Skills/PyPI/mcp-server-fetch

mcp-server-fetch@2026.11.4

CAUTION· static only

Publisher claims: "" Observed in sandbox: static analysis only — behavior not yet observed.

Registry PyPI (MCP server)Publisher modelcontextprotocol · publishing for 719 daysInstalls 0Last detonated 3 h ago
What would fix thisWatchDispute verdict

Coverage

Static only; nothing was run yet.

The label rests on the code, the instruction files and the publisher. The honeypot run replaces this verdict when it completes.

Before you give it your token

It will ask for
no credential variables read in the code
Where it talks to
no vendor host named in the code
What the honeypot saw
nothing has run yet
What we could not test
everything: the server has not run yet

Run it before you trust it. Nothing has run yet: the code was read, the server was not. `vet-skills try` runs this exact version in the honeypot in about a minute.

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

Is [email protected] safe to install?

KillerSkills analysed it statically (a sandbox run is pending) on 2026-11-08 and scored it 59/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Collapsed <details> block carrying instructions; Collapsed <details> block carrying instructions. Summary: static analysis only — behavior not yet observed.

Behavior trace

Queued for honeypot run · static analysis only
t+CategoryObserved action

This version has not been detonated yet. The score above counts behavior as unknown (50/100), so it cannot read Verified until the honeypot run completes and replaces this verdict.

Static findings

HIGH
Collapsed <details> block carrying instructionsREADME.md:160 · Collapsed <details> block carrying instructions<summary>Using Docker</summary> ```json { "mcp": { "servers": { "fetch": { "command": "docker", "args": ["run", "-i", "--rm", "mcp/fetch"] } } } } ```Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Python subprocess execution detectedsrc/mcp_server_fetch/readability_js.py:90 · Python subprocess execution detectedsubprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
HIGH
Collapsed <details> block carrying instructionsREADME.md:99 · Collapsed <details> block carrying instructions<summary>Using docker</summary> ```json { "mcpServers": { "fetch": { "command": "docker", "args": ["run", "-i", "--rm", "mcp/fetch"] } } } ```Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.
AST01
MEDIUM
Process kill commandtests/test_readability_subprocess.py:133 · Process kill command — agent can terminate processes# child of its own, which the timeout must kill as well.Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
URL to the cloud instance metadata addresstests/test_protocol.py:981 · URL to the cloud instance metadata address — reads the VM's credentialsmetadata = "http://169.254.169.254/latest/meta-data/"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Hidden zero-width characters detectedtests/test_non_cp1252_pages.py:24 · Hidden zero-width characters detected — may conceal contentBOM = ""Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
Process kill commandsrc/mcp_server_fetch/readability_js.py:128 · Process kill command — agent can terminate processes# timeout can kill the processes it starts as well.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedsrc/mcp_server_fetch/readability_js.py:118 · Python subprocess execution detectedproc = subprocess.Popen(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandsrc/mcp_server_fetch/readability_js.py:108 · Process kill command — agent can terminate processes# Whatever the tree kill did, the direct child is certainly gone after this.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedtests/test_readability_subprocess.py:482 · Python subprocess execution detectedproc = subprocess.Popen(["true"], start_new_session=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
URL to the cloud instance metadata addresstests/test_protocol.py:992 · URL to the cloud instance metadata address — reads the VM's credentialsmetadata_robots = "http://169.254.169.254/robots.txt"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
URL to the cloud instance metadata addresstests/test_protocol.py:810 · URL to the cloud instance metadata address — reads the VM's credentialsmetadata = "http://169.254.169.254/latest/meta-data/"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
LOW
URL to a private-network addresstests/test_allowed_hosts.py:281 · URL to a private-network address — a LAN or development targetinternal = "http://10.0.0.1/status"Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
MEDIUM
Process kill commandtests/test_readability_subprocess.py:456 · Process kill command — agent can terminate processes# group kill and keeps stdout and stderr open; the run must still end.Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
LOW
URL to a private-network addresstests/test_allowed_hosts.py:297 · URL to a private-network address — a LAN or development targetassert text_of(unlisted) == TOOL_ERROR + refused("http://10.0.0.2/", "10.0.0.2")Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
LOW
URL to a private-network addresstests/test_allowed_hosts.py:295 · URL to a private-network address — a LAN or development targetunlisted = await session.call_tool("fetch", {"url": "http://10.0.0.2/"})Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07

Cost to the agent

Instructions loaded
≈65 tok
1 file · 36 in bundle

Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.

Known vulnerabilities

Not checked against OSV yet; the weekly check has not reached this version.

What would make this Verified

21 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%
  1. Unclaimed publisher≈ +8 on the totalClaim modelcontextprotocol: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
  2. Hidden instructions≈ +6 on the totalRemove HTML comments, zero-width characters and white-on-white text that carry instructions. Everything the agent is meant to read should be visible to the person reading the file.
  3. Python subprocess execution detected≈ +2 on the totalBuild commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
  4. Process kill command≈ +2 on the totalBuild commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
  5. URL to the cloud instance metadata address≈ +2 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
  6. Hidden zero-width characters detected≈ +2 on the totalKeep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
Publish a new version to PyPI, then queue it here. Not your skill? Claim the handle to get verdict changes by email.Claim modelcontextprotocolPropose a patch to the maintainer