Coding & Development
Agentinel
1h agoAgentinel is a free, local security layer for autonomous coding agents. It registers a PreToolUse hook that intercepts every package install command, resolves the full dependency tree, and scans it against an extensive local OSV threat database plus heuristic checks for slopsquatting and zero‑day threats. Malicious or hallucinated packages are blocked instantly with ultra‑low latency, and a structured JSON feedback block is injected back into the AI context so the agent can self‑correct without human intervention. It also works in CI/CD pipelines (GitHub Actions, GitLab CI) to guard deployments, all without any cloud calls.
Local Zero‑Cloud ScanningExtensive OSV Threat DatabaseHeuristic Slopsquatting DetectionFull Dependency‑Tree ScanningPreToolUse Hook InterceptionUltra‑Low Latency ScansAI Self‑Correction Feedback LoopCI/CD Pipeline IntegrationZero‑Cost Free Model
Key Capabilities
01
Local Zero‑Cloud Scanning
All scans are performed locally with no telemetry sent to the cloud.
02
Extensive OSV Threat Database
Uses a local copy of the OSV database (~216k packages) to match known malicious packages.
03
Heuristic Slopsquatting Detection
Heuristics such as package age, download count, and publisher drift catch newly created or fake packages.
04
Full Dependency‑Tree Scanning
Resolves and scans the entire npm, pip, or cargo dependency tree, not just top‑level packages.
05
PreToolUse Hook Interception
Registers as a PreToolUse hook with the AI agent runtime to intercept commands before execution.
06
Ultra‑Low Latency Scans
Provides near‑instant feedback so development flow is not disrupted.
07
AI Self‑Correction Feedback Loop
Feeds a JSON block back into the AI context describing the block reason, enabling autonomous correction.
08
CI/CD Pipeline Integration
Can be used in GitHub Actions, GitLab CI, or any pipeline to scan manifests before deployment.
09
Zero‑Cost Free Model
Offers all features without subscription or licensing fees.
Use Cases
Prevent Malicious Packages in AI‑Generated CodeBlocks hallucinated or malicious npm packages before they are installed, protecting local development environments.Software developers using AI coding assistants
Secure CI/CD DeploymentsScans dependency manifests in CI pipelines to stop supply‑chain attacks before code reaches production.DevOps and release engineering teams
Enable Autonomous AI Self‑CorrectionProvides structured feedback so AI agents can automatically adjust their package selections without manual intervention.AI tool builders and researchers
Maintain Privacy‑Focused DevelopmentOperates entirely locally with no cloud calls, ensuring code and dependency data never leave the developer’s machine.Privacy‑concerned engineers and organizations
Update History
Updated 1h ago▲ 0% mentionsProduct HuntWeb