refactor: rename to firewall-escape.md, add 80% novelty requirement, and increase timeout (#4789)

* Initial plan

* feat: enhance test-firewall-escape.md with deep AWF implementation knowledge

Apply advanced prompt engineering techniques to guide the security testing agent:
- Provide detailed AWF architecture overview (Squid, iptables, Docker wrapper)
- Document network topology with ASCII diagram
- Explain key implementation details and their attack surfaces
- Include 13+ specific attack vectors with bash commands
- Add Phase 1 reconnaissance, Phase 2 systematic exploitation
- Reference actual code paths and configuration files
- Include creative thinking section for out-of-box approaches
- Extended timeout from 15 to 30 minutes for thorough testing

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

* fix: address code review feedback for test-firewall-escape.md

- Replace hardcoded IP (93.184.216.34) with dynamic resolution using dig
- Use actual allowed domain 'github.com' instead of 'allowed-domain.com'
- Add clarifying comments for URL encoding tests (%2E is URL-encoded ".")

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

* refactor: simplify prompt to direct agent to explore AWF repo

Reverted the detailed architecture documentation and instead instructed the agent to:
1. Explore https://github.com/githubnext/gh-aw-firewall
2. Read key source files to understand the firewall architecture
3. Devise escape strategies based on their understanding
4. Execute attacks methodically and document results

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

* refactor: rename test-firewall-escape to firewall-escape and add 80% novelty requirement

- Renamed workflow from test-firewall-escape.md to firewall-escape.md
- Updated workflow name and description
- Added instruction for agent to prioritize new strategies (80% novel attempts)
- Agent now uses cache-memory to track and avoid repeating previous attempts

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

* chore: increase workflow timeout to 30 minutes

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

* chore: increase workflow timeout to 60 minutes

Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: Mossaka <5447827+Mossaka@users.noreply.github.com>

Author: Copilot