="${FILE}.tmp.$$"
sed "s/^field: .*/field: $NEW_VALUE/" "$FILE" > "$TEMP_FILE"
mv "$TEMP_FILE" "$FILE"

Restart requirement — settings changes require Claude Code session restart. Document this in the plugin README.

Security Considerations

Sanitize user input when writing settings files.

SAFE_VALUE=$(echo "$USER_INPUT" | sed 's/"/\\"/g')

Validate file paths stored in settings to prevent path traversal.

if [[ "$FILE_PATH" == *".."* ]]; then
  echo "Invalid path in settings (path traversal)" >&2
  exit 2
fi

Permissions — settings files should be readable by user only (chmod 600), not committed to git, not shared between users.

Deep Reference

For detailed parsing techniques (field extraction, body parsing, atomic updates, edge cases, yq integration), read Parsing Techniques Reference.

For production examples showing complete settings file lifecycles in real plugins, read Real-World Examples.

Related Skills

  • For hook authoring patterns that consume settings files, use /plugin-creator:hooks-guide
  • For component selection guidance (when to use hooks vs skills vs agents), use /plugin-creator:component-patterns
  • For the legacy commands format that can create settings files, use /plugin-creator:command-development

SOURCE: Adapted from Anthropic plugin-dev plugin skills/plugin-settings/ (8 files, ~545 lines). Accessed 2026-03-24.

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

Add automated documentation updater to any Claude skill. Creates a Python sync script that downloads upstream docs, processes markdown for AI consumption, and maintains local cache with configurable refresh. Collects template variables, then delegates implementation through 5-phase workflow. Use when adding auto-updating reference documentation to plugins or skills.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

SAM-style feature initiation workflow — discovery through codebase analysis, architecture spec, task decomposition, validation, and context manifest. Use when a user asks to add a feature, plan a feature, or convert an idea into an executable SAM plan.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

Browser automation for AI agents using the agent-browser CLI and Playwright. Use when navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, logging into sites, or automating any browser task. Triggers on "open a website", "fill out a form", "click a button", "scrape data", "test this web app", "automate browser actions", or any programmatic web interaction request.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

Runs the description-drift experiment — spawns all Claude Code agents simultaneously to collect self-reported capabilities, then compares them against static frontmatter descriptions to reveal how reliable orchestrator routing based on descriptions actually is. Use when measuring description drift across the agent fleet, re-running the capability collection experiment, analyzing a specific agent's self-reported capabilities, or auditing whether frontmatter descriptions accurately reflect agent behavior.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

Create or adapt Claude Code agent definitions. Use when creating an agent, changing subagent configuration, selecting agent scope, or designing a specialized delegation role.

日本語の概要は準備中です。原文の説明を表示しています。

Jamie-BitFlight/claude_skills672026年10月9日 更新

Jamie-BitFlight のスキルをすべて見る

このスキルの問題を報告する