test(tools): comprehensive test suite for search_for_exploits module (+53 tests)

[manus-use]

Summary

Adds a comprehensive test suite for the search_for_exploits module (src/manus_agent/tools/search_for_exploits.py) — 53 fully-mocked tests covering all code paths in this security-critical GitHub PoC search tool.

What's tested

Category	Tests	Coverage
TOOL_SPEC contract	5	Name, description, inputSchema structure, required fields, types
Input validation	7	None, empty string, numeric, missing CVE- prefix, random string, missing key, toolUseId preservation
Successful search	7	Results returned, summary includes count, top-5 limiting, result field structure, case variants, single result
Zero results	3	Success status, CVE mentioned in summary, empty links list
Token handling	5	Env var injection, config fallback, no-token path, env priority over config, config exception fallback
HTTP errors	5	Timeout, connection error, 403 rate limit, 500 server error, 422 unprocessable
JSON errors	2	JSONDecodeError, unexpected response structure (KeyError)
Unexpected exceptions	2	RuntimeError, OSError catch-all
Output logging	4	Logging integration doesn't crash on success, error, invalid input, zero results
Request construction	5	URL includes CVE ID, targets search API, Accept header, timeout param, sort=updated
Result parsing edge cases	5	Null description, zero stars, exactly 5 results, fewer than 5, summary mentions "top 5"
ToolUse ID propagation	3	ID preserved on success, network error, validation error

Why this matters

search_for_exploits is a security-critical tool that searches GitHub for public Proof-of-Concept exploits. It handles authentication tokens, makes network requests, and parses untrusted JSON responses. Despite this, it had zero dedicated test coverage (only 2 importability checks in unrelated files). This suite covers:

• All error paths (network failures, malformed responses, auth failures)
• Token injection security (env var priority, config fallback, exception safety)\n- Response parsing robustness (null fields, missing keys, count boundaries)
• Request construction correctness (URL format, headers, timeout)

Test design notes

• No real HTTP calls — all requests.get calls are mocked
• Resilient to test ordering — the TestOutputLogging class avoids assert_called_once patterns that break when test_packageability.py (which uses importlib._fresh_import) runs earlier in the suite and invalidates module-level mock patches
• Config isolation — Config.from_file() is mocked to prevent filesystem dependencies

Existing PRs checked (no overlap)

• PR test(tools): comprehensive test suite for get_github_advisory and search_for_exploits #92: tests get_github_advisory and search_for_exploits helpers — but only 2 importability/wiring checks, not functional tests
• PR test(tools): comprehensive test suite for http_request truncation wrapper (+36 tests) #140: tests http_request truncation wrapper (different module)
• PR test(tools): comprehensive test suites for search_exploit_db, search_packetstorm, and get_cwe_details (+125 tests) #173: tests search_exploit_db, search_packetstorm, get_cwe_details (different modules)
• PR test(tools): comprehensive test suite for query_threat_intelligence_feeds module (+49 tests) #174: tests query_threat_intelligence_feeds (different module)
• PRs feat(cli): add manus-agent poc-week subcommand for trending CVE digest #131–feat: add cve-neighbors tool and CLI subcommand (+51 tests) #180: none target search_for_exploits functional test coverage

This is the first dedicated functional test suite for search_for_exploits.

Test results

1211 passed, 3 deselected, 3 warnings in 24.56s

Baseline: 1158 tests. New: 53 tests. All pass, ruff check/format clean.