Pi Agent: Security & Project Trust

セキュリティ3層モデル
├── 第1層：プロジェクト信頼 — Agentがプロジェクトにアクセスできるか
├── 第2層：ツール権限 — どのツールが許可されているか
└── 第3層：実行サンドボックス — ツール実行のセキュリティ境界

pi-agent trust set ./my-project --level trusted
pi-agent trust list

New project detected: /home/alice/my-project
Trust this project? [y/N]

trusted    - Full access (read/write files, execute commands)
readonly   - Read-only access
untrusted  - No access

Select trust level:

from pi_agent import Agent, TrustLevel

agent = Agent(
    name="safe_agent",
    project_path="./my-project",
    trust_level=TrustLevel.READONLY
)

permissions:
  tools:
    safe: auto
    medium: confirm
    dangerous: deny

  allow:
    - calculator
    - search
    - file_reader
  deny:
    - shell
    - file_write

  confirm_dangerous: true
  confirm_file_write: true
  confirm_shell: true

you> Delete all temporary files
Agent wants to execute: rm -rf /tmp/project_*
Allow? [y/N/a(always)]

from pi_agent import Agent

agent = Agent(
    name="sandboxed",
    sandbox=True,
    sandbox_dir="/tmp/pi_sandbox",
    allowed_commands=["python", "pytest"],
    max_execution_time=30
)

agent = Agent(
    sandbox=True,
    sandbox_dir="./workspace/sandbox",
    allowed_paths=["./data/", "./output/"]
)

agent = Agent(
    sandbox=True,
    network_access=False,
    allowed_domains=["api.github.com"]
)

agent = Agent(
    audit_log=True,
    audit_file="./audit.log"
)