registry / autopilot-ai / 2.4.1
autopilot-ai@2.4.1
autopilot (formerly kimiflare): a terminal coding agent for any model with your own key: OpenRouter, Anthropic, OpenAI, Gemini, Bedrock, Vertex, Ollama and more. Per-turn cost, web search, GitHub, headless browser, LSP, MCP, local memory.
AI Security Review
scanned 13d ago · by lpm-firewall-aiSource inspection found no malicious behavior supporting the scanner’s labels. The apparent credential transfer belongs to an explicitly invoked inbox feature, telemetry requires a supplied endpoint, and the package has no install hook.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
The CLI starts for normal use; telemetry requires a configured endpoint, while inbox access requires the user to open the inbox and submit its form.
Impact
These inspected paths support intended product behavior; they do not establish covert credential theft, automatic installation activity, or hidden remote code execution.
Mechanism
The inbox sends the handle and secret supplied in its form to the inbox service. Telemetry sends logs only after an endpoint is configured.
AI rationale
Source inspection found no malicious behavior supporting the scanner’s labels. The apparent credential transfer belongs to an explicitly invoked inbox feature, telemetry requires a supplied endpoint, and the package has no install hook.
Evidence
package.jsondist/index.jsdist/sdk/index.jsdist/sdk/index.d.ts.lpm-dependency-evidence-425ef879-c3c1-4fca-bd19-c88348115258/0/npm-linux/package.json.lpm-dependency-evidence-425ef879-c3c1-4fca-bd19-c88348115258/0/npm-linux/dist/index.js.lpm-dependency-evidence-425ef879-c3c1-4fca-bd19-c88348115258/1/npm-linux/package.json.lpm-dependency-evidence-425ef879-c3c1-4fca-bd19-c88348115258/2/npm-linux/package.json.lpm-dependency-evidence-425ef879-c3c1-4fca-bd19-c88348115258/2/npm-linux/lib/index.js
Network endpoints1
hello.kimiflare.com
Decision evidence
public snapshotAI called this Clean at 88.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json has no install hook; its prepublishOnly script runs the build, and it declares the package entrypoints and dependencies.
- OpenTelemetry export is disabled unless an endpoint is supplied through options or KIMIFLARE_OTEL_ENDPOINT; absent that, configuration remains null.
- The CLI’s inbox prompts for a handle and secret, then uses them to check the inbox; this is an explicit feature interaction.
- The SDK exports agent-session and run-management APIs as part of the package’s stated coding-agent purpose.
- The inspected dependency manifests identify the MCP and SQLite libraries; better-sqlite3’s install script is native prebuild or node-gyp setup.
Evidence against
- The inbox sends the entered handle and secret to a fixed service endpoint, but the source ties this transfer to the user-requested inbox flow.
- The bundle contains isolated evaluation and shell execution capabilities for agent features; the inspected excerpts do not show an automatic install-time trigger or hidden payload execution.
Behavioral surface
ChildProcess
Source & flagged code
15 flaggeddist/index.jsView file
41Source sends the broad process environment to a literal external destination.
L41: function defaultLogDir() {
L42: const xdg = process.env.XDG_CONFIG_HOME || join(homedir(), ".config");
L43: return join(xdg, "kimiflare", "logs");
...
L58: if (currentStream) {
L59: currentStream.end();
L60: currentStream = null;
...
L188: function setFetchForTesting(f) {
L189: fetchImpl = f ?? ((url, init) => fetch(url, init));
L190: }
...
L989: import { createSign } from "crypto";
L990: import { execFile } from "child_process";
L991: import { readFile } from "fs/promises";
Critical
Hardcoded Runtime Data Exfiltration
Source sends credentials or rich application records to a package-controlled external receiver enabled by default.
dist/index.jsView on unpkg · L411055const iat = Math.floor(now2 / 1e3);
L1056: const payload = enc({ iss: sa.client_email, scope: SCOPE, aud: sa.token_uri ?? "https://oauth2.googleapis.com/token", iat, exp: iat + 3600 });
L1057: const signature = createSign("RSA-SHA256").update(`${header}.${payload}`).sign(sa.private_key).toString("base64url");
...
L1062: const tokenUrl = creds.type === "service_account" ? creds.token_uri ?? "https://oauth2.googleapis.com/token" : "https://oauth2.googleapis.com/token";
L1063: const body = creds.type === "service_account" ? new URLSearchParams({ grant_type: "urn:ietf:params:oauth:grant-type:jwt-bearer", assertion: signJwt(creds, now2) }) : new URLSearchP...
L1064: const res = await fetchImpl2(tokenUrl, { method: "POST", headers: { "Content-Type": "application/x-www-form-urlencoded" },
Critical
Command Output Exfiltration
Findings
6 Critical8 High4 Medium6 Low
CriticalHardcoded Runtime Data Exfiltrationdist/index.js
CriticalCommand Output Exfiltrationdist/index.js
CriticalDns Exfiltrationdist/index.js
CriticalRemote Response Code Executiondist/index.js
CriticalTrigger Reachable Dangerous Capabilitydist/index.js
CriticalPrevious Version Dangerous Deltadist/sdk/index.d.ts
HighChild Processdist/index.js
HighShelldist/index.js
HighEval
HighSame File Env Network Executiondist/index.js
HighCredential Redirect Persistencedist/index.js
HighRemote Agent Bridgedist/index.js
Affected versions and remediation
This report applies to autopilot-ai@2.4.1.
See all versions of autopilot-ai for other recorded verdicts.
- Review the evidence and your use of autopilot-ai@2.4.1 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.