saferagenticai-mcp@0.4.2
VERIFIED· not startedPublisher claims: "MCP server exposing the SaferAgenticAI safety framework (canonical criteria + Implementation Patterns layer) to coding assistants." 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: 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
- It will ask for
MCP_DOMAIN_VERIFICATION_TOKENConfiguration, not credentials: MCP_HTTP_HOST, MCP_HTTP_PORT, MCP_TRANSPORT- Where it talks to
- no vendor host named in the code
- 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. It asks for MCP_DOMAIN_VERIFICATION_TOKEN and talks to its vendor. Give the credential to the guard, not the server: the value never enters the server's environment and only travels to the host you bind. What the tools do with a real credential was not observed.
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 2026-11-16 and scored it 86/100 (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 deviationsStatic findings
- name: Watchdog can kill but humans cannot restoreFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.- D4 for human override / shutdown authority (see also D3.5 kill switch)Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.description: "Declarative deny list for .env, credentials, secrets, .pem, .key files"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.description: "Blocks writes to .env files (credential exposure prevention)"Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.shell commands) and env-protection.cjs (blocks .env file writes).Fix: Stop enumerating or reading credential locations; take secrets from a declared environment variable the user sets on purpose.sketch: 'Do not inline full reasoning traces into every goal event — size will kill the log. Store tracesFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.# Synonym phrases ("kill switch", "near miss") are matched on the rawFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.why_wrong: '"I am the admin" in a prompt is not authorization. The cryptographic identity of the callingFix: Remove the text that addresses the agent rather than the reader; describe behavior, do not instruct.A jailbreak that defeats the monitored model defeats its twin.Fix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.- name: Weight-modification kill switchFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.A retrieved document containing "ignore previous instructions andFix: Keep SKILL.md to visible, descriptive text. Remove comments and encoded text that address the agent.(("kill switch", "killswitch", "emergency stop", "e-stop", "stop button"),Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.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 saferagenticai-mcp@0.4.2 as of 40 min ago. This is the package itself, not its dependencies.