@arihantdeva/heimdall@0.2.1
Verified, self-healing knowledge layer for AI coding agents (orchestration over Graft-style semantic memory backends)
AI Security Review
scanned 11d ago · by lpm-firewall-aiAgent configuration writes are real, but the inspected caller exposes them through the explicit `heimdall init` command and the package has no install lifecycle hook. This is user-invoked setup, so no malicious behavior was established.
Decision evidence
public snapshot- The manifest defines a CLI and has no preinstall, install, or postinstall hook.
- The explicit `init` command invokes the adapter; the CLI dispatch reaches it only for `init`.
- The Codex adapter appends Heimdall search instructions to the user's `AGENTS.md` when they are absent.
- The Claude Code adapter adds a post-tool hook to the user's settings file.
- No automatic lifecycle trigger for agent configuration writes was found.
- The inspected agent configuration changes are reached through explicit CLI setup.
Source & flagged code
6 flagged · loading sourceManifest-trigger-reachable source writes behavior-bearing configuration into a user or project AI-agent control surface.
bin/lib/adapters.mjsView on unpkgThe Codex adapter appends Heimdall search instructions to the user's `AGENTS.md` when they are absent.
bin/lib/adapters.mjsView on unpkg · L69Findings
Affected versions and remediation
This report applies to @arihantdeva/heimdall@0.2.1.
See all versions of @arihantdeva/heimdall for other recorded verdicts.
- Review the evidence and your use of @arihantdeva/heimdall@0.2.1 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.