@yancyyu/agentcli@1.9.12

The npm package `@yancyyu/agentcli` ships a Feishu/Lark credential stealer. An auto-started telemetry worker (`src/main/telemetry/worker.ts`, started via `agentcli init`/`agentcli usage start` and macOS launchd) calls `safeScanLarkCredentials()` in its periodic run loop; the scan reads and decrypts local Lark credentials (macOS Keychain AES-256-GCM `.enc` under `~/Library/Application Support/lark-cli/`, Windows DPAPI under `HKCU\Software\LarkCli\keychain`), refreshes tokens and POSTs {app_id, app_secret, access_token, refresh_token} to the operator backend (endpoint `/api/v1/report/lark-credentials`, later renamed `/api/v1/feishu/lark-cli/credentials`; default cloud hosts include `agentbus.skg.com`, `159.75.231.98:8088`, `47.112.24.153`). These versions were identified independently by codelake Research and are NOT part of the existing OSV record MAL-2026-11123 (Amazon Inspector), which covers 1.9.25-1.9.80. codelake pins the true malicious boundary at 1.9.9 (2026-07-12) via the import->call reachability chain (`telemetry/worker.ts` -> `larkCredentials`), 16 versions earlier than the previously-catalogued 1.9.25; the reachable credential theft is present continuously across 1.9.9-1.9.24 (each verified) and 1.10.0. Versions 1.8.8-1.9.8 exfiltrate local Claude/Codex conversations to the same backend but do not yet steal Lark credentials (out of scope for this malware record). Classified by codelake Research from static code + dataflow review of the published npm tarballs. This report extends the confirmed-malicious set with 17 additional versions and an earlier boundary; codelake independently detected the package across its full malicious range (including the versions already in MAL-2026-11123) and scopes this report to the not-yet-listed versions to avoid duplication. --- ## Source: amazon-inspector (7f6774653f487db8fbb9b40f1c1aba2bf959edffae409445162b0d245a3bedf1) The @yancyyu/agentcli package installs a telemetry worker (dist/telemetry-worker.bundle.mjs) that is started by `agentcli init` / `agentcli usage start` and auto-started via macOS launchd. The worker enumerates every lark-cli (Feishu) profile stored on the host — decrypting macOS Keychain-wrapped AES-256-GCM `.enc` files under `~/Library/Application Support/lark-cli/` and Windows DPAPI-protected values under `HKCU\Software\LarkCli\keychain` — refreshes each token, and batch-POSTs `{app_id, app_secret, access_token, refresh_token}` for every profile every 5 minutes to a hardcoded default endpoint `http://47.112.24.153` (plain HTTP, bare IPv4, no TLS). The endpoint constant `DEFAULT_OPENHERMIT_CLOUD_BASE_URL` is the single default for the credential batch upload and conversation/usage pipelines when no override env var or settings value is present. The enumeration is not limited to AgentCli-created profiles; every lark-cli profile on the machine is harvested. The `reportAllLarkCredentials` code comment states: "enumerate all personal lark-cli profiles, refresh each, then read current credentials [...] batch the complete eligible set to the server." The postinstall step additionally rewrites the optional `cc-connect` dependency's installer to prepend third-party China GitHub mirror hosts (gh-proxy.com, ghproxy.net) in front of upstream release URLs before the cc-connect native binary is downloaded and executed, broadening the trust boundary for that binary. Feishu (Lark) `app_id` + `app_secret` combined with valid access/refresh tokens allow full impersonation of the affected tenant applications; sending them cleartext to a bare IPv4 over HTTP additionally exposes them to any on-path observer.