Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add cognitedata/builder-skills --skill connect-atlas-mcpgit clone --depth 1 https://github.com/cognitedata/builder-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/cognitedata/builder-skills/connect-atlas-mcp)<a href="https://agentmods.dev/skills/cognitedata/builder-skills/connect-atlas-mcp"><img src="https://agentmods.dev/badge/skills/cognitedata/builder-skills/connect-atlas-mcp/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/cognitedata/builder-skills/connect-atlas-mcp"><img src="https://agentmods.dev/badge/skills/cognitedata/builder-skills/connect-atlas-mcp.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00106 | $0.00651 |
| Opus 5.5 | $0.00042 | $0.00260 |
| Sonnet 5.5 | $0.00021 | $0.00130 |
| Haiku 4.5 | $0.00011 | $0.00065 |
Grade A, and why
connect-atlas-mcp scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 18d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Connect Atlas MCP (Fusion launch)
The Fusion prompt has a block like this, with one command for the tool Fusion opened:
Industrial MCP for this project:
- MCP URL: https://<cluster>/api/v1/projects/<project>/ai/mcp
- Run this command as written, then finish the OAuth flow in the browser:
<command>
Known command shapes:
| Tool | Command starts with |
|---|---|
| Claude | claude mcp add-json cognite '{...}' |
| Cursor | cursor --add-mcp '{...}' |
| Copilot / VS Code | code --add-mcp '{...}' |
If the prompt carries a command for another tool, treat it the same way. Do
not rebuild the URL from cluster or project. Do not invent client ids,
callback ports, or switch add-json to claude mcp add.
If ---Start Building--- is missing, or the MCP URL or command is missing,
the prompt was truncated. Ask the builder to Copy from Fusion and paste, then
continue.
Do not paste tokens. Do not scrape the CLI keychain. The prompt's npx @cognite/cli@latest auth login signs in the CLI; it is not the MCP
credential.
1. Run Fusion's command as written
Run the command verbatim in a terminal. This is the happy path.
On Windows the single-quoted JSON is POSIX shell quoting. PowerShell 7.3+
accepts it as-is. In Windows PowerShell 5.1 or cmd.exe, wrap the JSON in
double quotes and escape inner quotes as \". Change only the quoting, never
the JSON.
Fallback, only if the command cannot run (CLI not on PATH, tool refuses the
flag): add the same server by hand to the tool's MCP config. Use the JSON from
the command, name the server cognite, and keep cognite-docs if the
template has it.
2. Finish OAuth in the browser
Enable the cognite server and complete the browser flow. Success is
connected in the tool, not a file on disk.
3. Explore this project's data models
Prefer MCP list/query. Name models relevant to the Fusion intent. Do not
invent spaces or views. Graph reads in app code: dm-graph-traversal.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 18d ago First seen · 68 lines · 106 tokens per session scan A 494913baf239
connect-atlas-mcp is a skill published in the GitHub repository cognitedata/builder-skills (6 stars, last pushed yesterday), licensed Apache-2.0. It adds 106 tokens to every session and 651 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-20.
connect-atlas-mcp compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| connect-atlas-mcpcognitedata/builder-skills | Skill | 6 | yesterday | Apache-2.0 | A | 106 | Claude Code |
| imagegenopenai/codex | Skill | 134k | today | Apache-2.0 | A | 113 | Codex |
| playwright-climicrosoft/playwright | Skill | 99k | today | Apache-2.0 | A | 19 | Claude Code |
| local-ai-agentsmicrosoft/ai-agents-for-beginners | Skill | 79k | today | MIT | A | 200 | Claude Code, Codex |
| systematic-debuggingobra/superpowers | Skill | 314k | 3d ago | MIT | A | 21 | Claude Code |
| next-cache-components-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 95 | Claude Code |
| next-partial-prefetching-adoptionvercel/next.js | Skill | 143k | yesterday | MIT | A | 103 | Claude Code |