@henryqw/pi-task-models

Shared fast, balanced, and frontier model profiles for HenryQW Pi extensions.

Install

pi install npm:@henryqw/pi-task-models

With

| Package | Why | | --- | --- | | @henryqw/pi-auto-compact | Consumer. Compaction defaults to the fast profile. | | @henryqw/pi-auto-dag | Consumer. Implement uses balanced; review uses frontier. | | @henryqw/pi-herdr-rename | Consumer. Rename uses the fast profile. | | @henryqw/pi-subagent | Consumer. Delegation defaults to balanced; caller may override it. | | @henryqw/pi-multi-codex | Improves. Numbered Codex slots dedupe to one route. |

Use

| Surface | Type | Purpose | | --- | --- | --- | | /task-models | command | Edit a profile or assign a task to a profile. |

Selecting a profile sets primary model, primary thinking, optional fallback model, then fallback thinking. One completed flow writes the whole profile. Selecting a task changes its assignment.

Menus and resolution use the current session's ctx.scopedModels, including pinned thinking. Empty scope uses Pi's full available model registry. Numbered Codex account aliases are deduplicated. Fallback choices exclude the selected primary. Hidden task assignments stay stored when a package is disabled.

Package also exports config and route-resolution helpers for consumers.

Config

~/.pi/agent/config/pi-task-models.json

{
  "profiles": {
    "fast": {
      "primary": { "model": "openai-codex/gpt-fast", "thinkingLevel": "low" },
      "fallback": { "model": "other-provider/fast-model", "thinkingLevel": "low" }
    },
    "balanced": {
      "primary": { "model": "openai-codex/gpt-balanced", "thinkingLevel": "high" }
    },
    "frontier": {
      "primary": { "model": "openai-codex/gpt-frontier", "thinkingLevel": "max" }
    }
  },
  "tasks": {
    "pi-herdr-rename/rename": "fast",
    "pi-auto-compact/autoCompact": "fast",
    "pi-subagent/delegateTask": "balanced",
    "pi-auto-dag/implement": "balanced",
    "pi-auto-dag/review": "frontier"
  }
}

Each configured profile needs one primary model and thinking level. Fallback is optional. Model references use canonical provider/model; numbered openai-codex-N routes store as openai-codex/model.

Reads are strict. Malformed or unknown values fail visibly and never rewrite the file. Only explicit /task-models actions write config.

Consumers can call resolveConfiguredTaskRoute(ctx, taskId) to read strict shared config, choose usable primary or fallback route, and fail with /task-models guidance when assignment, profile, or route is unavailable.

Remove

pi remove npm:@henryqw/pi-task-models

Development

npm test --workspace @henryqw/pi-task-models
npm run typecheck --workspace @henryqw/pi-task-models
npm run pack:check --workspace @henryqw/pi-task-models