APAI.runv0.1
Agents/Codex CLI

Codex CLI

OpenAI Codex CLI: local coding agent with shell access. Three approval modes (suggest, auto-edit, full-auto) control how much the agent acts without prompting.

local-tool
by OpenAIvaries (rolling release)Official docs →

Agent Passport

APAI.agent-passport.v0.1Agent Passport
local-toolunverified v0.1
Agent: Codex CLIVendor: OpenAIVersion: varies (rolling release)
Install modes:local-tool

Can read

  • ·Files and directories in current working directory
  • ·Git status and history
  • ·Shell command output

Can write

  • ·Files in current working directory (in auto-edit and full-auto modes)
  • ·Git commits (in full-auto mode)

Can access

  • ·Network endpoints reachable from the local machine (in full-auto mode)
  • ·OpenAI API (for model inference)
$

Can spend

  • ·OpenAI API tokens per session (billed to operator's OpenAI account)

Can expose

  • ·File contents and shell output sent to OpenAI API for inference

Will request approval for

  • ·All file writes in suggest mode
  • ·Destructive shell commands in auto-edit mode
  • ·All actions in suggest mode (shows diff before applying)

v0.1 note: This passport is descriptive, not enforceable. Runtime enforcement is the Policy Pack + Gateway story (Phase 4+). Fields sourced from publicly documented agent behavior as of May 2026.

Compatible packages (v0.1)

APAI packages that declare support for local-tool environments compatible with this agent. v0.1 match is by platform overlap; full capability-field matching ships with Phase 4.