Build effective agents using Model Context Protocol and simple workflow patterns
- capability exposureinferred+35
- recent driftinferred+12
- tool safetyinferred+12
- abandonmentinferred+6
inferred
The A–E grade is our heuristic synthesis — a "review this" prompt, not a verdict. Each factor is tagged by what backs it: attested (a verifiable record), reported (a third party's claim), or inferred (our own heuristic, e.g. permissions). See methodology.
graded 9m ago · see ecosystem CVEs →
- A · 6 → D · 65
no known CVEs for this server.
- highdangerous code
dynamic exec: eval()/exec(), __import__(), unsafe yaml.load()
analyzed commit f62d849 · analyzer v17 · 1h ago
skills & prompt files 3
danger signals4
- dynamic code executioneval()/exec()lastmile-ai-mcp-agent-f62d849/scripts/gen_schema.py:161
exec(content, namespace) - dynamic code execution__import__()lastmile-ai-mcp-agent-f62d849/src/mcp_agent/cli/commands/build.py:133
module = __import__(package) - dynamic code executionunsafe yaml.load()lastmile-ai-mcp-agent-f62d849/src/mcp_agent/cli/secrets/yaml_tags.py:107
return yaml.load(yaml_str, Loader=SecretYamlLoader) - dynamic code executioneval()/exec()lastmile-ai-mcp-agent-f62d849/src/mcp_agent/utils/pydantic_type_serializer.py:758
exec(func_code, namespace)
- recent drift+12 capability drift →
- abandonment+6 abandonment hub →
[](https://mcpobservatory.com/servers/github:lastmile-ai/mcp-agent/security) Heuristic, inferred signals — false positives (legitimately powerful tools, forks, language ports) are expected. Treat each as "review this", not a verdict. See the ecosystem-wide picture on the security hub, or the fleet security of lastmile-ai.