Warn by default; block when configured. The package changes AI-agent instructions for the user without an explicit install command, creating an unconsented agent control surface.
AI security review
scanned · by lpm-firewall-ai
The npm postinstall hook automatically writes this package’s skill into the user-wide AI-agent skills directory and forces replacement of differing content. This is an unconsented install-time mutation of an AI-agent control surface.
npm install or another package-manager action that runs the postinstall hook, unless DCR_SKIP_INSTALL=1 is set.
Impact
The package changes AI-agent instructions for the user without an explicit install command, creating an unconsented agent control surface.
Mechanism
The hook calls installSkill with force enabled. By default, it writes the bundled SKILL.md to the user-wide ~/.agents/skills/dev-change-report/SKILL.md path, replacing differing existing content.
Policy narrative
During package installation, the postinstall hook writes the bundled skill into the user-wide AI-agent skills directory. It forces replacement of differing existing content. This modifies an AI-agent control surface outside the package without an explicit user command; the environment-variable opt-out does not make the default action requested.
AI identified this as a dangerous AI-agent capability at 99.0% confidence with low false-positive risk.
Evidence for policy risk
The package runs bin/postinstall.js automatically through its npm postinstall lifecycle script.
The postinstall hook calls installSkill by default; skipping it requires setting DCR_SKIP_INSTALL=1.
installSkill targets ~/.agents/skills/dev-change-report/SKILL.md by default and writes the package’s SKILL.md there, overwriting changed content because the hook passes force: true.
Evidence against
The installed skill describes features aligned with the package’s stated purpose: generating development reports and technical articles.
The postinstall hook provides an opt-out through DCR_SKIP_INSTALL=1.
Behavioral surface
SourceChild process · Environment variables · Filesystem
Supply chainHigh entropy strings · URL strings
ManifestNo manifest risk signals triggered.
Scanned 12 files · 61.7 KB of source
3 external domains referenced
purl.org
schemas.openxmlformats.org
www.w3.org
Findings
1 critical · 3 high · 4 medium · 4 low
Where they are, most severe file first. Code opens a file's flagged lines, with their evidence, below.
Avoid installing dev-change-report@1.3.0. 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, check your AI agents' configuration and instruction files (the files this report lists) for changes you did not make, and remove them. This finding does not by itself establish malware intent.