claude-plugin-mcp-manager — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited claude-plugin-mcp-manager (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Bundled commands use $PLUGIN_ROOT ($env:PLUGIN_ROOT in PowerShell; same path suffix) for the plugin root. Set it once: the host's plugin-root variable when defined (Claude Code: PLUGIN_ROOT="$CLAUDE_PLUGIN_ROOT"), otherwise this plugin's absolute root path.
Covers plugin marketplaces, installed plugins, token-cost details, validation, release tags, scoped installs, session-only plugin loading, MCP server configuration, project approval state, stdio/http/sse transports, headers, env vars, OAuth, and strict-MCP-config runs.
Check the local command surface:
python3 "$PLUGIN_ROOT/scripts/claude_code_inspector.py" --commands plugin "plugin marketplace" "plugin install" "plugin validate" mcp "mcp add" --jsonRead current state before mutating it:
claude plugin list
claude plugin marketplace list
claude mcp listUse claude plugin details <name> and claude mcp get <name> before changing an existing item.
Validate source:
claude plugin validate /path/to/plugin
claude plugin validate /path/to/plugin --strictInspect installed plugin cost/components:
claude plugin details <plugin>Install from a configured marketplace:
claude plugin install <plugin>@<marketplace> --scope user
claude plugin install <plugin> --scope projectLoad a plugin only for one session:
claude --plugin-dir /path/to/plugin
claude --plugin-url https://example.com/plugin.zipDo not run plugin update, plugin uninstall, plugin prune, marketplace update/remove, or tag creation without an explicit target and user intent.
Stdio MCP servers:
claude mcp add <name> --scope local -- <command> <args>
claude mcp add <name> --scope project -e KEY=VALUE -- <command> <args>HTTP/SSE servers:
claude mcp add --transport http <name> https://example.com/mcp
claude mcp add --transport sse <name> https://example.com/sseFor auth-bearing servers, avoid literal secrets in commands; prefer env vars or prompts:
claude mcp add --transport http <name> https://example.com/mcp --header "Authorization: Bearer $TOKEN"
claude mcp add --transport http <name> https://example.com/mcp --client-id "$CLIENT_ID" --client-secretUse --strict-mcp-config on session startup when the task must ignore all MCP servers except explicit --mcp-config files or JSON.
.mcp.json servers unless the user asks for that project.validate, list, details, and get before mutating plugin or MCP state.--strict-mcp-config with explicit config.Report state inspected, exact commands run, scopes used, validation/details/get evidence, any source/config/cache path touched, and any remaining user action such as auth or setting an env var.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.