Openlmlib MCP Server
openlmlib
PyPI
v0.2.6
Published by vedant9500 — no publish provenance, so origin is unverified, but the source is public: the repository link below is self-declared yet readable, so you can inspect the code before adopting it.
The grade answers one question — how safe is this server for you to adopt — so it is computed in two auditable stages. Nothing below is an opinion or an LLM's guess; every line is a real term the deterministic engine applied, and the same input always yields the same number.
1. Threat score — 100 − 9.5 = 91. What the published surface and source actually contain:
| Points | What was found | Category |
|---|---|---|
| −9.5 | Unsafe deserialization ×2 MTC-SRC-007 | permissions |
2. Client adoption risk — 91 − 11 = 80. Three small, subtract-only factors that reflect your risk in adopting it — a clean scan proves less on a powerful, unverified or barely-inspectable package, so the grade says so plainly:
| Points | Adoption-risk factor |
|---|---|
| −10 | capability blast radius (critical) — client exposure if the model is manipulated |
| −1 | publisher verification (public source) — no provenance, but the source is public and inspectable |
| ✓ | inspection depth (source) — how much of the target the scan could see |
Capability observations and info notes are shown under Findings but never scored.
Open any row's finding below for the file, line and evidence behind a deduction.
Findings 8
high Reads a sensitive credential path or dumps the environment (openlmlib/mcp_setup.py)MTC-SRC-006
In the server's implementation (`openlmlib/mcp_setup.py:130`): Reading private keys / cloud credentials, or serializing the whole environment, is a sensitive-data source that becomes exfiltration when combined with any egress. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.
Evidence: dict(os.environ)
Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.
Location: server openlmlib/mcp_setup.py
high Tool "evaluate_retrieval" exposes command/code executionMTC-CAP-001
Tool "evaluate_retrieval" appears to run shell commands or evaluate code (keyword "eval"). Arbitrary execution driven by model input is one of the most dangerous MCP capabilities; combined with any untrusted input it becomes RCE.
Fix: Sandbox execution, allowlist commands/arguments, and never pass model output to a shell unescaped.
Location: tool evaluate_retrieval
high Shell/command execution in server code (openlmlib/memory/retrogit_ingest.py)MTC-SRC-002
In the server's implementation (`openlmlib/memory/retrogit_ingest.py:28`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.
Evidence: subprocess.run(
Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.
Location: server openlmlib/memory/retrogit_ingest.py
high Shell/command execution in server code (openlmlib/tui_setup.py)MTC-SRC-002
In the server's implementation (`openlmlib/tui_setup.py:32`): Spawning a shell/process is command-execution capability; with unsanitized tool input it is command injection / RCE. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.
Evidence: subprocess.run(
Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.
Location: server openlmlib/tui_setup.py
medium Untrusted input can drive an external actionMTC-FLOW-005
Untrusted-input tools ([list_sessions, read_messages, tail_messages, read_message_range, list_findings]) co-exist with external-action tools ([send_message, export_to_library, evaluate_retrieval]). A prompt injection could cause unwanted external actions, though no direct sensitive-data leak path was found.
Evidence: untrusted [list_sessions, read_messages, tail_messages, read_message_range, list_findings] → sinks [send_message, export_to_library, evaluate_retrieval]
Fix: Require confirmation for state-changing/egress actions triggered after processing untrusted content.
Location: flow list_sessions → send_message
medium Unsafe deserialization (openlmlib/embeddings.py)MTC-SRC-007
In the server's implementation (`openlmlib/embeddings.py:30`): Deserializing untrusted data with these APIs can execute arbitrary code (a well-known RCE gadget class). This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.
Evidence: pickle.load(
Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.
Location: server openlmlib/embeddings.py
medium Unsafe deserialization (openlmlib/vector_store.py)MTC-SRC-007
In the server's implementation (`openlmlib/vector_store.py:178`): Deserializing untrusted data with these APIs can execute arbitrary code (a well-known RCE gadget class). This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.
