AI assistants
Connect Lodeline AI to the assistant or coding agent you already use and ask what AI coding tools cost your team.
Server URL
https://mcp.lodeline.ai/mcp
Estimates need no sign-in. Questions about your own workspace ask you to sign in to Lodeline AI the first time.
Connect
claude mcp add --transport http lodeline https://mcp.lodeline.ai/mcp
When a workspace question asks you to sign in, run /mcp in Claude Code and choose Lodeline AI.
codex mcp add lodeline --url https://mcp.lodeline.ai/mcp
To sign in for workspace questions, run codex mcp login lodeline.
Or add it by hand to ~/.cursor/mcp.json:
{
"mcpServers": {
"lodeline": {
"url": "https://mcp.lodeline.ai/mcp"
}
}
}Or run:
code --add-mcp '{"name":"lodeline","type":"http","url":"https://mcp.lodeline.ai/mcp"}'Add a custom connector in Claude on the web or desktop and paste the server URL.
Any client that supports remote MCP servers over HTTP can use the server URL.
For workspace questions, sign in when the client asks if it supports OAuth. If it cannot sign in, use an API key (see Sign-in and access below) and send it as Authorization: Bearer <key>. The examples below read it from the environment variable LODELINE_MCP_KEY.
Cursor with an API key
{
"mcpServers": {
"lodeline": {
"url": "https://mcp.lodeline.ai/mcp",
"headers": {
"Authorization": "Bearer ${env:LODELINE_MCP_KEY}"
}
}
}
}Codex with an API key, in ~/.codex/config.toml
[mcp_servers.lodeline] url = "https://mcp.lodeline.ai/mcp" bearer_token_env_var = "LODELINE_MCP_KEY"
Claude Code with an API key
claude mcp add --transport http lodeline https://mcp.lodeline.ai/mcp --header "Authorization: Bearer $LODELINE_MCP_KEY"
Tools
| Tool | What it returns | Sign-in |
|---|---|---|
Estimate AI coding tool cost estimate_ai_coding_cost | Monthly seat cost for a team on GitHub Copilot, Claude or Cursor business plans, plus usage billed beyond the seats where the vendor publishes how it is billed. | Sign-in: None |
Compare a vendor's AI coding plans compare_plans | One vendor's plans side by side for your team size and usage. A cheapest plan is named only when at least two plans can be fully costed. | Sign-in: None |
Spend summary get_spend_summary | Spend for a window: total, money paid under your plans, change against the previous window, by tool and by team. | Sign-in: Required |
What changed what_changed | The models and tools behind a change in spend, recorded spikes, and data connections that are not reporting. | Sign-in: Required |
Month-end forecast get_forecast | Projected spend for this calendar month, with a low and high bound and how confident the projection is. | Sign-in: Required |
Budget status get_budget_status | Each active budget this month: its limit, money paid so far and whether it is on track, near or over. | Sign-in: Required |
Invoice reconciliation get_reconciliation | For a closed month, each vendor invoice against the spend measured for that tool, and whether the difference needs a look. | Sign-in: Required |
Every tool only reads. None can change anything in Lodeline AI or in your vendors' accounts.
Estimates come from each vendor's published list prices, last checked 2026-11-18. Where a vendor does not publish a usage limit, the answer says so instead of guessing.
What you can ask
- “What would GitHub Copilot Business cost for 25 developers?”
- “Compare Cursor's plans for 40 developers using $30 of extra usage each.”
- “What did we spend on AI coding tools in the last 30 days, by team?”
- “Why did our AI coding spend go up this month?”
- “Will we stay within our budgets this month?”
- “Does last month's Cursor invoice match the spend Lodeline AI measured?”
Clients with a prompt menu also offer ready questions: Why did AI coding spend change? · Month-end forecast and budget risk · Invoice reconciliation summary · Weekly summary for leadership.
Sign-in and access
- Sign in. The assistant opens Lodeline AI's sign-in and reads as you, with your role. Viewers cannot connect.
- API key. For clients that cannot sign in. An admin creates one in Settings › AI assistants, limited to the teams it needs and with an end date. It is shown once.
- Turn off. An admin turns assistant access on or off in Settings › AI assistants. While it is off, every assistant's next workspace question is refused. A single API key is revoked on the same page.
- See what was read. Recent reads on that page lists which figures were read and when. Questions are not stored.
GitHub Copilot coding agent
The coding agent cannot sign in, so it uses an API key.
- Create a key in Settings › AI assistants. Limit it to the teams the repository needs.
- Add an Agents secret named
COPILOT_MCP_LODELINE_AUTHwith the valueBearerfollowed by a space and the key. - In the repository's Settings › Copilot › MCP servers, add:
{
"mcpServers": {
"lodeline": {
"type": "http",
"url": "https://mcp.lodeline.ai/mcp",
"headers": {
"Authorization": "$COPILOT_MCP_LODELINE_AUTH"
},
"tools": [
"get_spend_summary",
"what_changed",
"get_forecast",
"get_budget_status",
"get_reconciliation"
]
}
}
}If your organization uses GitHub secret scanning, a custom pattern ll_mcp_[A-Za-z0-9_-]{43} finds a key committed by mistake.
Security
Keep tool approval on in your client, so you see each call before it runs. A tool result is data from Lodeline AI; your assistant should not follow instructions that appear inside one. Treat an API key like a password: store it as a secret and revoke it when it is no longer used.
What it receives
Only what your assistant sends with a question: team size, plans and a usage figure. No account, no source code and no conversation text. Your assistant's provider handles the conversation under its own terms. See Privacy.
Troubleshooting
A workspace question says I am not authorized
Sign in again from your client (/mcp in Claude Code, codex mcp login lodeline in Codex). If it still fails, ask an admin whether assistant access is on and whether your role is viewer; viewers cannot connect. A revoked or expired API key needs a new key.
My company network blocks the server
Ask IT to allow mcp.lodeline.ai over HTTPS.
Measured spend
The estimate tools use list prices. Your team's real spend, forecast and invoice reconciliation are in the Lodeline AI dashboard. Book a demo.