registry / @koda-sl/baker-cli / 0.632.0
@koda-sl/baker-cli@0.632.0
AI-agent-first CLI for interacting with Baker, including the Baker creative canvas.
AI Security Review
scanned 4d ago · by lpm-firewall-aiThe package presents as an AI CLI, has no install lifecycle hook, and the reviewed credential use sends its configured OpenRouter key only to OpenRouter for model requests. The flagged command execution and flow code fit the package’s command and creative workflow; the routing hints did not establish an attack.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
A user invokes a CLI command that needs a model review or answer, and an OpenRouter key is configured.
Impact
The intended model service receives the content needed for the requested AI feature; no unrelated credential exfiltration or unauthorized account action was identified.
Mechanism
The command sends its model prompt and, where relevant, images to OpenRouter using the configured key. Optional debug logging writes locally when enabled.
AI rationale
The package presents as an AI CLI, has no install lifecycle hook, and the reviewed credential use sends its configured OpenRouter key only to OpenRouter for model requests. The flagged command execution and flow code fit the package’s command and creative workflow; the routing hints did not establish an attack.
Network endpoints2
openrouter.ai/api/v1/chat/completionsopenrouter.ai/api/v1/systemone
Decision evidence
public snapshotAI called this Clean at 87.0% confidence as Benign with low false-positive risk.
Evidence for AI clean decision
- The manifest identifies this as an AI CLI and declares only build, development, typecheck and test scripts; it has no install lifecycle hook.
- Model requests use OPENROUTER_API_KEY and send prompts or images to OpenRouter, matching the CLI’s AI review and rendering purpose.
- The README documents that OpenRouter requests include Baker feature, command and runtime attribution fields.
- Debug logging is disabled unless explicitly enabled through its setting.
Evidence against
- The CLI entry loads and runs a command from the supplied arguments, consistent with its stated purpose.
- The scanner’s command-output and credential-flow leads resolve to command/model functions; the reviewed requests send model inputs to OpenRouter rather than forwarding credentials to an unrelated host.
- The flagged flow file contains local flow-status and configuration helpers; the inspected excerpt shows no automatic account action or install-time write.
Behavioral surface
ChildProcessCryptoDynamicRequire
Source & flagged code
9 flaggeddist/chunk-3WEWQWRV.jsView file
59patternName = generic_password
severity = medium
line = 59
Medium
60patternName = generic_password
severity = medium
line = 60
Medium
dist/organization-VVOJFHR4.jsView file
37// src/commands/organization/pull.ts
L38: import { execFile } from "child_process";
L39: import {
High
Child Process
Package source references child process execution.
dist/organization-VVOJFHR4.jsView on unpkg · L37Findings
3 Critical4 High6 Medium6 Low
CriticalTrojan Source Unicodedist/chunk-QCY7IUWS.js
CriticalTrigger Reachable Dangerous Capabilitydist/chunk-QCY7IUWS.js
CriticalPrevious Version Dangerous Deltadist/add-DL3HVDIV.js
HighChild Processdist/organization-VVOJFHR4.js
HighCommand Output Exfiltrationdist/chunk-Z45HTBZB.js#virtual:normalized:round1
HighTrigger Reachable Command Output Exfiltrationdist/chunk-Z45HTBZB.js#virtual:normalized:round1
HighSemantic Analysis Limiteddist/add-DL3HVDIV.js#virtual:string-array:round1
MediumSecret Patterndist/chunk-3WEWQWRV.js
MediumDynamic Require
MediumNetwork
MediumEnvironment Vars
Medium
Affected versions and remediation
This report applies to @koda-sl/baker-cli@0.632.0.
See all versions of @koda-sl/baker-cli for other recorded verdicts.
- Review the evidence and your use of @koda-sl/baker-cli@0.632.0 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.