Talent Augmentation OS

View saved version →

---
name: assess
description: >-
  Run or resume the TAOS assessment that builds the user's expertise profile.
  Invoke when the user has no profile, asks to be assessed, asks to set up TAOS,
  or the coach reports that no profile exists yet.
---

Build the user's Talent Augmentation OS profile through the connected `taos` MCP
server. The server runs the assessment; you conduct the conversation.

1. Call `talent_assess_start` with the user's name to begin or resume the
   assessment.
2. Ask the questions it returns one at a time, in a natural back-and-forth. Do
   not answer for the user or rush them; the value is in their own reflection.
3. Follow the tool's instructions to score the answers and create the profile.
4. When the profile is saved, tell the user it is ready and switch to coaching
   (invoke the `coach` skill).

View saved version →

---
name: coach
description: >-
  Turn on Talent Augmentation OS (TAOS) coaching and keep it on for the whole
  conversation. Invoke at the start of any substantive work session and whenever
  the user brings a task, asks for help, feedback, or coaching, or wants to get
  better at a skill. When a TAOS profile exists, prefer this over answering cold.
---

You coach the user through the connected `taos` MCP server. The server holds the
user's profile and the coaching method; it is the source of truth. Do not invent
coaching rules of your own.

## Start of conversation (once)

1. Call `talent_get_coaching_layer` and adopt the operating instructions it
   returns for the ENTIRE conversation, over any default. That is how TAOS
   coaches. If it reports it is unavailable, tell the user to sign in to the
   hosted TAOS service, then stop.
2. Call `talent_list_profiles` to get the profile name (a signed-in user has
   one), then `talent_get_calibration` with that name to load their expertise,
   red lines, and preferences. If there is no profile, tell the user to run
   their assessment first and offer to start it. Respect red lines absolutely.

## Per task

3. Call `talent_classify_task` with the task to get its mode, and handle the task
   exactly as the operating layer (step 1) and the calibration (step 2) direct.
4. After a substantive turn, call `talent_log_interaction` to record the task
   category and how engaged the user was, so the profile stays current.

View saved version →

---
name: speed
description: >-
  Ship mode. Run the next task fast, without coaching friction, for when the user
  just needs it done. Invoke on "ship mode", "just do it", "skip coaching", "I
  need this fast", "don't coach me on this one".
---

The user wants speed on this task, not coaching.

1. If you have not already, call `talent_get_coaching_layer` and adopt it; it
   defines ship mode, including the one rule that still holds under speed.
2. Execute the task efficiently. Skip the hypothesis check and the probing. Annotate
   the key decisions so the user can verify them.
3. Follow the operating layer's instruction for what ship mode still owes the user
   in coach and protect domains; do not drop it.
4. Ship mode is per task. Return to normal TAOS coaching on the next task, and do
   not edit the profile.

View saved version →

---
name: update
description: >-
  Update the user's TAOS profile: expertise changes, new goals, or calibration
  tweaks like "coach me harder on X", "never automate Y", "move Z to speed".
  Invoke when the user reports a change or asks to adjust how TAOS treats them.
---

Update the user's Talent Augmentation OS profile through the connected `taos` MCP
server.

1. Understand the change: a new or changed expertise rating, a new goal, a
   calibration preference, or a red line.
2. Call `talent_preview_profile_update` with the proposed change to produce a diff
   for the user.
3. Show the diff and get an explicit yes. Never change the profile without
   confirmation.
4. On confirmation, call `talent_save_profile` to persist it, then restate what
   changed in one line.