greydgl--pentestgpt
ab5fbb4d90
* first refactor
* feat: dockerized tool with persistent Claude+Codex login + multi-model benchmark
Run the autonomous CTF/pentest tool in Docker with a one-time, persistent login for
BOTH Claude Code and Codex, and add a multi-model benchmark harness.
Backend (multi-model):
- Add `--backend {claude,codex}` to the CTF pipeline. CodexBackend (pentestgpt/core/
backend.py) wraps unified_agent's Codex backend and translates its events into
AgentMessages, so the same pipeline runs on Claude (opus/sonnet) or Codex
(gpt-5.5/gpt-5.4-mini). Wired through config.backend, pipeline stage construction,
and the CLI (+ PENTESTGPT_CODEX_EFFORT; greppable [CODEX_USAGE] under PENTESTGPT_BENCH=1).
Docker tool (tool-only image; the benchmark stays OUTSIDE the image):
- Extend Dockerfile: Codex CLI (@openai/codex) + openai_codex SDK + unified_agent/
pentestgpt_agent/pentestgpt_legacy packages + gobuster/dirb + socat. Add .dockerignore
(keeps creds/benchmark/workspace out of the build context).
- Persistent dual login (the hard part) — asymmetric by token model:
* Claude: `setup-token` -> token stored in the pentestgpt-claude volume; entrypoint
exports CLAUDE_CODE_OAUTH_TOKEN (setup-token does not write .credentials.json; macOS
host creds live in the Keychain and can't be copied).
* Codex: the container does its OWN `codex login` (NOT seeding -- ChatGPT refresh tokens
are single-use, so a shared/copied login 401s on first refresh). The 127.0.0.1:1455
OAuth callback is forwarded into the container via a socat hop (-p 1455:8455).
* scripts/docker-login.sh is idempotent: checks logins live, logs in only the missing one(s).
- docker-compose codex-config volume (+ pinned names); entrypoint token-export + non-blocking
preflight; scripts/docker-auth-status.sh; Make targets (docker-build/login/auth-status/
run/shell/down/nuke).
- Verified end-to-end: one `make docker-login` -> a fresh container reports claude+codex
logged in with live round-trips; the CTF pipeline (Codex) captured a flag against an
isolated fixture and the pentest pipeline ran cleanly; persists across recreation, no re-login.
Benchmark (multi-model, host-side):
- benchmark/pilot/ harness (run_pilot.py + report.py): builds each xbow challenge, discovers
the loopback port, runs the pipeline across the 4 model combos, judges by the baked
FLAG{sha256(UPPER-dir)}, and renders REPORT.md (infra failures excluded from solve rates).
Includes the partial pilot's results (results.jsonl + REPORT.md).
Docs: docs/docker-dev-plan.md (full plan + implementation status); CLAUDE.md and README
docker quickstart; benchmark/pilot/README.md; design-doc roadmap (docs/redesign).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* fix: fail controller on backend error messages
* fix: allow listing sessions without target
* docs: add docker xbow benchmark report
* fix: infer concrete backend constructor type
* docs: refresh docker benchmark documentation
* feat(benchmark): add pure single-agent baseline + pipeline comparison
Add a "pure single agent" benchmark variant -- one bare `claude -p` /
`codex exec` call per target (no pipeline) -- to quantify what the 3-stage
PentestGPT pipeline buys over an un-orchestrated agent on the xbow targets.
- pentestgpt/prompts/stages.py: ctf_single_agent_{system,task}_prompt -- the
pipeline's shared fragments collapsed into ONE turn, so prompt content is
held constant and the only variable is the multi-stage decomposition.
- benchmark/pilot/run_docker_bench.py: docker-network runner
(--variant single|pipeline). Brings the target up, discovers the container's
internal IP+network (skips DB side-cars/ports), docker-runs the tool image on
that network, and scores the ground-truth flag against the agent's *assistant
text* only (parity with the pipeline's raw streaming). Reads stdout in chunks
to handle >64KB JSON lines. Resumable; --dry-run supported.
