gh-fine-grained-pat

No direct evidence of classic malware behavior (no keylogging, clipboard theft, reverse shells, external exfiltration, or encoded payloads) is present. However, the code increases attack surface by launching Chrome with `--remote-allow-origins=*` and by connecting/tearing down based on a port that can be influenced via environment variables. If an attacker can control env vars or ensure a malicious service listens on the selected port, it could lead to unintended CDP connections and termination of arbitrary processes. Otherwise, the risk is moderate and primarily operational/sandboxing related.

This module is a high-privilege GitHub token management automation tool that can generate and then explicitly harvest the resulting github_pat_… secret from the browser DOM, returning it to the caller. It also automates token creation scope changes and token deletion, and can optionally dynamic-import and run an additional “autonomous sudo” confirmation path controlled by GH_PAT_AUTONOMOUS. No obvious overt malware behaviors (network exfiltration, shells, or destructive system actions) are present in the shown fragment, but the credential-capture capability and privileged automation make it a significant security risk and a key supply-chain review target—especially the contents and behavior of ./autosudo.mjs and how returned tokens are stored/used downstream.

SUSPICIOUS: the skill is purpose-aligned, but its footprint is inherently high-trust because it automates creation of real GitHub credentials, persists a session cookie, and optionally handles full web-auth secrets. No clear third-party exfiltration or malicious install path is shown, so this is not confirmed malware, but it carries meaningful security risk due to credential scope and autonomous account actions.

This code is a straightforward local IPC credential/blob cache agent, not overt malware. However, it exposes substantial security risk in multi-process environments: the Unix socket RPC is completely unauthenticated, allowing any local connector to read/overwrite stored 'blob' values and to trigger shutdown via an unauthenticated 'stop' operation (server.close + process.exit). Additionally, the server has unbounded request buffering (possible DoS). Risk is therefore dominated by local privilege/isolation requirements (socket permissions) rather than by remote exploitation or embedded payload behavior.

This fragment does not contain malware or obfuscation; it is a declarative authorization scope. However, it requests a highly privileged token (administration write and contents write) across all repositories, which substantially increases impact if provisioned token credentials are leaked or misused. Require least-privilege scoping, short lifetimes with clear units, and strong secret handling/auditing around the provisioning and usage workflows.