Evidence: pickle.load(
Fix: Review this call path: confirm it never receives unsanitized tool input, constrain it, or remove it. Treat a server whose code reaches these sinks as high-capability regardless of what its tools claim.
Location: server openlmlib/vector_store.py
low Mutating tool "evaluate_retrieval" declares no destructiveHintMTC-CAP-005
Tool "evaluate_retrieval" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.
Fix: Declare accurate annotations, and gate destructive tools on user confirmation regardless.
Location: tool evaluate_retrieval
Tools 60
Each tool and what it can reach — statically extracted from the published source.
evaluate_retrievalruns code / shellexport_to_librarynetwork egresslist_findingsingests untrusted inputlist_sessionsingests untrusted inputread_message_rangeingests untrusted inputread_messagesingests untrusted inputsend_messagenetwork egresstail_messagesingests untrusted inputcheck_contextno sensitive capabilitycollab_claim_taskno sensitive capabilitycreate_from_templateno sensitive capabilitycreate_sessionno sensitive capabilitydelete_findingno sensitive capabilityend_sessionno sensitive capabilityget_agent_sessionsno sensitive capabilityget_artifactno sensitive capabilityget_findingno sensitive capabilityget_model_detailsno sensitive capabilityget_observationsno sensitive capabilityget_session_stateno sensitive capabilityget_templateno sensitive capabilityget_usage_analyticsno sensitive capabilitygrep_artifactsno sensitive capabilitygrep_messagesno sensitive capabilityhealthno sensitive capabilityhelp_collabno sensitive capabilityhelp_libraryno sensitive capabilityingest_git_historyno sensitive capabilityinit_libraryno sensitive capabilityinject_contextno sensitive capabilityjoin_sessionno sensitive capabilityleave_sessionno sensitive capabilitylist_artifactsno sensitive capabilitylist_modelsno sensitive capabilitylist_templatesno sensitive capabilitylog_observationno sensitive capabilitymemory_timelineno sensitive capabilitypoll_messagesno sensitive capabilityquery_memoryno sensitive capabilityrecommended_modelsno sensitive capabilityretrieve_contextno sensitive capabilityretrieve_findingsno sensitive capabilitysave_artifactno sensitive capabilitysave_findingno sensitive capabilitysave_finding_autono sensitive capabilitysearch_findingsno sensitive capabilitysearch_knowledgeno sensitive capabilitysearch_memoryno sensitive capabilitysearch_sessionsno sensitive capabilitysession_contextno sensitive capabilitysession_endno sensitive capabilitysession_recapno sensitive capabilitysession_relationshipsno sensitive capabilitysession_startno sensitive capabilitysession_statisticsno sensitive capabilitysessions_summaryno sensitive capabilitystart_researchno sensitive capabilityterminate_sessionno sensitive capabilitytopic_contextno sensitive capabilityupdate_session_stateno sensitive capability
Toxic flows 1
Cross-tool combinations that form a data-exfiltration primitive (untrusted input → sensitive source → external sink).
What this scan could not see
- Tools were statically extracted from the published source (60 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Versions
Scan history per published version. The engine is deterministic — the same version always yields the same score, so a changed score means the package itself changed.
| Version | Score | Findings | Engine | Scanned |
|---|---|---|---|---|
v0.2.6 latest |
B 80/100 | 8 | 1.8.0 | 2026-07-23 |
Embed this score
Show this server's live Trust Score in your README, docs or website. The badge is served straight from the registry and updates automatically after every rescan — no API key needed. It links back to this page, so anyone who sees the grade can also read the findings behind it instead of taking a number on faith.
Verify this score yourself
The score above is reproducible: the same package version always yields the same result. Run it locally or over the free API — no account, no LLM, fully deterministic.
npx mcptrustchecker scan openlmlib --online --registry pypiMore in Developer Tools
Security scan results for the 1stay MCP server.
Security scan results for the Adguard Home MCP server.
Security scan results for the Ado Browser MCP server.
Security scan results for the Adobe Experience Dev MCP server.
Security scan results for the Aemet MCP server.
Security scan results for the Agent MCP server.