registry / @sema-agent/core / 1.308.0

@sema-agent/core@1.308.0

Stateless, task-oriented AI agent core

AI Security Review

scanned 1d ago · by lpm-firewall-ai

No confirmed malicious attack surface. The package provides user-invoked AI-agent, shell, workflow, LSP, and web-fetch capabilities rather than install-time execution or covert payload delivery.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

Consumer imports APIs or explicitly runs the `sema-tb` CLI/agent tools.

Impact

Capabilities can run commands or access caller-selected web resources only when a consuming application or user enables them; no unconsented persistence, credential exfiltration, or foreign agent-control mutation was found.

Mechanism

Explicitly configured AI-agent execution and guarded web fetching.

Rationale

Static hints misclassified security-policy prompt text and package-aligned agent primitives as malicious behavior. Direct inspection found no install-time payload, covert exfiltration, remote payload loading, or unconsented AI-agent control-surface mutation.

Evidence

package.jsondist/index.jsdist/core/auto-mode-prompt-assets.jsdist/tools/web.jsdist/engine/execution-env/node-execution-env.jsdist/orchestration/run-workflow-tool.jsdist/bin/sema-tb.jsdist/core/auto-promote.jsdist/core/session-store.jsdist/engine/execution-env/kill-tree.jsdist/engine/lsp/node-lsp-manager.jsdist/core/spec-contract.jsdist/core/memory-engine/file-backend.js

@sema-agent/core@1.308.0

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code