choose

An interactive selector for choosing whether AnyModel should delegate a task or run an adversarial review, then choosing a provider and model. A provider is the service supplying the model; the model is the specific AI system doing the work.

In plain words
What is it for?
Starting implementation or debugging work, requesting a challenge or audit of work, and selecting an available provider/model combination.
Why use it?
It avoids having to remember provider names, model names, and the correct action format when starting a job.

Command

Part of the anymodel plugin — 11 commands, 1 agent shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/pealmeida/anymodel-plugin/choose
Clone the repo
git clone --depth 1 https://github.com/pealmeida/anymodel-plugin

Or install anymodel, the plugin that ships this one along with the rest of its 11 commands, 1 agent.

Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 793 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00015 $0.00793
Opus 5 $0.00008 $0.00396
Sonnet 5 $0.00003 $0.00159
Haiku 4.5 $0.00002 $0.00079

Measured 2d ago against content hash 25b758c17146, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

choose scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/anymodel/commands/choose.md · 63 lines

What it actually says

Guide the user through a semantic AnyModel chain, then dispatch to the existing runtime. This command is a selector only; do not implement delegation or review logic here.

Raw user request: $ARGUMENTS

Semantic chain:

  1. Determine the action.

    • If the request clearly says delegate, execute, run, fix, implement, or solve, choose delegate.
    • If it says review, adversarial, challenge, critique, or audit, choose adversarial-review.
    • Otherwise use AskUserQuestion once:
      • Delegate task
      • Adversarial review
  2. Determine provider/model.

    • First preserve any explicit --model <provider/model> argument.
    • Also accept natural forms like with zai/glm-5.2, using ollama/qwen3-coder, or on opencode-go/kimi.
    • If an exact <provider>/<model> is not present, run:
      node "${CLAUDE_PLUGIN_ROOT}/scripts/companion.mjs" models --json
      
    • Parse the JSON. Ignore providers with ok: false or an empty models list.
    • If no usable provider/model is available, stop and return the models output verbatim, then tell the user to run /anymodel:setup.
    • If the user named a provider but not a model, ask them to choose one reachable model for that provider.
    • If the user named neither provider nor model, ask them to choose from the most useful reachable options. Prefer showing up to three concrete choices as <provider>/<model>, using the provider/model names exactly as returned by models --json.
  3. Determine the engine.

    • Preserve any explicit --engine <id>.
    • If omitted, do not ask. Let the companion runtime use its configured default engine.
  4. Determine the work text.

    • For delegate, require a task. If the request does not include one after removing action/provider/model selection words, ask what the selected model should investigate, solve, or implement.
    • For adversarial-review, focus text is optional. Preserve any provided focus text exactly.
  5. Dispatch.

    • Build a final argument string from the original flags that are still relevant, the selected --model <provider/model>, and the task/focus text.
    • Preserve --engine, --bridge, --write, --base, --scope, --wait, --background, --resume, and --fresh when present.
    • Do not include selector-only words like delegate with, review using, or choose.

Delegate dispatch:

  • Invoke the anymodel:anymodel-runner subagent via the Agent tool (subagent_type: "anymodel:anymodel-runner"), forwarding the final request as if the user had called /anymodel:delegate.
  • The subagent must make exactly one Bash call to:
    node "${CLAUDE_PLUGIN_ROOT}/scripts/companion.mjs" delegate <final arguments>
    
  • Return the companion stdout verbatim.

Adversarial review dispatch:

  • Run:
    node "${CLAUDE_PLUGIN_ROOT}/scripts/companion.mjs" adversarial-review <final arguments>
    
  • Return stdout verbatim. Do not summarize, fix, or add commentary after the review.
Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 63 lines · 15 tokens per session scan A 25b758c17146

Subscribe to this mod's changes

choose is a command published in the GitHub repository pealmeida/anymodel-plugin (0 stars, last pushed 24d ago), licensed Apache-2.0. It adds 15 tokens to every session and 793 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other commands, from other repositories

/commit-push

Commit and push all changes from this chat session using Conventional Commits.

DataCovey/nornweave · 18 tokens

/release

Release a new version — updates CHANGELOG, pyproject.toml, creates git tag, and pushes.

DataCovey/nornweave · 23 tokens

mega-approve

Approve the mega-plan and start feature execution. Creates worktrees and generates PRDs for each feature. Usage: /plan-cascade:mega-approve [--flow ] [--tdd ] [--confirm] [--no-confirm] [--spec ] [--first-principles] [--max-questions N] [--auto-prd] [--agent ] [--prd-agent ] [--impl-agent ].

Taoidle/plan-cascade · 110 tokens

auto

AI auto strategy executor. Analyzes task and automatically selects and executes the best strategy: direct execution, hybrid-auto PRD generation, hybrid-worktree isolated development, or mega-plan multi-feature orchestration.

Taoidle/plan-cascade · 41 tokens

hybrid-auto

Generate PRD from task description and enter review mode. Auto-generates user stories with priorities, dependencies, and acceptance criteria for parallel execution. Usage: /plan-cascade:hybrid-auto [--flow ] [--tdd ] [--confirm] [--no-confirm] [--spec ] [--first-principles] [--max-questions N] [--agent ]…

Taoidle/plan-cascade · 107 tokens

hybrid-worktree

Start a new task in an isolated Git worktree with Hybrid Ralph PRD mode. Creates worktree, branch, loads existing PRD or auto-generates from description. Usage: /plan-cascade:hybrid-worktree [--flow ] [--tdd ] [--confirm] [--no-confirm] [--spec ] [--first-principles] [--max-questions N] [--agent ] [design-doc-path].

Taoidle/plan-cascade · 125 tokens