registry / savant-code / 0.0.35
savant-code@0.0.35
Multi-Agent AI Coding Assistant with ECHO Protocol
AI Security Review
scanned 3d ago · by lpm-firewall-aiThe package is a CLI launcher that downloads and starts its own release binary, and its only shell execution is a local CPU-feature probe. The flagged shell commands are design reference content, while optional PostHog reporting requires explicitly supplied environment settings.
Static reason
One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source
Trigger
The user invokes the savant-code CLI.
Impact
No malicious impact identified. Optional update-failure telemetry requires PostHog settings in the environment.
Mechanism
The launcher checks for its binary, downloads official release assets when needed, and starts the binary; its Windows shell call checks CPU support.
AI rationale
The package is a CLI launcher that downloads and starts its own release binary, and its only shell execution is a local CPU-feature probe. The flagged shell commands are design reference content, while optional PostHog reporting requires explicitly supplied environment settings.
Evidence
~/.config/savant/<packageName>~/.config/savant/<packageName>-metadata.json~/.config/savant/.download-temp
Network endpoints3
registry.npmjs.org/${packageName}/latestgithub.com/savant0x/savant-code/releases/downloadPostHog host configured by SAVANT_CODE_POSTHOG_HOST or NEXT_PUBLIC_POSTHOG_HOST_URL
Decision evidence
public snapshotAI called this Clean at 91.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The package has CLI bin entries and only prepack/postpack scripts; it declares no install lifecycle hook.
- The Windows child-process call runs a local PowerShell CPU-feature check and reads only its Boolean result.
- The launcher downloads the product's own release assets from GitHub, with an environment override for custom release servers.
- The launcher bootstraps and runs its binary when the user invokes the CLI.
- Update-failure telemetry is sent only when both a PostHog API key and host are provided through environment variables.
- The flagged curl-to-shell commands appear as literal install snippets in design reference JSON.
Evidence against
- The CLI fetches and executes a binary during normal startup, and may stage updates in the background.
- When configured for PostHog, the failure event includes the home-directory path in its distinct ID.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetwork
Source & flagged code
6 flaggedlauncher.jsView file
•matchType = previous_version_dangerous_delta
matchedPackage = savant-code@0.0.22
matchedIdentity = npm:c2F2YW50LWNvZGU:0.0.22
similarity = 0.667
summary = stored previous version shares package body but lacks this dangerous source file
Critical
Previous Version Dangerous Delta
This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
launcher.jsView on unpkg2L3: const { spawn, execFileSync } = require('child_process')
L4: const crypto = require('crypto')
High
index.jsView file
2L3: const fs = require('fs')
L4: const path = require('path')
Medium
Findings
1 Critical4 High3 Medium4 Low
CriticalPrevious Version Dangerous Deltalauncher.js
HighChild Processlauncher.js
HighSame File Env Network Executionlauncher.js#virtual:normalized:round1
HighCommand Output Exfiltrationlauncher.js#virtual:normalized:round1
HighSemantic Analysis Limitedlauncher.js#virtual:normalized:round1
MediumDynamic Requireindex.js
MediumNetwork
MediumEnvironment Vars
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
Affected versions and remediation
This report applies to savant-code@0.0.35.
See all versions of savant-code for other recorded verdicts.
- Review the evidence and your use of savant-code@0.0.35 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.