@recallai/desktop-sdk@2.0.30
Recall Desktop SDK
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package has a material remote-code and desktop-capture capability whose native payload cannot be verified from this snapshot. Its behavior is consistent with a desktop SDK, but absence of integrity verification warrants a warning rather than a clean result.
Static reason
One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source
Trigger
npm install on macOS or Windows, followed by an application call to init
Impact
A replaced remote archive could run arbitrary native code with the application's permissions; the SDK can request sensitive desktop permissions.
Mechanism
remote native payload download, extraction, and child-process launch
AI rationale
The package has a material remote-code and desktop-capture capability whose native payload cannot be verified from this snapshot. Its behavior is consistent with a desktop SDK, but absence of integrity verification warrants a warning rather than a clean result.
Evidence
package.jsonsetup.jsindex.jsindex.d.tsdesktop_sdk_macos.tardesktop_sdk_win32.tarFrameworksgstreamer-1.0desktop_sdk_macos_exeagent-windows.exe
Network endpoints2
recallai-desktop-sdk-releases.s3.us-east-1.amazonaws.comapi.recall.ai
Decision evidence
public snapshotAI called this Suspicious at 90.0% confidence as Dangerous Capability with low false-positive risk.
Evidence for warning
- The automatic install hook downloads a platform archive from a remote S3 host selected by manifest data.
- The installer removes package-native directories and extracts the downloaded archive without a checksum or signature check.
- Runtime code launches the installed native executable and forwards SDK commands to it.
- The public API includes screen, microphone, accessibility, browser automation, and full-disk-access permissions.
Evidence against
- The installer writes only inside the package directory and has no self-dependency or consumer-project mutation.
- No credential harvesting, shell execution, evaluator, or AI-agent control-surface changes appear in the inspected JavaScript.
- Native process launch occurs through the explicit init API, not during installation.
Behavioral surface
ChildProcessEnvironmentVarsFilesystemNetwork
HighEntropyStringsUrlStrings
Source & flagged code
4 flaggedpackage.jsonView file
•scripts.install = node ./setup.js
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg14"scripts": {
"install": "node ./setup.js",
"validate-types": "tsc --noEmit",
Medium
Suspicious Lifecycle Evidence
The automatic install hook downloads a platform archive from a remote S3 host selected by manifest data.
package.jsonView on unpkg · L14index.jsView file
•matchType = previous_version_dangerous_delta
matchedPackage = @recallai/desktop-sdk@2.0.28
matchedIdentity = npm:QHJlY2FsbGFpL2Rlc2t0b3Atc2Rr:2.0.28
similarity = 0.500
summary = stored previous version shares package body but lacks this dangerous source file
High
Previous Version Dangerous Delta
This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.
Findings
2 High5 Medium5 Low
HighInstall Time Lifecycle Scriptspackage.json
HighPrevious Version Dangerous Deltaindex.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
MediumSuspicious Lifecycle Evidencepackage.json
MediumSuspicious Lifecycle Evidencesetup.js
LowScripts Present
LowFilesystem
LowHigh Entropy Strings
LowUrl Strings
LowNo License
Affected versions and remediation
This report applies to @recallai/desktop-sdk@2.0.30.
See all versions of @recallai/desktop-sdk for other recorded verdicts.
- Review the evidence and your use of @recallai/desktop-sdk@2.0.30 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.