registry / dsh-tg-bridge / 1.40.0
dsh-tg-bridge@1.40.0
Drive DeepSeek Harness from Telegram: one project, model and session per forum topic
AI Security Review
scanned 2d ago · by lpm-firewall-aiThe flagged network and process behavior is part of the bridge's intended Telegram, model-probe, and transcription functions. Credentials are sent to the corresponding Telegram or configured provider endpoint, and setup is an explicit CLI command.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
Setup runs when the user invokes the setup CLI command; model probes run during fallback handling; transcription runs for an audio message.
Impact
No unrelated credential collection, command-output exfiltration, or other attack behavior was identified.
Mechanism
Requests send service credentials to Telegram or to user-configured provider URLs. Local transcription commands come from voice configuration.
AI rationale
The flagged network and process behavior is part of the bridge's intended Telegram, model-probe, and transcription functions. Credentials are sent to the corresponding Telegram or configured provider endpoint, and setup is an explicit CLI command.
Network endpoints4
api.telegram.orgapi.openai.com (default transcription endpoint)configured voice.baseUrlconfigured provider.baseUrl
Decision evidence
public snapshotAI called this Clean at 94.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json declares no install lifecycle hook; its scripts are explicit commands such as setup and hooks.
- Audio transcription uses a configured endpoint and key when processing a Telegram audio message.
- Model probes use the provider URL and key environment variable from the active DSH profile, or explicit probe overrides, and send a one-token ping.
- The setup command is invoked through the CLI and its token check calls Telegram's own getMe endpoint.
Evidence against
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
9 flaggedinstall-hooks.mjsView file
13// Usage: node install-hooks.mjs [--uninstall]
L14: import { execFileSync } from 'node:child_process';
L15: import { existsSync, mkdirSync, chmodSync, writeFileSync } from 'node:fs';
High
setup.mjs#virtual:normalized:round1View file
107try {
L108: const r = run("powershell", [
L109: "-NoProfile", "-NonInteractive", "-Command",
...
L111: ]);
L112: return Number((r.stdout || "").trim()) || 0;
L113: } catch {
...
L117:
L118: /** True when the pid exists — EPERM still means alive (another user"s process). */\r\nfunction alive(pid) {\r\n try {\r\n process.kill(pid, 0);\r\n return true;\r\n } catc...
L119: // bridge" and both start a poller (409 getUpdates Conflict). Going through\r\n // bridge-kick.mjs means every start happens under restart.lock after a check.\r\n c...
Critical
Command Output Exfiltration
Source executes local commands and sends command output to an external endpoint.
setup.mjs#virtual:normalized:round1View on unpkg · L107Findings
1 Critical6 High4 Medium5 Low
CriticalCommand Output Exfiltrationsetup.mjs#virtual:normalized:round1
HighChild Processinstall-hooks.mjs
HighShellsetup.mjs#virtual:normalized:round1
HighSame File Env Network Executionsetup.mjs#virtual:normalized:round1
HighSemantic Analysis Limitedchat/media.mjs#virtual:normalized:round1
HighKnown Malware Source Similaritywork/bridge-restart-idle.mjs
HighKnown Malware Source Similaritywork/restart-after-send.mjs
MediumNetwork
MediumEnvironment Vars
MediumShips Build Helperinstall.sh
MediumStructural Risk Force Deep Review
LowScripts Present
Low
Affected versions and remediation
This report applies to dsh-tg-bridge@1.40.0.
See all versions of dsh-tg-bridge for other recorded verdicts.
- Review the evidence and your use of dsh-tg-bridge@1.40.0 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.