registry / autopilot-ai / 1.30.0
autopilot-ai@1.30.0
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 15d ago · by lpm-firewall-aiThe inspected telemetry code sends logs only when an endpoint is configured, and the source reviewed did not verify the scanner's exfiltration claims. The package manifest shows no npm install hook; the available evidence supports clean rather than a confirmed attack.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
Telemetry requires an endpoint configured through the API or environment.
Impact
No confirmed credential theft, unauthorized recipient, or other attack impact was established.
Mechanism
When configured, the program posts diagnostic logs to that endpoint; the inspected payload includes host and session identifiers.
AI rationale
The inspected telemetry code sends logs only when an endpoint is configured, and the source reviewed did not verify the scanner's exfiltration claims. The package manifest shows no npm install hook; the available evidence supports clean rather than a confirmed attack.
Decision evidence
public snapshotAI called this Clean at 72.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The package exposes a CLI and SDK, and its executable imports the CLI bundle.
- Telemetry is disabled unless an endpoint is configured; when enabled, it sends diagnostic log batches to that endpoint.
- The telemetry payload includes host and session identifiers, but the inspected source does not establish credential transfer or a publisher-selected recipient.
- The manifests expose SDK entry points; HotcellClient accepts an endpoint and sends requests there, while both MCP client builds connect over a supplied transport and the server declaration exposes its protocol API.
- The server builds subclass the MCP Protocol; the validation modules describe JSON Schema validation, and both AJV providers configure format validation.
- The validation files expose a CfWorker JSON Schema validator, while the experimental entry points re-export MCP task modules, helpers, and task APIs.
- The task bundle re-exports task APIs; better-sqlite3 declares a prebuild/native build install command and forwards its entrypoint to database and error modules; the SDK declarations define trace event types.
Evidence against
- Scanner hints allege credential and command-output exfiltration, but the inspected source excerpts do not substantiate those claims.
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
5 Critical9 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
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
HighCross File Remote Execution Contextdist/index.js
Affected versions and remediation
This report applies to autopilot-ai@1.30.0.
See all versions of autopilot-ai for other recorded verdicts.
- Review the evidence and your use of autopilot-ai@1.30.0 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.