agency-lang@0.18.2
The Agency language
AI Security Review
scanned 2mo ago · by lpm-firewall-aiSource inspection found no lifecycle script, self-dependency chain, or automatic exfiltration or persistence. Flagged capabilities are tied to the package's language runtime and explicit CLI commands.
Decision evidence
public snapshot- The manifest exposes library and CLI entrypoints, with no install lifecycle hook or runtime dependency on itself.
- The CLI's dynamic execution path adds a V8 restriction when the untrusted-output mode is selected.
- Subprocess code validates generated paths and confines temporary writes to its own package directory.
- Optimizer writeback checks source hashes before writing changed candidate files.
- OAuth behavior is user-initiated, uses a loopback callback, and stores tokens with restrictive file permissions.
- launchd persistence is reachable only through the explicit schedule command.
- No credential harvesting, covert exfiltration, install-time execution, or foreign agent-control-surface mutation was found.
- Network and process APIs implement explicit CLI/runtime features rather than an automatic package-install payload.
Source & flagged code
7 flaggedPackage source references child process execution.
dist/lib/optimize/workspace.jsView on unpkg · L6Findings
Affected versions and remediation
This report applies to agency-lang@0.18.2.
See all versions of agency-lang for other recorded verdicts.
- Review the evidence and your use of agency-lang@0.18.2 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.