feat(cli): add manus-agent exploit-search subcommand for unified exploit database search

[manus-use]

Summary

Adds a new exploit-search CLI subcommand that queries both Exploit-DB and Packet Storm Security in a single command, presenting a unified view of public exploits for a CVE ID or keyword.

What it does

manus-agent exploit-search CVE-2024-3094
manus-agent exploit-search CVE-2024-3094 --sources exploitdb,packetstorm
manus-agent exploit-search "apache struts" --max-results 10
manus-agent exploit-search CVE-2024-3094 --output json

• Queries both exploit databases by default (configurable via --sources)
• Supports both CVE IDs and free-text keyword queries
• --max-results N controls per-source limit (default: 5, max: 20)
• Graceful partial failure: if one source errors, the other still returns results (exit 0)
• Exit 1 only when ALL sources fail
• Text and JSON output formats

Refactoring

Also refactors search_exploit_db.py and search_packetstorm.py to extract reusable library functions:

• fetch_exploitdb(query, *, max_results=5) → structured dict
• fetch_packetstorm(query, *, max_results=5) → structured dict

These functions return a clean {"source": ..., "query": ..., "exploits": [...], "error": ...} structure that both the Strands tool entry points and the CLI can consume. The Strands tool interfaces are fully backward-compatible.

Tests

28 new tests in tests/test_cli_exploit_search.py:

• TestFetchExploitdb (5 tests): unit tests for the extracted library function
• TestFetchPacketstorm (5 tests): unit tests for the extracted library function
• TestExploitSearchCLI (12 tests): CLI subcommand integration tests (text/json output, source filtering, error handling, partial failure, max-results, help flag)
• TestStrandsToolCompat (6 tests): backward compatibility verification for the refactored Strands tool entry points

All tests are 100% mocked (no real HTTP calls). Full suite: 1186 passed, 0 failures.

Files changed

• src/manus_agent/tools/search_exploit_db.py — refactored: extracted fetch_exploitdb() + kept Strands tool compat
• src/manus_agent/tools/search_packetstorm.py — refactored: extracted fetch_packetstorm() + kept Strands tool compat
• src/manus_agent/cli.py — added exploit-search subcommand (+120 lines: parser, runner, _SUBCOMMANDS, main dispatch)
• README.md — added TOC entry + CLI reference section
• tests/test_cli_exploit_search.py — new test file (28 tests)

Open PRs checked for overlap (none)

Checked all 48 open PRs: #51, #53, #54, #58, #60, #64, #65, #67, #74–90, #96, #98, #100, #103–126.

• PR test(tools): comprehensive test suite for search_exploit_db, search_packetstorm, query_threat_intelligence_feeds, obtain_cves #90 adds tool-level tests for search_exploit_db and search_packetstorm but no CLI subcommand
• PR feat(cli): add manus-agent threat-feeds subcommand for threat intelligence feed lookup #126 adds threat-feeds CLI (different tool)
• poc-search (already merged in feat: multi-source PoC aggregator (search_poc_sources) #62) is a multi-source aggregator that includes Exploit-DB among 5 sources — exploit-search is a dedicated, standalone subcommand focused exclusively on exploit database search with per-source filtering and unified output

No existing open or merged PR provides a dedicated exploit-search CLI subcommand.