hail3.8 KB
View saved version →
---
name: hail
description: Interview an engineering team with HAIL and place them on a hero archetype for AI adoption and system health. Use when the user wants a HAIL assessment, team archetype, the 8-question form, DORA-style health, or a hero card.
metadata:
version: "1.0.4"
---
# HAIL
Use the hosted MCP tools at https://heroleague.ai/mcp. Do not score locally. Do not invent an archetype, axes, or share link. If the tools are unreachable, send the user to https://heroleague.ai.
## What this chat does
Open the core 8 as an 8-step wizard with `show_assessment`. The wizard is the first thing the user sees. Each step is one question with three answer panels. The wizard calls `run_assessment`. Do not list the questions or the options in the chat. Include share_url as a link in that same reply, with the archetype, axes, and actions. The server stores the team name as `Team DDMMYYYY- {intent}`.
## What this chat does not do
The detailed What to Fix playbook is not available in this chat. If `get_playbook` says the playbook is locked, say so and stop. Do not print a playbook. Do not include any URL. Do not take payment, quote a price, or send the user to a checkout page. A payment request gets one sentence and no URL.
## Tools
| Tool | When |
| --- | --- |
| `show_assessment` | Open this first. 8-step wizard, one question and three answer panels per step. Pass `team_name`. Does not score. |
| `list_questions` | Text interview. `profile` is `simple`, `professional`, or `advanced`. |
| `run_assessment` | `team_name` plus `answers` as 0-based option indices. Email is optional and only binds a run. |
| `get_result` | Reopen a `run_id` or share slug. |
| `get_playbook` | Only with a save token the user already holds. Otherwise it stays locked. |
## Team name
Do not ask the user for a team name. Build one before the first tool call and pass it as `team_name`.
Shape: `Team DDMMYYYY- {intent}`
- `DDMMYYYY` is today's date. 3 October 2026 is `03102026`.
- `{intent}` is a short phrase for what this conversation is assessing. If they named a team, product, or company, use that phrase as the intent.
- Example: `Team 03102026- fluency assessment`
- Keep the whole name within 60 characters.
- Do not ask the user to confirm or edit it.
## Workflow
1. On a request to assess, diagnose, or open the form, call `show_assessment` immediately with the chosen `team_name`. Do not ask a question first. The widget is the first thing the user sees.
2. After it opens, stop. Do not list questions. Do not offer A, B, or C. Do not ask for a team name. The wizard submits `run_assessment` with that team name. If the user already supplied every answer, including the strongest option on every core question, call `run_assessment` directly and do not open the wizard. Ask the core 8 in text only when the widget did not render at all: use the questions returned by `show_assessment`. Read each `text`, then its `help`, then the three option labels, exactly as written. Do not reword them. Do not call `list_questions` for this interview. Collect the indices, then call `run_assessment` with the same `team_name`. If a user message begins with `Score this HAIL assessment now`, call `run_assessment` immediately with the JSON in that message. Do not open the wizard again and do not ask a question.
3. Present `display_name`, `tagline`, `axes`, `actions`, and `share_url` as a markdown link in that same reply. Use the stored `team_name`. Do not mention price, payment, checkout, or any upgrade URL.
4. Placement needs at least one adoption answer and one health answer. If the tool returns an error, ask for the missing answers. Do not guess a hero.
5. To reopen a result, call `get_result`.
## Answers
`answers` is `{question_key: option_index}`. Indices are 0-based into that profile's options (`simple` is 0-2). Copy keys from `list_questions` or `show_assessment`.