- benchmark/pilot/report_comparison.py -> DOCKER_COMPARISON.md: head-to-head
pipeline-vs-single per model on the common non-infra set.
- tests/unit/test_single_agent_prompt.py: prompt-builder coverage.
- docs: README, CLAUDE.md, benchmark README, DOCKER_REPORT updated.
Recorded result (10 medium/hard targets x 4 models, container-to-container,
same baseline image digest 0c4c0f3e..., commit dca0019 image):
Model Pipeline Single
Claude Opus 5/10 7/10 (single +2)
Claude Sonnet 6/10 4/10 (pipeline +2)
Codex gpt-5.5 7/10 7/10 (tie)
Codex gpt-5.4-mini 3/10 4/10 (single +1)
TOTAL 21/40 22/40
Single agent matches the pipeline on solve rate (55% vs 52%) while using
~40% fewer Codex tokens (13.0M vs 21.8M) and solving faster. The pipeline
only clearly helps Claude Sonnet (which times out solo); Opus is better solo.
Full per-challenge grid in DOCKER_COMPARISON.md; raw records in
docker_single_results.jsonl.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
* feat(benchmark): add pentestgpt_agent docker harness
* bench: refresh pentestgpt_agent smoke result
* fix(benchmark): make repeat rows variant-aware
* fix(agent): fall back for semantic executor labels
* fix(agent): tolerate executor prose evidence
* fix(benchmark): score accepted framework findings
* bench: append partial framework repeat results
* bench: complete framework repeat sweep
* bench: expose framework executor concurrency
* bench: add extended parallel framework sweep
* checkpoint: preserve working agent and benchmark state
* feat: harden durable agent loop and xbow qualification
* fix: reserve an exploit result turn
* docs: record clean xbow qualification
* build: consume unified-agent from the git wrapper repo
Repoint pentestgpt_agent_new's unified-agent dependency from the local
editable path (../../UnifiedAgentPoC, now renamed and gone) to the pinned
git source PentestGPT-Project/UnifedAgentWrapper@d05d21f. Regenerate uv.lock
and update test_dependency.py to assert the external package is installed
from that VCS URL (not the repo-root vendored copy) at version 0.2.0.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* refactor: make pentestgpt_agent_new the sole framework
Remove the retired ledger-based pentestgpt_agent package (instructor/executor/
judge) and its orphaned unit + smoke tests. The nested pentestgpt_agent_new
project (Supervisor/Executor over a durable SQLite loop, consuming unified-agent
from the git wrapper) is now the single maintained framework.
Repoint the top-level tooling to it:
- pyproject: drop the pentestgpt-agent console script and pentestgpt_agent from
the wheel packages.
- Makefile: lint/format target parent code only; typecheck/check/ci now run the
nested framework's own gate (ruff, format, mypy, pytest) via test-agent-new /
check-agent-new, so `make check` finally covers it; `make run` delegates to the
pentestgpt-agent-new CLI.
- Dockerfile: stop copying the removed package (kept the build working); note the
framework is not baked into the image yet.
- docker container-health test: import the substrate packages that actually ship.
- CLAUDE.md / AGENT.md: describe the new framework, the git-sourced wrapper, and
the deprioritized benchmark/Docker rewire.
The XBOW `--variant framework` path and docker-bench Makefile targets still point
at the old in-image framework and are left as a pending rewire (benchmarks
deprioritized); the naive `--variant single` path is unaffected.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* refactor: rename pentestgpt_agent_new -> pentestgpt_agent
The framework reclaims the clean name now that the old ledger-based package is
gone. Rename the nested project folder, its src package, the distribution
(pentestgpt-agent-new -> pentestgpt-agent) and CLI, and every import/reference in
the package, the umbrella Makefile, the Dockerfile, the docker health test, and
CLAUDE.md / AGENT.md. Regenerate uv.lock. The audit CLI stays pentestgpt-agent-audit;
the git-sourced unified-agent dependency is unchanged. `make check` is green
(108 nested tests). The two historical *_REPORT.md files keep the old name as
dated records.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: extract benchmark harness to sibling xbow-benchmark repo
Move PentestGPT/benchmark/ out to ../xbow-benchmark (its own repo) to keep this
project clean. The harness was decoupled from the framework code (it scores
container output, never imports pentestgpt_agent/unified_agent), so only
operational ties remain and they now live in the sibling repo.
