Anthropic Analytics Mcp — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Anthropic Analytics Mcp (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.
A local MCP server that lets Claude (Desktop and Code) answer plain-language questions about your Anthropic usage and spend — both the developer API platform and the Claude Enterprise product (chat, Claude Code, Cowork).
It wraps two Anthropic APIs behind one server:
usage across Claude, Claude Code, and Cowork.
API keys are never stored on disk. They live in 1Password and are injected at launch by the 1Password CLI (op run), which prompts for Touch ID.
Platform (Admin key):
| Tool | Description |
|---|---|
get_cost_report | USD cost over time, grouped by workspace/description. |
get_usage_report | Token usage over time, grouped/filtered by model, workspace, key, tier. |
list_workspaces | Map wrkspc_… ids to names. |
list_api_keys | Map apikey_… ids to names/status. |
Enterprise (Analytics key):
| Tool | Description |
|---|---|
get_activity_summary | Org-wide DAU/WAU/MAU, seat utilization, pending invites. |
get_user_activity | Per-user engagement (messages, projects, Claude Code commits/PRs, Cowork sessions). |
get_skill_usage | Skill adoption across surfaces. |
get_connector_usage | Connector adoption across surfaces. |
get_chat_project_usage | Conversation/user counts per chat project. |
get_user_cost | Users ranked by USD spend (usage credits on seat plans). |
get_user_token_usage | Users ranked by token usage. |
get_usage_by_subsidiary | Rolls per-user usage up by email domain = subsidiary. |
Units note: both APIs return costamountvalues in USD cents. The server converts totals to dollars (total_amount_usd,value_usd); raw rows insidedataremain in cents.
op)installed and its desktop app integration enabled (1Password → Settings → Developer → Integrate with 1Password CLI).
Admin key (platform usage/cost) — requires the org Admin/Primary Owner role:
https://platform.claude.com/settings/admin-keys).sk-ant-admin….Enterprise Analytics key (Enterprise product) — requires the Primary Owner:
https://claude.ai/analytics/api-keys).The Enterprise Analytics cost/usage endpoints reflect usage credits on seat-based plans; the engagement/adoption endpoints are the main value there.
Create two items (any vault), each with the key in a field (e.g. credential):
Right-click each field → Copy Secret Reference to get its op://… path.
Copy the example and fill in your two references:
cp config.local.sh.example config.local.sh
$EDITOR config.local.shconfig.local.sh is git-ignored, so your vault/item names never get committed.
./install.sh # creates .venv and installs depsSmoke test (triggers Touch ID, makes one live call):
echo '{"jsonrpc":"2.0","id":1,"method":"tools/list"}' | ./run.shUse the absolute path to run.sh.
Claude Desktop — ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"anthropic-analytics": {
"command": "/absolute/path/to/anthropic-analytics-mcp/run.sh"
}
}
}Claude Code:
claude mcp add anthropic-analytics --scope user /absolute/path/to/anthropic-analytics-mcp/run.shRestart Claude Desktop after editing its config.
can revise for up to 30 days. For invoice-grade Enterprise totals, query dates ≥30 days old.
op: command not found → install the 1Password CLI; enable the app integration.invalid character in secret reference → 1Password references can't contain' or &; rename the item.
401 invalid x-api-key → wrong key type/role. Admin endpoints need ansk-ant-admin… key; Enterprise endpoints need the Analytics key.
command path is absolute and run.sh isexecutable. Logs: ~/Library/Logs/Claude/mcp*.log.
MIT — see LICENSE.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.