Skip to main content
The CLI is designed to be operated by AI coding agents: stable JSON in and out, one error shape, exit codes to branch on, --dry-run to preview, and a hard --yes gate on anything that spends money. The local tools are covered too: streams test, streams forward and upgrade honour -o json and send their progress output to stderr, so stdout stays parseable.

​
Install the skill

plivo skill install writes a single reference file (SKILL.md) that agents load automatically. It documents every command, the JSON and error envelopes, the exit codes, headless auth, and the safety rules: without a network call, from the copy embedded in the binary.

​
Authentication

Since CLI v1.1.2 there is no headless login. PLIVO_AUTH_ID and PLIVO_AUTH_TOKEN are no longer read from the environment, so a person has to run plivo login once on the machine the agent works on; the agent then uses the stored profile. See Run in CI and from agents. Set PLIVO_FEEDBACK_PROMPT=0 and CI=1 to suppress interactive prompts.

​
Rules of the road for agents

  • Always pass -o json; parse data on success and error.code on failure.
  • Never pass --yes unless the human asked for the spend or deletion. Use --dry-run to show what would happen.
  • plivo <command> --help is the source of truth for flags. Do not invent flag names.
  • Use plivo ask for how-to questions and plivo diagnose for a failed call or message; both are rate-limited per account.
  • plivo docs search <keywords>, plivo docs list and plivo docs show <page> read the documentation with no credentials, so an agent can look something up before it has any. See Read the docs from your terminal.