项目文件夹

文件
Gelei Deng ab5fbb4d90 feat: ship the durable multi-model autonomous PentestGPT runtime (#493)
* 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>
2026-07-13 16:49:08 +08:00

166 行
5.7 KiB
Python

"""Agent Skills (agentskills.io): one canonical skill dir, both agents discover it.
Claude Code reads project skills from ``<ws>/.claude/skills/``; Codex reads the
cross-agent location ``<ws>/.agents/skills/``. ``install_skills`` validates each
skill against the open spec and links it into both, so a single SKILL.md serves
both agents. Validation enforces the spec's shared baseline; ``lint_skill``
flags Claude-only syntax that would degrade on Codex.
"""
from __future__ import annotations
import re
import shutil
from dataclasses import dataclass, field
from pathlib import Path
import yaml
from .types import SkillError
NAME_RE = re.compile(r"^[a-z0-9]+(-[a-z0-9]+)*$")
MAX_NAME_LEN = 64
MAX_DESCRIPTION_LEN = 1024
# Discovery roots inside a workspace, per host.
CLAUDE_SKILLS_DIR = Path(".claude") / "skills"
CODEX_SKILLS_DIR = Path(".agents") / "skills"
# Claude-only constructs that Codex (and other agentskills hosts) ignore.
_PORTABILITY_PATTERNS = [
("$ARGUMENTS", "argument substitution ($ARGUMENTS/$1...) is Claude-only"),
(
"!`",
"dynamic shell injection (!`cmd`) is Claude-only and runs before the model sees content",
),
("${CLAUDE_", "${CLAUDE_*} variables are Claude-only"),
]
@dataclass
class Skill:
name: str
description: str
path: Path
body: str
frontmatter: dict = field(default_factory=dict)
def _parse_frontmatter(text: str, where: Path) -> tuple[dict, str]:
if not text.startswith("---"):
raise SkillError(f"{where}: SKILL.md must start with YAML frontmatter (--- ... ---)")
end = text.find("\n---", 3)
if end == -1:
raise SkillError(f"{where}: unterminated YAML frontmatter")
raw = text[3:end]
body = text[end + 4 :].lstrip("\n")
try:
data = yaml.safe_load(raw) or {}
except yaml.YAMLError as e:
raise SkillError(f"{where}: invalid YAML frontmatter: {e}") from e
if not isinstance(data, dict):
raise SkillError(f"{where}: frontmatter must be a YAML mapping")
return data, body
def load_skill(skill_dir: Path) -> Skill:
"""Load and validate one ``<skill-dir>/SKILL.md`` per the agentskills spec."""
skill_dir = Path(skill_dir)
skill_md = skill_dir / "SKILL.md"
if not skill_md.is_file():
raise SkillError(f"{skill_dir}: no SKILL.md")
front, body = _parse_frontmatter(skill_md.read_text(encoding="utf-8"), skill_md)
name = str(front.get("name") or "")
description = str(front.get("description") or "")
if not NAME_RE.match(name) or len(name) > MAX_NAME_LEN:
raise SkillError(
f"{skill_md}: invalid skill name {name!r} "
f"(lowercase/digits/single-hyphens, <= {MAX_NAME_LEN} chars)"
)
if name != skill_dir.name:
raise SkillError(
f"{skill_md}: skill name {name!r} must match its directory name {skill_dir.name!r}"
)
if not description.strip():
raise SkillError(f"{skill_md}: description is required")
if len(description) > MAX_DESCRIPTION_LEN:
raise SkillError(f"{skill_md}: description exceeds {MAX_DESCRIPTION_LEN} chars")
return Skill(name=name, description=description, path=skill_dir, body=body, frontmatter=front)
def lint_skill(skill: Skill) -> list[str]:
"""Warnings for constructs that work in Claude Code but not in Codex."""
warnings = []
for needle, why in _PORTABILITY_PATTERNS:
if needle in skill.body:
warnings.append(f"{skill.name}: contains {needle!r}{why}")
return warnings
def discover_skills(source_dir: Path) -> list[Skill]:
source_dir = Path(source_dir)
if not source_dir.is_dir():
raise SkillError(f"skills source {source_dir} is not a directory")
skills = []
for child in sorted(source_dir.iterdir()):
if child.is_dir() and (child / "SKILL.md").is_file():
skills.append(load_skill(child))
return skills
def make_skill(parent: Path, name: str, description: str, body: str) -> Path:
"""Write a minimal spec-valid skill (used by examples and tests)."""
d = Path(parent) / name
d.mkdir(parents=True, exist_ok=True)
(d / "SKILL.md").write_text(
f"---\nname: {name}\ndescription: {description}\n---\n\n{body}\n",
encoding="utf-8",
)
return d
def _install_one(skill: Skill, target_root: Path, mode: str, force: bool) -> None:
target_root.mkdir(parents=True, exist_ok=True)
target = target_root / skill.name
source = skill.path.resolve()
if target.is_symlink():
if mode == "symlink" and target.resolve() == source:
return # already correct
target.unlink()
elif target.exists():
if not force:
raise SkillError(
f"{target} exists and is not a managed symlink; pass force=True to replace it"
)
shutil.rmtree(target)
if mode == "symlink":
target.symlink_to(source, target_is_directory=True)
elif mode == "copy":
shutil.copytree(source, target)
else:
raise SkillError(f"unknown install mode {mode!r} (use 'symlink' or 'copy')")
def install_skills(
workspace: Path,
source_dir: Path,
mode: str = "symlink",
force: bool = False,
) -> list[Skill]:
"""Install every skill under ``source_dir`` into BOTH hosts' discovery dirs.
Copy mode replaces previously-copied skills on reinstall (the target dirs
under ``.claude/skills`` / ``.agents/skills`` are treated as managed).
"""
workspace = Path(workspace)
skills = discover_skills(source_dir)
if not skills:
raise SkillError(f"no skills found under {source_dir}")
for skill in skills:
for root in (workspace / CLAUDE_SKILLS_DIR, workspace / CODEX_SKILLS_DIR):
_install_one(skill, root, mode=mode, force=force or mode == "copy")
return skills