bug-hunter

Comprehensive multi-agent bug hunting with parallel specialist agents, deduplication, regression guard, and prioritized reporting

Use this agent to hunt for API contract violations: type mismatches between caller and callee, wrong argument order, schema drift between client and server, incorrect return value handling, and interface/protocol violations. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the api-contract-checker to verify function signatures and API schemas match their usage." <Task tool invocation to launch api-contract-checker agent> </example>

Use this agent to hunt for edge case bugs: null/undefined paths, empty collections, boundary values, integer overflow, Unicode handling issues, and unexpected input types. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the edge-case-finder agent to check for unhandled edge cases." <Task tool invocation to launch edge-case-finder agent> </example>

Use this agent to hunt for error handling bugs: silent failures, swallowed exceptions, empty catch blocks, overly broad catches, missing error propagation, and fallback logic that masks real problems. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the error-handler agent to check for silent failures and inadequate error handling." <Task tool invocation to launch error-handler agent> </example>

Use this agent to hunt for logic bugs: off-by-one errors, wrong comparisons, inverted conditions, unreachable code, incorrect operator precedence, and flawed control flow. Launch this agent via the Task tool with a prompt specifying the files to analyze. The agent returns findings in a standardized format with severity and confidence scores. Examples: <example> Context: Reviewing recently changed files for logic errors. assistant: "I'll launch the logic-hunter agent to check for logic bugs in the changed files." <Task tool invocation to launch logic-hunter agent> </example>

Use this agent to hunt for concurrency bugs: race conditions on shared mutable state, TOCTOU vulnerabilities, deadlocks, async/await pitfalls, missing synchronization, and thread-safety violations. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the race-condition-detector to check for concurrency issues in the async code." <Task tool invocation to launch race-condition-detector agent> </example>

Use this agent after collecting bug findings to assess regression risk of fixing each bug. It checks test coverage, downstream consumers, public API impact, and classifies each fix as SAFE/CAUTION/BREAKING. It also suggests test cases to write before and after applying fixes. Launch via Task tool with the consolidated findings as context. Examples: <example> assistant: "I'll launch the regression-guard agent to analyze the regression risk of the proposed fixes." <Task tool invocation to launch regression-guard agent with findings> </example>

Use this agent to hunt for resource leaks: unclosed file handles, database connections not returned to pool, HTTP clients not closed, missing cleanup in error paths, goroutine/thread leaks, and memory leaks from retained references. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the resource-leak-hunter to check for unclosed connections and file handles." <Task tool invocation to launch resource-leak-hunter agent> </example>

Use this agent for deep security analysis: injection vulnerabilities (SQL, command, XSS, template), authentication/authorization bypass, path traversal, SSRF, hardcoded secrets, insecure deserialization, cryptographic misuse, and OWASP Top 10 issues. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the security-scanner agent for a deep security review of the auth module." <Task tool invocation to launch security-scanner agent> </example>

Use this agent to hunt for state management bugs: stale state, missing UI updates, cache invalidation errors, state machine violations, inconsistent state across components, and state synchronization failures. Launch via Task tool with files to analyze. Examples: <example> assistant: "I'll launch the state-bug-finder to check for stale state and missing UI updates." <Task tool invocation to launch state-bug-finder agent> </example>