- Remove benchmark/ and the 4 harness unit tests (relocated + repointed there).
- Strip the docker-bench-*/bench-* targets and their config vars from the
Makefile; keep the tool-image lifecycle (docker-build/login/run/...) and add a
help pointer to `make -C ../xbow-benchmark help`.
The sibling repo mounts this checkout read-only (--source-root ../PentestGPT) and
runs the pentestgpt:latest image built here.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* feat: harden autonomous framework and runtime integration
---------
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
364 行
11 KiB
Python
364 行
11 KiB
Python
from pathlib import Path
|
|
|
|
import pytest
|
|
|
|
from pentestgpt_agent.execution import ExecutionOutcome, ValidExecution
|
|
from pentestgpt_agent.memory import MemoryKernel, RunSpec, RunSpecMismatchError, RunStatus
|
|
from pentestgpt_agent.plan import (
|
|
SupervisorDecision,
|
|
TaskKind,
|
|
TaskProposal,
|
|
TaskStatus,
|
|
compile_plan,
|
|
)
|
|
|
|
|
|
def test_run_is_recoverable_from_a_new_memory_kernel_instance(tmp_path: Path) -> None:
|
|
database = tmp_path / "run-state.sqlite3"
|
|
created = MemoryKernel(database).create_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
)
|
|
)
|
|
|
|
recovered = MemoryKernel(database).snapshot("run-1")
|
|
|
|
assert recovered == created
|
|
assert recovered.revision == 0
|
|
assert recovered.status is RunStatus.RUNNING
|
|
assert recovered.tasks == ()
|
|
|
|
|
|
def test_open_run_reuses_only_an_identical_persisted_spec(tmp_path: Path) -> None:
|
|
database = tmp_path / "run-state.sqlite3"
|
|
spec = RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
max_attempts_per_task=2,
|
|
)
|
|
created = MemoryKernel(database).open_run(spec)
|
|
|
|
reopened = MemoryKernel(database).open_run(spec)
|
|
|
|
assert reopened == created
|
|
with pytest.raises(RunSpecMismatchError, match="does not match"):
|
|
MemoryKernel(database).open_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess a different target.",
|
|
allowed_targets=spec.allowed_targets,
|
|
max_attempts_per_task=spec.max_attempts_per_task,
|
|
)
|
|
)
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"spec, message",
|
|
[
|
|
(
|
|
RunSpec("run-1", "x" * 4_001, ("http://127.0.0.1:8080",)),
|
|
"goal must be",
|
|
),
|
|
(
|
|
RunSpec(
|
|
"run-1",
|
|
"Assess the target.",
|
|
tuple(f"http://127.0.0.1:{8000 + index}" for index in range(17)),
|
|
),
|
|
"at most 16",
|
|
),
|
|
],
|
|
)
|
|
def test_run_spec_memory_inputs_are_bounded(
|
|
tmp_path: Path,
|
|
spec: RunSpec,
|
|
message: str,
|
|
) -> None:
|
|
with pytest.raises(ValueError, match=message):
|
|
MemoryKernel(tmp_path / "state.sqlite3").open_run(spec)
|
|
|
|
|
|
def test_progress_task_can_be_selected_again_and_completed(tmp_path: Path) -> None:
|
|
memory = MemoryKernel(tmp_path / "state.sqlite3")
|
|
initial = memory.create_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
)
|
|
)
|
|
target = "http://127.0.0.1:8080"
|
|
first_commit = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=0,
|
|
new_tasks=(
|
|
TaskProposal(
|
|
"discover",
|
|
TaskKind.DISCOVER,
|
|
target,
|
|
"Inspect HTTP.",
|
|
"HTTP is recorded.",
|
|
),
|
|
),
|
|
