clone: classify CHOOSING_YOUR_AI_TOOL.md as mixed

[Jammy2211]

The problem

autolens_assistant's clone-boundary CI has been red since the file landed on its main in 64018f8 ("Release 2026.8.17.1"):

check_boundary: 1 file(s) in autolens_assistant fall on neither side of the template boundary:
  ✗ CHOOSING_YOUR_AI_TOOL.md

This currently fails every PR against that repo — it surfaced on autolens_assistant#112, an unrelated hpc/sync bug fix. And it is exactly the condition the checker exists to catch: while it stands, the Clone Agent refuses to birth any newborn.

The fix is not a new decision

autolens_assistant's modes/maintainer.md — the prose that owns the boundary — already classifies this file under **Mixed** (line 101):

the free-tier chat surface — CHOOSING_YOUR_AI_TOOL.md and the generated llms-chat.txt (the per-platform setup mechanics clone verbatim; the worked prompts, dataset names and API rules are domain)

Only the code side was missing, so the two sources drifted apart. This restores agreement rather than deciding anything new — no change to maintainer.md is needed.

Placed in _SHARED_MIXED beside llms-chat.txt and FREE_TIER_SETUP.md, the two files that same sentence groups it with. The page exists only in autolens_assistant today; an unmatched pattern is inert for the other references, and a sibling that grows the same page inherits the same call.

Verification

Ran the exact checker CI invokes, against a real autolens_assistant checkout:

	Result
Before (this branch's base)	exit 1 — CHOOSING_YOUR_AI_TOOL.md unclassified
After	exit 0 — boundary complete: generic 59 · domain 324 · mixed 92

No regression on the other references: autofit_assistant still reports boundary complete (generic 104 · domain 20 · mixed 21); autogalaxy_assistant has no profile and correctly skips.

Merge order

autolens_assistant#112 declares Brain-ref pointing at this branch, so it can go green before this merges. Merge this one first, then #112 picks the same classification up from main.

---

Generated by Claude Code