feat: add autonomous bounty-hunting agent (Bounty #861)

[liufang88789-ui]

Overview

Implements a full Autonomous Bounty-Hunting Agent for SolFoundry — a state-machine system that autonomously finds, analyzes, plans, implements, tests, and submits solutions for GitHub bounty issues.

This PR is a clean replacement for earlier draft work and contains only the files relevant to Bounty #861.

---

What’s Included

automaton/bounty_hunter/agent.py

• Main BountyHunterAgent state machine
• Flow: IDLE → SCANNING → ANALYZING → PLANNING → IMPLEMENTING → TESTING → SUBMITTING → DONE
• Supports scanning for bounties, selecting a target, generating code, testing, and creating PRs

automaton/bounty_hunter/github_client.py

• GitHub API wrapper for:
  • listing open bounty issues
  • reading issue details
  • creating branches
  • updating files
  • opening pull requests
• Includes bounty metadata parsing (tier, reward, domain)

automaton/bounty_hunter/planner.py

• LLM-backed planner that:
  • analyzes bounty requirements
  • produces a structured implementation plan
  • can generate code per implementation step
• Falls back to heuristic planning if no API key is configured

automaton/bounty_hunter/coder.py

• Applies generated file changes locally
• Handles git staging, commit, and push workflows
• Provides basic codebase exploration utilities

automaton/bounty_hunter/tester.py

• Detects test frameworks (pytest, jest, vitest, mocha, unittest)
• Runs tests and parses results
• Includes lightweight lint integration hook

automaton/bounty_hunter/test_bounty_hunter.py

• Minimal validation tests covering:
  • bounty metadata parsing
  • planner fallback behavior
  • test framework detection

automaton/bounty_hunter/README.md

• Usage, architecture, CLI examples, and integration notes

---

Acceptance Criteria Mapping

• ✅ Multi-LLM agent orchestration with planning
  Implemented through the planner-driven workflow and pluggable LLM integration.
• ✅ Automated solution implementation and testing
  Implemented via the coder + tester modules.
• ✅ Autonomous PR submission with proper formatting
  Implemented via GitHub client + branch/PR creation flow.

---

Validation

Ran locally:

python3 -m py_compile automaton/bounty_hunter/*.py
python3 -m pytest automaton/bounty_hunter/test_bounty_hunter.py -q

Result:

• py_compile ✅
• pytest ✅ 3 passed

---

Wallet

Solana: 7UqBdYyy9LG59Un6yzjAW8HPcTC4J63B9cZxBHWhhHsg

EVM: 0x7F3a01563C504bD57aa465dd6273Ef21AF8F7784

---

Closes #861