Approve a tool run

Safe by default

File writes and shell commands are risky. On an interactive terminal PraisonAI asks before they run.

In CI or a pipe, the same tools are denied unless you set an explicit policy.

Example

from praisonaiagents import Agent

agent = Agent(
    name="Dev Agent",
    instructions="Help with file ops",
    tools=["write_file", "execute_command"],
)
agent.start("create a hello.txt file")

Example explained

• tools — High-risk tools trigger an approval prompt on a TTY.
• agent.start — When the model picks a tool, you choose once, session, always, or no.

Deny in automation

For pipelines, turn off prompts and keep deny-by-default so a misconfigured runner cannot delete files silently.

Example

from praisonaiagents import Agent

agent = Agent(
    name="CI Agent",
    approval=False,
    tools=["write_file", "execute_command"],
)
agent.start("run tests and update report")

Example explained

• approval=False — No prompt; dangerous tools stay denied off-TTY style.

Declarative rules

For YAML runs you can allow and deny patterns up front instead of clicking each time.

Example

praisonai run agents.yaml \
  --allow 'read:*' \
  --allow 'bash:git *' \
  --deny 'bash:rm *' \
  --permission-default ask