registry / @sema-agent/core / 1.306.0

@sema-agent/core@1.306.0

Stateless, task-oriented AI agent core

AI Security Review

scanned 2d ago · by lpm-firewall-ai

No confirmed malicious attack surface is established. The package intentionally provides an AI-agent runtime with explicit shell and HTTP tooling, activated by consumer code or its CLI.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

Consumer invokes exported runtime tools or runs `sema-tb` with a user-supplied model/task.

Impact

Can execute configured agent actions under the host application's permissions; no unconsented execution or malicious chain was found.

Mechanism

User-configured AI-agent execution, web-fetch, and provider API calls.

Rationale

Static hits arise from an agent framework's deliberate execution and security-policy capabilities, including prose mentioning reverse shells. Source inspection found no install-time malicious behavior, stealth persistence, credential exfiltration, or foreign control-surface mutation.

Evidence

package.jsondist/index.jsdist/bin/sema-tb.jsdist/core/auto-promote.jsdist/core/auto-mode-prompt-assets.jsdist/engine/execution-env/node-execution-env.jsdist/tools/web.jsdist/core/session-store.jsdist/stores/file/session-store.jsdist/core/memory-engine/layout.jsdist/core/memory-engine/sync-client.jsdist/orchestration/workflow-script-runner.js

Network endpoints

@sema-agent/core@1.306.0

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code