sealclaw@2026.9.15
Multi-channel AI gateway with extensible messaging integrations
AI Security Review
scanned 15d ago · by lpm-firewall-aiThe source confirms an install-time mutation of installed pi-ai code, so a warning is warranted. The behavior inspected is narrowly scoped to cost handling and does not meet the blocking threshold for broad AI-agent control-surface mutation.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source
Trigger
Installing the package through npm with its postinstall hook enabled.
Impact
An installed dependency is modified during package installation, changing how reported model usage cost is handled.
Mechanism
The hook locates pi-ai files, checks for a marker and expected source anchor, then replaces the anchor and writes the patched file.
AI rationale
The source confirms an install-time mutation of installed pi-ai code, so a warning is warranted. The behavior inspected is narrowly scoped to cost handling and does not meet the blocking threshold for broad AI-agent control-surface mutation.
Evidence
package.jsonscripts/postinstall-patch-pi-ai.mjsinstalled node_modules/@mariozechner/pi-ai/dist/providers/openai-completions.jsinstalled node_modules/@mariozechner/pi-ai/dist/providers/openai-responses-shared.js
Decision evidence
public snapshotAI called this Suspicious at 88.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The package postinstall runs a script that searches for an installed pi-ai dependency and writes a patch into that dependency’s source file.
- The patch is narrowly targeted at pi-ai cost handling; the inspected script does not show a broad agent configuration change.
- The vault initializer writes an AGENTS.md file when the wiki-init command is invoked.
- The inspected audit helper only enumerates installed plugin directories.
Evidence against
- The package-manager preinstall hook only emits a warning based on detected package manager.
- The inspected postinstall patch proceeds only when its target and expected source anchor are found.
Behavioral surface
ChildProcessCryptoDynamicRequireEnvironmentVarsEvalFilesystemNetworkShellWebSocket
HighEntropyStringsMinifiedPossibleObfuscation
Source & flagged code
12 flagged · loading sourcepackage.jsonView file
•scripts.preinstall = node scripts/preinstall-package-manager-warning.mjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg•scripts.postinstall = node scripts/postinstall-bundled-plugins.mjs && node scripts/postinstall-patch-pi-ai.mjs
Medium
Ambiguous Install Lifecycle Script
Install-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgdist/proxy-cli.runtime-C5A6MhNT.jsView file
764patternName = generic_password
severity = medium
line = 764
Medium
Secret Pattern
Package contains a possible secret pattern.
dist/proxy-cli.runtime-C5A6MhNT.jsView on unpkg · L764Findings
1 Critical7 High7 Medium6 Low
CriticalTrojan Source Unicodedist/control-ui/assets/index-C4J4V6vf.js
HighInstall Time Lifecycle Scriptspackage.json
HighCloud Metadata Accessdist/control-ui/assets/index-C4J4V6vf.js
HighPayload In Excluded Dirskills/llm-wiki-ingest/tests/test_ingest_scripts.py
HighSemantic Analysis Limiteddist/control-ui/assets/index-C4J4V6vf.js
HighKnown Malware Source Similaritydist/plugin-sdk/plugin-test-contracts.js
HighKnown Malware Source Similarityscripts/preinstall-package-manager-warning.mjs
HighPrevious Version Dangerous Deltascripts/preinstall-package-manager-warning.mjs
MediumAmbiguous Install Lifecycle Scriptpackage.json
MediumSecret Patterndist/proxy-cli.runtime-C5A6MhNT.js
Affected versions and remediation
This report applies to sealclaw@2026.9.15.
See all versions of sealclaw for other recorded verdicts.
- Review the evidence and your use of sealclaw@2026.9.15 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.