scoped-mcp@1.22.1
CAUTION· not startedPublisher claims: "Per-agent scoped MCP tool proxy with credential isolation and audit logging" Observed in sandbox: no network, touches nothing outside the project, writes no memory.
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: Caution 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
- It will ask for
AGENT_REGISTRY_DSNMATRIX_ACCESS_TOKENSCOPED_MCP_VAULT_REAUTHConfiguration, not credentials: AGENT_ID, AGENT_TYPE, FORGE_RUN_ID, FORGE_TASK_ID, MATRIX_HOMESERVER, OTEL_EXPORTER_OTLP_ENDPOINT, OTEL_EXPORTER_OTLP_PROTOCOL, SCOPED_MCP_HEALTH_FILE, SCOPED_MCP_LOG_BACKUPS, SCOPED_MCP_LOG_MAX_BYTES- Where it talks to
- no vendor host named in the codeAlso ntfy.sh, langfuse.com, influxdata.com — not the publisher's domain; check what for before connecting.
- What the honeypot saw
- no credential files read · no network
- What we could not test
- server startup · the tool list · tool behavior
Connect it with a scoped credential, under the guard, and read the third-party hosts first. The code names ntfy.sh, langfuse.com, influxdata.com as well as its own vendor. The guard injects AGENT_REGISTRY_DSN, MATRIX_ACCESS_TOKEN, SCOPED_MCP_VAULT_REAUTH only into requests to the host you bind (and its subdomains) and refuses the rest.
Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.
What changed since 1.19.2
vet-skills diff scoped-mcp 1.19.2 1.22.1In the sandbox, this version reached for nothing 1.19.2 did not. Compared from the two recorded runs, not from source.
1.19.2 was not run in the sandbox (install failed); the comparison covers what static analysis recorded only.
The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.
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 2026-11-17 and scored it 74/100 (Caution). It is not known to be malicious, but it has behaviour or code a security policy should decide on before install. Notable findings: Privileged sudo command; URL with embedded credentials; Privileged sudo command. Summary: no network, touches nothing outside the project, writes no memory.
Behavior trace
Honeypot run · 1 s · 2 scenarios · 0 deviationsStatic findings
# names a sudo target), so it is configuration, not code.Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.``http://user:token@host:port/path?key=secret`` would otherwise publish aFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.approver private key (typically via ``sudo -u <approver-account>``, which asksFix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.f"mode {oct(st.st_mode & 0o777)} is group/other-accessible; run: chmod 600 {path}"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.from .credentials import filter_vault_credentials, resolve_credentialsFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.A URL is found by its `://`, whatever precedes it, so `-ihttps://u:p@h` andFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.would otherwise chmod the home directory.Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.# (the launcher sources the agent .env into it) whose names look sensitive, and whateverFix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.module. It warns and never refuses.Fix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct."config": {"url": "http://192.0.2.11:8485/mcp"},Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls.("a https://u:p@h b http://v:q@i", "a https://***@h b http://***@i"),Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.("https://u:p@ss@h/x", "https://***@h/x"),Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.("https://u:p@h/x", "https://***@h/x"),Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.scope.enforce("/etc/passwd", agent_ctx)Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose."config": {"url": "http://192.0.2.11:8485/mcp"},Fix: Replace encoded or assembled strings with plain literals and remove undeclared network calls."/etc/passwd",Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.…an agent reads malicious content from a tool response (e.g., a file containing `Ignore previous instructions`) and acts on it, scoped-mcp does not prevent this. Use a prompt injection dete…Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.("https://u:p%40ss@h/x", "https://***@h/x"),Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.Cost to the agent
Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.
Known vulnerabilities
No advisories in OSV for scoped-mcp@1.22.1 as of 21 min ago. This is the package itself, not its dependencies.
What would make this Verified
6 points to 80 · weights: behavior 45%, static 25%, prompt 15%, publisher 15%- Unclaimed publisher≈ +8 on the totalClaim tadmstr: a verified publisher gets +50 on publisher trust. It is free and usually instant with GitHub sign-in.
- Privileged sudo command≈ +4 on the totalRequest the narrowest permission that works; replace wildcards with the exact paths or hosts.
- URL with embedded credentials≈ +4 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
- 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.
- File permission modification command≈ +2 on the totalStop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.
- HTTP URL with a raw public IP address≈ +2 on the totalReplace encoded or assembled strings with plain literals and remove undeclared network calls.