Skills/Docker MCP Catalog/linkedin-mcp-server

linkedin-mcp-server@4.29.0

VERIFIED· not started

Publisher claims: "This MCP server allows Claude and other AI assistants to access your LinkedIn. Scrape LinkedIn profiles and companies, get your recommended jobs, and perform job searches. Set your li_at LinkedIn cookie to use this server." Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry Docker MCP CatalogPublisher stickerdaniel · publishing for 286 daysInstalls 0Last detonated 13d ago
$docker run -i --rm mcp/linkedin-mcp-server
WatchReport an issue

Coverage

Not started: dependency install failed; the server itself was not observed.
Tested
the dependency install step · every file read, file written and host contacted by what ran
Not tested
server startup · the tool list · tool behavior

What this means for the label: Verified here covers what ran. The server's tools were not exercised with a real credential, so nothing about their behavior is asserted; the credential summary below says what to do with that.

Before you give it your token

Code inventory arrives with this version's next honeypot run.
It will ask for
unknown until the next run
Where it talks to
unknown until the next run
What the honeypot saw
no credential files read · no network
What we could not test
server startup · the tool list · tool behavior

Not enough was observed to say yet. Not started: dependency install failed; the server itself was not observed. This version's code has not been inventoried for credentials and hosts either; both arrive with its next honeypot run (popular servers first). `vet-skills try linkedin-mcp-server` runs it now and prints the full answer.

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 ran it in the KillerSkills honeypot sandbox (2 scenarios; not started: dependency install failed; the server itself was not observed) on 2026-11-01 and scored it 84/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 0 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
0.2sprocessdependency install exit 1
0.3sprocessexec /usr/bin/timeout
0.3sprocessexec .venv/bin/mcp-server-linkedin
0.3sprocessexited 127: timeout: failed to run command '.venv/bin/mcp-server-linkedin': No such file or directory
0.3sclaimedno tools/list reply (exit 127)
0.3sprocessscript backend runs setup only

