ai-mcp-server@8.3.2
VERIFIEDPublisher claims: "MCP server for 2kw.ai — EU-hosted AI platform: OpenAI-compatible LLM gateway, schema-driven document extraction, transcription, agents with a knowledge base, and cost observability. 199 tools for Claude Code, Cursor, and Windsurf." Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Coverage
Started with a decoy AI_2KW_API_KEY, BACKBONE_API_KEY; tool list observed. Tool calls with a real credential not tested.- Tested
- server startup with a tagged decoy in AI_2KW_API_KEY, BACKBONE_API_KEY · MCP initialize and tools/list · every tool description, for hidden instructions · where the decoy credential went · every file read, file written and host contacted by what ran
- Not tested
- what the tools do when called with a real credential
- Needs
- AI_2KW_API_KEY, BACKBONE_API_KEY — give it to the server where it runs, through vet-skills run --secrets, so the value never enters the server's environment. Never paste a credential into a chat, a ticket or a report.
Before you give it your token
- It will ask for
- no credential variables read in the code; at startup it asked for AI_2KW_API_KEY, BACKBONE_API_KEY
- Where it talks to
- 2kw.ai (its own)
- What we saw when it ran
- no credential files read · no network
- What we could not test
- what the tools do when called with a real credential
Reasonable to connect. No credential asked for, no third-party host named, Started with a decoy AI_2KW_API_KEY, BACKBONE_API_KEY; tool list observed. Tool calls with a real credential not tested.
Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.
What changed since 8.3.0
vet-skills diff ai-mcp-server 8.3.0 8.3.2In the sandbox, this version reached for nothing 8.3.0 did not. Compared from the two recorded runs, not from source.
The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios) on 2026-11-21 and scored it 85/100 (Verified). It did what it claims and nothing else we could detect: no undeclared network egress, no credential reads, no persistent instruction writes. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 48 s · 2 scenarios · 0 deviationsStatic findings
},"instructions":{"description":"System instructions for the agent","type":"string"},"options":{"description":"Model and run oFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct."check:spec": "tsx --env-file=.env openapi/scripts/check-drift.ts",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.…as one ZIP and save it to a local path: a folder per succeeded item and a report.csv listing every item with its status, reason and outputs. While the batch runs, the ZIP holds the items finished so far.", { ba…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why."generate:spec": "tsx --env-file=.env openapi/scripts/fetch-spec.ts",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.},"instructions":{"description":"System instructions for the agent","type":"string"},"options":{"description":"Model and run oFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.… server.tool("2kw_create_agent_version", "Create a new version of an agent. Auto-updates the 'latest' label. Give its model as `model` (one model) or `models` (an ordered list whose first entry is the default); one of the two is required."…Fix: Pin versions and sources; vendor anything fetched at install time.},"instructions":{"description":"System instructions for the agent","type":"string"},"options":{"description":"Model and run oFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.Cost to the agent
Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.
Known vulnerabilities
No advisories in OSV for @2kw/ai-mcp-server@8.3.2 as of 22 min ago. This is the package itself, not its dependencies.