feat: add cve-enrich tool and CLI subcommand for multi-source CVE enrichment

[manus-use]

Summary

Add a lightweight, non-agent CVE enrichment tool that fetches data from multiple public sources in parallel and returns a unified risk snapshot.

Motivation

Currently, getting a complete picture of a CVE requires manually invoking multiple tools (get_nvd_data, EPSS APIs, check_cisa_kev, get_osv_data, etc.) and mentally combining their outputs. This tool provides a single-call enrichment pipeline that:

1. Removes the need for an LLM agent for basic CVE triage
2. Provides a programmatic building block for other tools and workflows
3. Adds a quick CLI subcommand for analysts who just want a snapshot

What's included

New tool: cve_enrich (src/manus_agent/tools/cve_enrich.py)

• Parallel fetching via ThreadPoolExecutor (5 sources, ~15s worst case)
• Sources: NVD (CVSS, CWE, references), EPSS (exploitation probability), CISA KEV (active exploitation), OSV.dev (affected packages + fixed versions), VulnCheck KEV (multi-source exploitation signal)
• Composite risk scoring: Weighted combination of CVSS severity, EPSS probability, KEV presence → produces critical/high/medium/low/unknown level
• Graceful partial failure: Individual source errors don't break the pipeline
• Strands @tool interface for direct agent use
• Public enrich_cve() function for programmatic consumption

New CLI subcommand: manus-agent enrich

# Quick triage with colored text output
manus-agent enrich CVE-2024-3094

# Machine-readable JSON
manus-agent enrich CVE-2024-3094 --output json

# Skip VulnCheck if no API key
manus-agent enrich CVE-2024-3094 --no-vulncheck

Test suite: 43 new tests (tests/test_cve_enrich.py)

All HTTP calls fully mocked. Covers:

• Each fetcher function (success, error, edge cases)
• Risk computation logic (all severity levels)
• Integration scenarios (partial failure, exceptions, case normalization)
• CLI text/JSON output modes
• Strands tool interface

Test results

43 passed in 1.35s (new tests)
1201 passed, 3 deselected, 3 warnings in 25.55s (full suite)

Usage as a building block

from manus_agent.tools.cve_enrich import enrich_cve

result = enrich_cve("CVE-2024-3094")
print(result["risk_assessment"]["level"])  # 'critical'
print(result["epss"]["score"])              # 0.975
print(result["cisa_kev"]["in_kev"])         # True