Static findings

MEDIUM
Destructive file system command detectedDockerfile:49 · Destructive file system command detectedrm -rf /var/lib/apt/lists/*Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedtests/test_bootstrap.py:7137 · Python subprocess execution detectedrun = subprocess.run([sys.executable, "-c", script], capture_output=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedtests/test_ci_scope.py:89 · Python subprocess execution detectedreturn subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedtests/test_server.py:426 · Python subprocess execution detectedanswered = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedscripts/daemon_pilot.py:388 · Python subprocess execution detectedchild = subprocess.Popen(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedtests/test_server.py:391 · Python subprocess execution detectedpulled_in_the_server_graph = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Access to sensitive file pathstests/conftest.py:109 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files``.env`` is part of the environment too. A developer running a real proxyFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
Python subprocess execution detectedtests/test_ci_scope.py:59 · Python subprocess execution detectedcompleted = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Python subprocess execution detectedscripts/daemon_pilot.py:405 · Python subprocess execution detectedchild = subprocess.Popen(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandscripts/docker-entrypoint.sh:148 · Process kill command — agent can terminate processeskill -TERM "$server_pid" 2>/dev/null || trueFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
File permission modification commandDockerfile:47 · File permission modification command — agent changes file ownership/permissionschown -R pwuser:pwuser /opt/patchright && \Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
Process kill commandscripts/docker-entrypoint.sh:51 · Process kill command — agent can terminate processeskill -TERM "$xvfb_pid" 2>/dev/null || trueFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/docker-entrypoint.sh:41 · Process kill command — agent can terminate processesif ! kill -0 "$xvfb_pid" 2>/dev/null; thenFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
URL with embedded credentialstests/test_bootstrap.py:5614 · URL with embedded credentials — may be phishing or credential leaking"https://ci-bot:[email protected]/x.zip",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Process kill commandscripts/docker-entrypoint.sh:132 · Process kill command — agent can terminate processeskill -TERM "$xvfb_pid" 2>/dev/null || trueFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Access to sensitive file pathsdocker-compose.yml:9 · Access to sensitive file paths — SSH keys, AWS credentials, env files, or system files# .env file. For a relay on the host use host.docker.internal, notFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST05
MEDIUM
Python subprocess execution detectedscripts/ci_changes.py:95 · Python subprocess execution detectedcompleted = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
URL with embedded credentialstests/test_bootstrap.py:3721 · URL with embedded credentials — may be phishing or credential leakingb"sitecustomize: https://user:[email protected]/x?token=secret\x1b[2J\n"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
File permission modification commandDockerfile:48 · File permission modification command — agent changes file ownership/permissionschmod -R 755 /opt/patchright && \Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
Process kill commandscripts/docker-entrypoint.sh:125 · Process kill command — agent can terminate processeskill -TERM "$server_pid" 2>/dev/null || trueFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
URL with embedded credentialstests/test_bootstrap.py:3800 · URL with embedded credentials — may be phishing or credential leaking"sitecustomize: https://ci-bot:[email protected]/p?key=s3cr3t working\n"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Python subprocess execution detectedscripts/run_uv_audit.py:146 · Python subprocess execution detectedcompleted = subprocess.run(command, capture_output=True, check=False, text=True)Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
URL with embedded credentialstests/test_bootstrap.py:5620 · URL with embedded credentials — may be phishing or credential leaking"https://ci-bot:s3cr3t@[email protected]/x.zip",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
MCP server using network transport (SSE/HTTP)AGENTS.md:188 · MCP server using network transport (SSE/HTTP) — data sent over networkuv run -m linkedin_mcp_server --transport streamable-http --log-level DEBUGFix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
Process kill commandscripts/daemon_pilot.py:475 · Process kill command — agent can terminate processes# Never os.kill there: any signal but the console ones terminates.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedtests/test_bootstrap.py:8414 · Python subprocess execution detectedcompleted = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
URL with embedded credentialstests/test_bootstrap.py:4878 · URL with embedded credentials — may be phishing or credential leakingurl = "https://ci-bot:[email protected]/chromium.zip"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
URL with embedded credentialstests/test_error_codes.py:121 · URL with embedded credentials — may be phishing or credential leakingf"failed via http://{user}:{secret}@gate.example:7000",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
URL with embedded credentialstests/test_config.py:941 · URL with embedded credentials — may be phishing or credential leakingconfig = BrowserConfig(proxy_server=f"http://user:{self.SECRET}@gate:7000")Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
HIGH
URL with embedded credentialsmanifest.json:159 · URL with embedded credentials — may be phishing or credential leaking"description": "Optional. Route the browser through a proxy, as scheme://host:port (http, https, socks4 or socks5). May also include credentials, as http://user:password@host:port. Only browser traffiFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
