registry / AI-agent control-surface risk / @riddance/env
@riddance/env
Too much code slows you down, creates risks, increases maintainability burdens, confuses AI. So let's commit less of it.
This report shows the evidence for @riddance/env@0.10.2, the only known version.
AI Security Review
scanned 3mo ago · by lpm-firewall-aiThis is a concrete unconsented postinstall mutation of foreign, broad AI-agent control surfaces, matching the block boundary. The absence of network exfiltration does not mitigate the control-surface hijack.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
npm postinstall
Impact
Can steer multiple AI coding agents and grant Claude Code project-local command permissions.
Mechanism
postinstall project-wide AI-agent configuration and instruction injection
Policy narrative
On npm installation, postinstall runs bin/install.js using INIT_CWD. It calls setup(), which calls setupAgents() to write or replace AI-agent instruction files and Claude Code local permissions in the consuming project. The generated content incorporates local README text and ai.instructions from dependencies, creating an unconsented, broad agent-control surface.
AI rationale
This is a concrete unconsented postinstall mutation of foreign, broad AI-agent control surfaces, matching the block boundary. The absence of network exfiltration does not mitigate the control-surface hijack.
Evidence
package.jsonbin/install.jslib/env.jslib/agents.jsCLAUDE.mdAGENTS.mdGEMINI.md.github/copilot-instructions.md.cursor/rules/.continue/rules/.claude/settings.local.json
Decision evidence
public snapshotAI identified this as a dangerous AI-agent capability at 99.0% confidence with low false-positive risk.
Evidence for policy risk
- package.json postinstall runs bin/install.js automatically.
- bin/install.js calls setup() against INIT_CWD.
- lib/env.js invokes setupAgents() during setup.
- lib/agents.js writes CLAUDE.md, AGENTS.md, GEMINI.md, Copilot, Cursor, Continue, and Claude settings.
- Agent files incorporate README and dependency ai.instructions content.
- Claude settings grant shell commands in the target project.
Evidence against
- No network endpoint or credential-exfiltration code found in inspected paths.
- No remote payload loading, eval, or binary loading found in inspected paths.
Behavioral surface
ChildProcessDynamicRequireEnvironmentVarsFilesystemNetworkShell
UrlStrings
Source & flagged code
6 flagged · loading sourcepackage.jsonView file
•scripts.postinstall = ([ -f ./bin/install.js ] || tsc) && node ./bin/install.js
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkg•scripts.postinstall = ([ -f ./bin/install.js ] || tsc) && node ./bin/install.js
Medium
Ambiguous Install Lifecycle Script
Install-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkglib/npm.jsView file
1import { exec } from 'node:child_process';
L2: import { stat } from 'node:fs/promises';
High
Findings
1 Critical4 High5 Medium4 Low
CriticalAi Agent Control Hijacklib/agents.js
HighInstall Time Lifecycle Scriptspackage.json
HighChild Processlib/npm.js
HighShelllib/git.js
HighRuntime Package Installlib/npm.js
MediumAmbiguous Install Lifecycle Scriptpackage.json
MediumDynamic Require
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowFilesystem
LowUrl Strings
Affected versions and remediation
This report applies to @riddance/env@0.10.2.
The versions list below shows the verdict for each recorded version.
- Avoid installing @riddance/env@0.10.2. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.