Autonomous OODA-loop pentesting agent. Hermes brain, Kali + Playwright tools over MCP, ATT&CK as the Orient language.
Python ai-agentsautonomous-agentskali-linuxllm-agentsoffensive-securityooda-looppentestingplaywrightsecurity
Commits on master
docs(ctf): add HTML pentest report example with Vulnerator-AI theme
05a0dd2 irraz committed 1 month ago
feat: pentest-report skill v0.2 — two-layer report (base + enrichment)
227a120 irraz committed 1 month ago
landing: animated boot-screen landing page under docs/ for GitHub Pages
86de016 irraz committed 1 month ago
feat: SOUL.md — read_findings at every ORIENT for durable context reload
27003a0 irraz committed 1 month ago
fix: SOUL.md — enforce mandatory OODA phase label in every response
c95d5c9 irraz committed 1 month ago
fix: SOUL.md — tighten evidence discipline and progress counter rules
d4fbc0c irraz committed 1 month ago
fix(ctf): mask linuxkit kernel in uname to prevent false-positive CVEs
47ce7eb irraz committed 1 month ago
fix: SOUL.md — remove 'no gates' language that triggered model refusals
7da1d91 irraz committed 1 month ago
feat(report): emit pentest-report in English + regenerate sample
56071b3 irraz committed 1 month ago
docs: add redacted English sample pentest-report + findings store (linked from README; jsonl via explicit docs/ gitignore exception)
1bf1c11 irraz committed 1 month ago
fix: COPY scope_guard.py into the kali image (fixes clean-build crash)
32763dc irraz committed 1 month ago
fix(mcp): fail-loud oversized SSE events + stream robustness/doc nits
da82253 irraz committed 1 month ago
fix(mcp): close iteration-3 review findings (3 blocking + 4 non-blocking)
15c01e2 irraz committed 1 month ago
fix(mcp): harden stream_run_events + run_id, fix all 7 review findings
494942c irraz committed 1 month ago