next_task_id="discover",
|
|
finish=False,
|
|
summary="Start discovery.",
|
|
),
|
|
initial,
|
|
)
|
|
)
|
|
assert first_commit.lease is not None
|
|
partial = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="discover",
|
|
attempt_id=first_commit.lease.attempt_id,
|
|
lease_revision=first_commit.lease.revision,
|
|
trace_episode_id="executor-1",
|
|
outcome=ExecutionOutcome.PROGRESS,
|
|
summary="Discovery needs one more request.",
|
|
observation="The first response linked to a second path.",
|
|
evidence_sequences=(1,),
|
|
)
|
|
)
|
|
second_commit = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=partial.revision,
|
|
new_tasks=(),
|
|
next_task_id="discover",
|
|
finish=False,
|
|
summary="Use the second bounded discovery attempt.",
|
|
),
|
|
partial,
|
|
)
|
|
)
|
|
assert second_commit.lease is not None
|
|
|
|
completed = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="discover",
|
|
attempt_id=second_commit.lease.attempt_id,
|
|
lease_revision=second_commit.lease.revision,
|
|
trace_episode_id="executor-2",
|
|
outcome=ExecutionOutcome.DONE,
|
|
summary="HTTP inspected.",
|
|
observation="HTTP returned 200 OK.",
|
|
evidence_sequences=(1,),
|
|
)
|
|
)
|
|
|
|
assert {task.id: task.status for task in completed.tasks} == {
|
|
"discover": TaskStatus.DONE,
|
|
}
|
|
|
|
|
|
def test_new_task_with_completed_dependencies_can_be_leased_immediately(
|
|
tmp_path: Path,
|
|
) -> None:
|
|
memory = MemoryKernel(tmp_path / "state.sqlite3")
|
|
initial = memory.create_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
)
|
|
)
|
|
target = "http://127.0.0.1:8080"
|
|
first_commit = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=0,
|
|
new_tasks=(
|
|
TaskProposal(
|
|
"discover",
|
|
TaskKind.DISCOVER,
|
|
target,
|
|
"Inspect HTTP.",
|
|
"HTTP is recorded.",
|
|
),
|
|
),
|
|
next_task_id="discover",
|
|
finish=False,
|
|
summary="Discover the target.",
|
|
),
|
|
initial,
|
|
)
|
|
)
|
|
assert first_commit.lease is not None
|
|
after_discovery = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="discover",
|
|
attempt_id=first_commit.lease.attempt_id,
|
|
lease_revision=first_commit.lease.revision,
|
|
trace_episode_id="executor-1",
|
|
outcome=ExecutionOutcome.DONE,
|
|
summary="HTTP inspected.",
|
|
observation="HTTP returned 200 OK.",
|
|
evidence_sequences=(1,),
|
|
)
|
|
)
|
|
next_plan = compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=after_discovery.revision,
|
|
new_tasks=(
|
|
TaskProposal(
|
|
"test-input",
|
|
TaskKind.TEST,
|
|
target,
|
|
"Test the discovered input.",
|
|
"The behavior is confirmed.",
|
|
depends_on=("discover",),
|
|
),
|
|
),
|
|
next_task_id="test-input",
|
|
finish=False,
|
|
summary="Test the discovered input.",
|
|
),
|
|
after_discovery,
|
|
)
|
|
|
|
second_commit = memory.commit_plan(next_plan)
|
|
recovered = memory.snapshot("run-1")
|
|
|
|
assert second_commit.lease is not None
|
|
assert second_commit.lease.task_id == "test-input"
|
|
assert {task.id: task.status for task in recovered.tasks}["test-input"] is TaskStatus.ACTIVE
|
|
|
|
|
|
def test_blocked_execution_does_not_reenter_the_dependency_ready_queue(
|
|
tmp_path: Path,
|
|
) -> None:
|
|
memory = MemoryKernel(tmp_path / "state.sqlite3")
