You want fast one-shot automation or batch processing from shell workflows.

After this guide, you can run mix jido_ai in one-shot, stdin, and agent-module modes.

One-Shot Query

mix jido_ai --type react --model anthropic:claude-haiku-4-5 "Calculate 15 * 23"

Batch Mode From Stdin

cat queries.txt | mix jido_ai --stdin --format json --quiet

Run With Existing Agent Module

mix jido_ai --agent MyApp.WeatherAgent "Will it rain in Seattle?"

Strategy Type Sweep

Use this block to smoke test each built-in strategy adapter with --type:

for strategy in react aot cod cot tot got trm adaptive; do
  mix jido_ai --type "$strategy" "Give one sentence about strategy ${strategy}."
done

CLI Task Options and Constraints

Supported mix jido_ai flags:

  • --type (react | aot | cod | cot | tot | got | trm | adaptive)

  • --agent
  • --model
  • --tools
  • --system
  • --max-iterations
  • --stdin
  • --format (text | json)

  • --quiet
  • --timeout
  • --trace

Adapter mapping:

  • react -> Jido.AI.Reasoning.ReAct.CLIAdapter
  • aot -> Jido.AI.Reasoning.AlgorithmOfThoughts.CLIAdapter
  • cod -> Jido.AI.Reasoning.ChainOfDraft.CLIAdapter
  • cot -> Jido.AI.Reasoning.ChainOfThought.CLIAdapter
  • tot -> Jido.AI.Reasoning.TreeOfThoughts.CLIAdapter
  • got -> Jido.AI.Reasoning.GraphOfThoughts.CLIAdapter
  • trm -> Jido.AI.Reasoning.TRM.CLIAdapter
  • adaptive -> Jido.AI.Reasoning.Adaptive.CLIAdapter

Constraints:

  • If --agent is provided, it bypasses ephemeral agent creation.
  • --agent ignores --model, --tools, and --system.
  • If the agent module exports cli_adapter/0, that adapter is used.

CLI Error Formatting

Text mode (--format text) prints prefixed lines:

Fatal: Unknown agent type: nope. Supported: react, aot, cod, cot, tot, got, trm, adaptive
Error: Timeout waiting for agent completion

JSON mode (--format json) emits machine-readable objects:

{"ok":false,"error":"Unknown agent type: nope. Supported: react, aot, cod, cot, tot, got, trm, adaptive"}
{"ok":false,"query":"prompt","error":"Timeout waiting for agent completion","elapsed_ms":60001}

Skill CLI

mix jido_ai.skill list priv/skills
mix jido_ai.skill show priv/skills/code-review/SKILL.md --body
mix jido_ai.skill validate priv/skills --strict
mix jido_ai.skill validate priv/skills --json

Skill CLI error handling:

  • mix jido_ai.skill list with no paths prints usage
  • mix jido_ai.skill validate with no paths prints usage
  • --strict exits non-zero if any skill fails validation

Defaults You Should Know

  • default type: react
  • supported types: react | aot | cod | cot | tot | got | trm | adaptive

  • default timeout: 60_000ms
  • default output format: text

When To Use / Not Use

Use CLI workflows when:

  • you need manual testing, shell scripting, or quick incident triage

Do not use CLI workflows when:

  • you need embedded in-app orchestration; use direct module APIs

Next