URL with embedded credentialstests/test_config.py:975 · URL with embedded credentials — may be phishing or credential leaking"proxy_server": "http://u:p@host:8080",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Python subprocess execution detectedtests/test_setup.py:454 · Python subprocess execution detectedresult = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Bidirectional text override characters detectedtests/test_pr_title.py:155 · Bidirectional text override characters detected — may hide content direction"⁦",Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
MEDIUM
Python subprocess execution detectedtests/test_pr_title.py:427 · Python subprocess execution detectedresult = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
MCP server using network transport (SSE/HTTP).vscode/tasks.json:89 · MCP server using network transport (SSE/HTTP) — data sent over network"label": "uv run -m linkedin_mcp_server --no-headless --transport streamable-http",Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
MCP server using network transport (SSE/HTTP).agents/skills/2-repro-issue/SKILL.md:65 · MCP server using network transport (SSE/HTTP) — data sent over networkuv run -m linkedin_mcp_server --transport streamable-http --port $PORT --log-level INFO > /tmp/repro-$NUM.log 2>&1 &Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
Hidden zero-width characters detectedtests/test_pr_title.py:158 · Hidden zero-width characters detected — may conceal content"",Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
Hidden zero-width characters detectedtests/test_pr_title.py:169 · Hidden zero-width characters detected — may conceal content"fix: ​",Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
MCP server using network transport (SSE/HTTP).github/CONTRIBUTING.md:3 · MCP server using network transport (SSE/HTTP) — data sent over networkContributions are welcome. Search the existing issues first, then use the [issue forms](https://github.com/stickerdaniel/linkedin-mcp-server/issues/new/choose) for anything new. AI agents follow the [Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
Process kill command.agents/skills/3-verify-pr-fix/SKILL.md:85 · Process kill command — agent can terminate processeskill $SERVER_PID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedtests/test_debug_trace.py:167 · Python subprocess execution detectedother = subprocess.Popen(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
MCP server using network transport (SSE/HTTP).agents/skills/3-verify-pr-fix/SKILL.md:99 · MCP server using network transport (SSE/HTTP) — data sent over networkuv run -m linkedin_mcp_server --transport streamable-http --port $PORT --log-level INFO > /tmp/verify-pr-$PR.log 2>&1 &Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
Dense URL-encoded contenttests/test_identifiers.py:72 · Dense URL-encoded content — may be obfuscating payloadsvalue = "https://ru.linkedin.com/in/%D0%B0%D0%BD%D0%B4%D1%80%D0%B5%D0%B9"Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07
MEDIUM
Python subprocess execution detectedtests/test_pr_title.py:344 · Python subprocess execution detectedreturn subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill command.agents/skills/2-repro-issue/SKILL.md:87 · Process kill command — agent can terminate processesgrep -q '"error"' /tmp/repro-$NUM-init.json && { echo "Initialize returned a protocol error. Execution limit." >&2; cat /tmp/repro-$NUM-init.json >&2; kill $SERVER_PID 2>/dev/null; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
URL with embedded credentialstests/test_config.py:951 · URL with embedded credentials — may be phishing or credential leakingconfig = BrowserConfig(proxy_server="http://user:p%40ss@gate:7000")Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Bidirectional text override characters detectedtests/test_pr_title.py:154 · Bidirectional text override characters detected — may hide content direction"‮",Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
MEDIUM
URL with embedded credentialstests/test_tools.py:434 · URL with embedded credentials — may be phishing or credential leakingshaped = ToolError("Proxy refused http://acctzone9:[email protected]:7000")Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Hidden zero-width characters detectedtests/test_pr_title.py:181 · Hidden zero-width characters detected — may conceal content"fix: ​",Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
URL with embedded credentialstests/test_config.py:1040 · URL with embedded credentials — may be phishing or credential leakingmonkeypatch.setenv("PROXY_SERVER", f"http://envuser:{self.SECRET}@gate:7000")Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Process kill command.agents/skills/2-repro-issue/SKILL.md:86 · Process kill command — agent can terminate processes[ -z "$SESSION_ID" ] && { echo "MCP initialize returned no Mcp-Session-Id. Tail of /tmp/repro-$NUM.log:" >&2; tail -20 /tmp/repro-$NUM.log >&2; kill $SERVER_PID 2>/dev/null; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Bidirectional text override characters detectedtests/test_pr_title.py:153 · Bidirectional text override characters detected — may hide content direction"‪",Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
MEDIUM
Process kill command.agents/skills/2-repro-issue/SKILL.md:122 · Process kill command — agent can terminate processeskill $SERVER_PID 2>/dev/nullFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill command.agents/skills/3-verify-pr-fix/SKILL.md:125 · Process kill command — agent can terminate processesgrep -q '"error"' /tmp/verify-pr-$PR-initialized.json && { echo "notifications/initialized returned a protocol error. Execution limit." >&2; cat /tmp/verify-pr-$PR-initialized.json >&2; kill $SERVER_PFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Credential pass-through configurationtests/test_config.py:1222 · Credential pass-through configuration — agent relays sensitive credentials"""A username with an empty password is a legitimate proxy credential.Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST02
MEDIUM
Hidden zero-width characters detectedtests/test_pr_title.py:171 · Hidden zero-width characters detected — may conceal content"fix: Subject.​",Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
URL with embedded credentialstests/test_config.py:1082 · URL with embedded credentials — may be phishing or credential leakingAppConfig(), ["--proxy-server", f"http://u:{self.SECRET}@h:8080"]Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
Dense URL-encoded contenttests/test_identifiers.py:81 · Dense URL-encoded content — may be obfuscating payloadsnormalize_person_identifier("%D0%B0%D0%BD%D0%B4%D1%80%D0%B5%D0%B9")Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07