|
|
initial = memory.create_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
)
|
|
)
|
|
target = "http://127.0.0.1:8080"
|
|
plan_commit = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=0,
|
|
new_tasks=(
|
|
TaskProposal(
|
|
"exploit",
|
|
TaskKind.TEST,
|
|
target,
|
|
"Exploit the tested input.",
|
|
"The hypothesis is resolved.",
|
|
),
|
|
),
|
|
next_task_id="exploit",
|
|
finish=False,
|
|
summary="Try the exploit.",
|
|
),
|
|
initial,
|
|
)
|
|
)
|
|
assert plan_commit.lease is not None
|
|
|
|
after_attempt = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="exploit",
|
|
attempt_id=plan_commit.lease.attempt_id,
|
|
lease_revision=plan_commit.lease.revision,
|
|
trace_episode_id="executor-1",
|
|
outcome=ExecutionOutcome.BLOCKED,
|
|
summary="The prerequisite hypothesis was disproved.",
|
|
observation="The candidate behavior is a hard-coded decoy.",
|
|
evidence_sequences=(1,),
|
|
)
|
|
)
|
|
|
|
assert after_attempt.tasks[0].status is TaskStatus.FAILED
|
|
|
|
|
|
def test_progress_cannot_reopen_a_task_past_its_attempt_budget(tmp_path: Path) -> None:
|
|
memory = MemoryKernel(tmp_path / "state.sqlite3")
|
|
initial = memory.create_run(
|
|
RunSpec(
|
|
run_id="run-1",
|
|
goal="Assess the authorized target.",
|
|
allowed_targets=("http://127.0.0.1:8080",),
|
|
max_attempts_per_task=2,
|
|
)
|
|
)
|
|
first = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=0,
|
|
new_tasks=(
|
|
TaskProposal(
|
|
"enumerate",
|
|
TaskKind.ENUMERATE,
|
|
"http://127.0.0.1:8080",
|
|
"Enumerate the named surface.",
|
|
"The surface is recorded.",
|
|
),
|
|
),
|
|
next_task_id="enumerate",
|
|
finish=False,
|
|
summary="Begin enumeration.",
|
|
),
|
|
initial,
|
|
)
|
|
)
|
|
assert first.lease is not None
|
|
after_first = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="enumerate",
|
|
attempt_id=first.lease.attempt_id,
|
|
lease_revision=first.lease.revision,
|
|
trace_episode_id="executor-1",
|
|
outcome=ExecutionOutcome.PROGRESS,
|
|
summary="One bounded request remains.",
|
|
observation="First response",
|
|
evidence_sequences=(1,),
|
|
)
|
|
)
|
|
second = memory.commit_plan(
|
|
compile_plan(
|
|
SupervisorDecision(
|
|
base_revision=after_first.revision,
|
|
new_tasks=(),
|
|
next_task_id="enumerate",
|
|
finish=False,
|
|
summary="Use the final attempt.",
|
|
),
|
|
after_first,
|
|
)
|
|
)
|
|
assert second.lease is not None
|
|
|
|
exhausted = memory.commit_execution(
|
|
ValidExecution(
|
|
run_id="run-1",
|
|
task_id="enumerate",
|
|
attempt_id=second.lease.attempt_id,
|
|
lease_revision=second.lease.revision,
|
|
trace_episode_id="executor-2",
|
|
outcome=ExecutionOutcome.PROGRESS,
|
|
summary="The proposed completion evidence could not be resolved.",
|
|
observation=None,
|
|
evidence_sequences=(),
|
|
evidence_unresolved=True,
|
|
)
|
|
)
|
|
|
|
assert exhausted.status is RunStatus.RUNNING
|
|
assert exhausted.tasks[0].status is TaskStatus.FAILED
|
|
assert [attempt.status.value for attempt in exhausted.attempts] == ["progress", "progress"]
|
|
assert exhausted.transitions[-1].detail["attempt_limit_reached"] is True
|
|
assert exhausted.transitions[-1].detail["evidence_unresolved"] is True
|