registry / @netnodeag/kraftwerk / 0.56.0
@netnodeag/kraftwerk@0.56.0
Open-source agentic workspace for teams — where people and AI agents get work done together: agents (Claude Code, Codex, Pi), skills, shared knowledge, repeatable workflows, and deterministic verification
AI Security Review
scanned 17d ago · by lpm-firewall-aiSource inspection found no install-time attack behavior: the manifest’s only lifecycle entry is a prepublishOnly build command. The scanner’s execution and Unicode hints correspond to runtime features and bundled editor code, with no concrete attack established.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
Publishing runs the prepublishOnly build; runtime commands require a configured feature or an inspector workflow action.
Impact
No confirmed unauthorized mutation, credential transfer, or payload execution was established.
Mechanism
The manifest has no install hook. Inspected runtime code reads skills and run records, records local instance metadata, and launches configured commands or workflows.
AI rationale
Source inspection found no install-time attack behavior: the manifest’s only lifecycle entry is a prepublishOnly build command. The scanner’s execution and Unicode hints correspond to runtime features and bundled editor code, with no concrete attack established.
Evidence
package.jsondist/inspector/skills.jsdist/inspector/runs.jsinspector/dist/assets/editor-B3IL5KL7.jsinspector/dist/assets/index-BPl3rBgX.jsinspector/dist/assets/dist-H3rerbMg.jsdist/remote.jsdist/inspector/vibeables.jsdist/inspector/runner.jsdist/cli/tunnel.jsdist/inspector/instances.js
Decision evidence
public snapshotAI called this Clean at 92.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- package.json defines only a prepublishOnly build command; it has no preinstall, install, or postinstall hook.
- The cited skills location discovers project and user skills by reading their directories; it does not write agent configuration there.
- The runtime hint in runs.js lists local run directories.
- The inspector assets contain bundled editor and UI code, including a React root render.
- Remote git operations are in the remote resolver, while vibeable commands and workflow runs use configured or explicitly triggered runtime paths.
- The inspector records local instance metadata; its workflow and tunnel processes are invoked through runtime features.
Evidence against
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetworkShell
HighEntropyStringsMinifiedPossibleObfuscation
Source & flagged code
8 flaggeddist/remote.jsView file
1import { spawnSync } from "node:child_process";
L2: import { mkdir, stat } from "node:fs/promises";
High
dist/inspector/vibeables.jsView file
546cwd: r.dir,
L547: shell: true,
L548: detached: process.platform !== "win32",
High
bin/kraftwerk.jsView file
15if (existsSync(src) && process.env.KRAFTWERK_DIST !== "1") {
L16: const { register } = await import("tsx/esm/api");
L17: register();
Findings
2 Critical4 High4 Medium7 Low
CriticalTrojan Source Unicodeinspector/dist/assets/editor-B3IL5KL7.js
CriticalPrevious Version Dangerous Deltadist/inspector/instances.js
HighChild Processdist/remote.js
HighShelldist/inspector/vibeables.js
HighRuntime Package Installdist/inspector/runner.js
HighSemantic Analysis Limiteddist/cli/tunnel.js#virtual:normalized:round1
MediumDynamic Requirebin/kraftwerk.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowEval
Low
Affected versions and remediation
This report applies to @netnodeag/kraftwerk@0.56.0.
See all versions of @netnodeag/kraftwerk for other recorded verdicts.
- Review the evidence and your use of @netnodeag/kraftwerk@0.56.0 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.