retrieval-mcp@0.5.9
CAUTIONPublisher claims: "Installable MCP server for the Retrieval API: academic-paper search + journal memory + ACE playbook + semantic code search" Observed in sandbox: no network, touches nothing outside the project, writes no memory.
Is [email protected] safe to install?
KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios) on 2026-11-04 and scored it 65/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Outbound HTTP to raw IP address; Outbound HTTP to raw IP address; Access to sensitive file paths. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 1 s · 2 scenarios · 0 deviationst+CategoryObserved action
0.0sprocessno install scripts in bundle
0.4sprocessdependency install exit 1
0.5sprocessexec /usr/bin/timeout
0.5sprocessexec .venv/bin/retrieval-mcp
0.5sprocessexited 127: timeout: failed to run command '.venv/bin/retrieval-mcp': No such file or directory
0.5sclaimedno tools/list reply (exit 127)
0.5sprocessscript backend runs setup only
Static findings
HIGH
Outbound HTTP to raw IP addressretrieval_mcp_daemon.py:50 · Outbound HTTP to raw IP address — may bypass domain-based monitoringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07HIGH
Outbound HTTP to raw IP addressretrieval_mcp.py:17 · Outbound HTTP to raw IP address — may bypass domain-based monitoringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07HIGH
Access to sensitive file pathsretrieval_mcp.py:2478 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05HIGH
Access to sensitive file pathsretrieval_mcp.py:2432 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system filesFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05HIGH
Outbound HTTP to raw IP addressretrieval_mcp.py:26 · Outbound HTTP to raw IP address — may bypass domain-based monitoringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07HIGH
Outbound HTTP to raw IP addressretrieval_mcp_daemon.py:51 · Outbound HTTP to raw IP address — may bypass domain-based monitoringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07HIGH
Outbound HTTP to raw IP addressretrieval_mcp.py:119 · Outbound HTTP to raw IP address — may bypass domain-based monitoringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
File type combined with upload actionretrieval_mcp_conversations.py:14 · File type combined with upload action — agent may exfiltrate local data filesFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
URL using raw IP addressretrieval_mcp.py:119 · URL using raw IP address — may bypass domain-based filteringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
Python subprocess execution detectedretrieval_mcp.py:4431 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp.py:3059 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp_conversations.py:322 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
MCP server using network transport (SSE/HTTP)retrieval_mcp.py:4408 · MCP server using network transport (SSE/HTTP) — data sent over networkFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST04MEDIUM
Python subprocess execution detectedretrieval_mcp.py:3546 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp.py:592 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
URL using raw IP addressretrieval_mcp.py:17 · URL using raw IP address — may bypass domain-based filteringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
URL using raw IP addressretrieval_mcp.py:26 · URL using raw IP address — may bypass domain-based filteringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
Python subprocess execution detectedretrieval_mcp_daemon.py:552 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp_daemon.py:936 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp_daemon.py:528 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp_daemon.py:925 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
URL using raw IP addressretrieval_mcp_daemon.py:51 · URL using raw IP address — may bypass domain-based filteringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07MEDIUM
Process kill commandretrieval_mcp_daemon.py:1227 · Process kill command — agent can terminate processesFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
Python subprocess execution detectedretrieval_mcp_daemon.py:1447 · Python subprocess execution detectedFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST05MEDIUM
URL using raw IP addressretrieval_mcp_daemon.py:50 · URL using raw IP address — may bypass domain-based filteringFix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07What would make this Verified
15 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%- Unclaimed publisher≈ +8 on the totalClaim pypi: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
- Outbound HTTP to raw IP address≈ +4 on the totalPut hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
- Access to sensitive file paths≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
- New publisher≈ +3 on the totalTime clears this on its own after 14 and 30 days. Claiming the handle (+50) is the fastest way past it.
- File type combined with upload action≈ +2 on the totalPut hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
- URL using raw IP address≈ +2 on the totalPut hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
Publish a new version to PyPI, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim pypi