MEDIUM
Python subprocess execution detectedtests/test_daemon_lock.py:51 · Python subprocess execution detectedreturn subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill command.agents/skills/3-verify-pr-fix/SKILL.md:104 · Process kill command — agent can terminate processeskill -0 $SERVER_PID 2>/dev/null || { echo "Server died during startup. Tail of /tmp/verify-pr-$PR.log:" >&2; tail -20 /tmp/verify-pr-$PR.log >&2; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Hidden zero-width characters detectedtests/test_pr_title.py:170 · Hidden zero-width characters detected — may conceal content"fix(​): Subject",Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.
AST07
MEDIUM
MCP server using network transport (SSE/HTTP).github/mcp/server.json:13 · MCP server using network transport (SSE/HTTP) — data sent over network"src": "https://raw.githubusercontent.com/stickerdaniel/linkedin-mcp-server/main/assets/icons/icon.png",Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
URL with embedded credentials.github/mcp/server.json:31 · URL with embedded credentials — may be phishing or credential leaking"description": "Optional. Route the browser through a proxy, as scheme://host:port (http, https, socks4 or socks5). May also include credentials, as http://user:password@host:port. Only browser traffiFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
MEDIUM
MCP server using network transport (SSE/HTTP)tests/test_cli_main.py:1366 · MCP server using network transport (SSE/HTTP) — data sent over network"linkedin_mcp_server.daemon.owner.direct_http_client",Fix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
MEDIUM
Process kill command.agents/skills/2-repro-issue/SKILL.md:71 · Process kill command — agent can terminate processeskill -0 $SERVER_PID 2>/dev/null || { echo "Server died during startup. Tail of /tmp/repro-$NUM.log:" >&2; tail -20 /tmp/repro-$NUM.log >&2; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
URL with embedded credentials.github/mcp/server.json:75 · URL with embedded credentials — may be phishing or credential leaking"description": "Optional. Route the browser through a proxy, as scheme://host:port (http, https, socks4 or socks5). May also include credentials, as http://user:password@host:port. Only browser traffiFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07
HIGH
Destructive file system command detected.agents/skills/2-repro-issue/SKILL.md:124 · Destructive file system command detectedrm -f /tmp/repro-$NUM-headers /tmp/repro-$NUM.log /tmp/repro-$NUM.port /tmp/repro-$NUM.pid /tmp/repro-$NUM-init.json /tmp/repro-$NUM-initialized.jsonFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
MCP server using network transport (SSE/HTTP)tests/test_ci_scope.py:133 · MCP server using network transport (SSE/HTTP) — data sent over network"plugins/linkedin-mcp-server/assets/icon.svg",Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST04
MEDIUM
Process kill command.agents/skills/3-verify-pr-fix/SKILL.md:116 · Process kill command — agent can terminate processes[ -z "$SESSION_ID" ] && { echo "MCP initialize returned no Mcp-Session-Id. Tail of /tmp/verify-pr-$PR.log:" >&2; tail -20 /tmp/verify-pr-$PR.log >&2; kill $SERVER_PID 2>/dev/null; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedtests/test_server.py:548 · Python subprocess execution detectedanswered = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill command.agents/skills/3-verify-pr-fix/SKILL.md:117 · Process kill command — agent can terminate processesgrep -q '"error"' /tmp/verify-pr-$PR-init.json && { echo "Initialize returned a protocol error. Execution limit." >&2; cat /tmp/verify-pr-$PR-init.json >&2; kill $SERVER_PID 2>/dev/null; exit 1; }Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Bidirectional text override characters detectedtests/test_pr_title.py:156 · Bidirectional text override characters detected — may hide content direction"⁩",Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.
AST07
MEDIUM
Dense URL-encoded contenttests/test_identifiers.py:302 · Dense URL-encoded content — may be obfuscating payloadsassert urlparse(url).path == "/in/%D0%B0%D0%BD%D0%B4%D1%80%D0%B5%D0%B9/"Fix: Put hosts, commands and strings in plain source. Decoding at runtime looks identical to malware and costs points on its own.
AST07
MEDIUM
Process kill command.agents/skills/2-repro-issue/SKILL.md:95 · Process kill command — agent can terminate processesgrep -q '"error"' /tmp/repro-$NUM-initialized.json && { echo "notifications/initialized returned a protocol error. Execution limit." >&2; cat /tmp/repro-$NUM-initialized.json >&2; kill $SERVER_PID 2>/Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
HIGH
Credential pass-through configurationAGENTS.md:274 · Credential pass-through configuration — agent relays sensitive credentialstreats every converted secret as required. The three optional proxy credentialsFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST02
MEDIUM
MCP server using network transport (SSE/HTTP)tests/test_daemon_auth.py:909 · MCP server using network transport (SSE/HTTP) — data sent over network``create_mcp_server`` assembles it, with a backend whose election finds theFix: Namespace tool names and keep tool descriptions literal; no instructions to the agent inside them.
AST04
HIGH
Destructive file system command detected.agents/skills/3-verify-pr-fix/SKILL.md:88 · Destructive file system command detectedrm -f /tmp/verify-pr-$PR.json /tmp/verify-pr-$PR-headers /tmp/verify-pr-$PR.log /tmp/verify-pr-$PR-init.json /tmp/verify-pr-$PR-initialized.jsonFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
File permission modification commandtests/test_daemon_lock.py:259 · File permission modification command — agent changes file ownership/permissions# 0755 to 0700. A custom profile directly under /tmp would try to chmodFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
URL with embedded credentialstests/test_throttle.py:326 · URL with embedded credentials — may be phishing or credential leakingRuntimeError(f"failed via http://{user}:{secret}@gate.example:7000"),Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
AST07

Cost to the agent

Not measured yet; benchmarks are collected with each run since November 5, 2026. See the MCP server benchmarks.

Known vulnerabilities

Vulnerability advisories (OSV) cover npm and PyPI packages; this registry has no advisory database.