Documentation Index
Fetch the complete documentation index at: https://mintlify.com/xxyoudeadpunkxx/a.d.a.m.-adaptive-depth-and-mode/llms.txt
Use this file to discover all available pages before exploring further.
TRACE INPUT is an operator-grade diagnostic command that emits a strict one-line structural trace of the last visible user message. It is used to verify what the structural kernel actually sees — useful when debugging unexpected routing behavior.
INPUT_TRACE: len=<L> head="<H>" tail="<T>" struct=[opt:<n> step:<n> crit:<n> num:<Y|N>]
TRACE INPUT command turn itself):
| Field | Meaning |
|---|
len | Raw character count of the traced message as seen by the model. If unreliable, len=0. |
head | First 16 characters of the traced message after normalization. Newlines replaced with spaces; " replaced with '. |
tail | Last 16 characters of the traced message after normalization. Same replacement rules as head. |
struct | Structural summary recomputed from the traced message using the same primitives as section S of the spec. |
struct field contains four sub-fields:
| Sub-field | Meaning |
|---|
opt:<n> | Count of line-start option blocks (A), B), 1), 2), …) |
step:<n> | Count of line-start step blocks (1., 2., 3., …) |
crit:<n> | Count of criteria items (line-start - or *, excluding option and step blocks) |
num:<Y|N> | Y if the traced message would set HAS_NUM true (2+ digits, digit range, comparator with digit, percent with digit, or dates); N otherwise |
How to use it
Send the exact command while ACTIVE:
TRACE INPUT traces the previous user turn — not the TRACE INPUT command message itself. If you want to inspect a specific input, send that input first, then send TRACE INPUT on the following turn.
Normative examples
| Prior message | Expected struct output |
|---|
A message with 2 option blocks (A) / B)), 3 criteria items, and one number | struct=[opt:2 step:0 crit:3 num:Y] |
| Plain prose with no options, steps, criteria, or numbers | struct=[opt:0 step:0 crit:0 num:N] |
A message with 3 step blocks (1. / 2. / 3.) and a date | struct=[opt:0 step:3 crit:0 num:Y] |
Automatic structural overlay
TRACE INPUT as a command and the automatic structural overlay that may appear in MODE: MID -> POSSIBLE DEEP or MODE: DEEP replies are related but serve different purposes:
| TRACE INPUT (command) | Automatic overlay |
|---|
| Who triggers it | Operator sends the exact command | Emitted automatically by the protocol |
| Output format | Strict one-line machine-readable INPUT_TRACE: form | Human-readable chip labels (e.g. [confronto tra opzioni]) |
| When emitted | Any time while ACTIVE | Only in MODE: MID -> POSSIBLE DEEP and qualifying MODE: DEEP replies |
| Shown when | Always (or UNAVAILABLE if no prior turn) | Only when structural thresholds are met |
| Purpose | Forensic/operator-grade trace for debugging | Presentation-only structural mirror for user context |
opt, step, crit, num) and the same structural counting logic. However, the automatic overlay is presentation-only — it does not imply that TRACE INPUT was explicitly invoked, does not change routing or gating, and is not a substitute for the strict forensic command.
Unavailable case
If no prior user-authored turn is visible before TRACE INPUT, the output is exactly:
TRACE INPUT is unavailable in TRANSPORT and OFF states. While TRANSPORT, it
returns the standard 2-line transport gate response. While OFF, it returns:Use TRACE INPUT when a reply is routed differently than you expect. The
struct field shows the exact structural classification that drove the routing
decision — if opt, crit, or num are higher than expected, the kernel
will have seen a stronger structural signal than the message appeared to carry
visually.