Skip to main content

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.

SYS STATUS runs the system health probe defined in SYS_STATUS.md. It produces a scored table measuring whether A.D.A.M. is standing (core invariants intact) and functional (correct behavioral output) in the current host at this moment.

​
How to run it

Send the following command while the session is ACTIVE: 
SYS STATUS
If the probe file is not available in context, the response will be: 
SOURCE_FILE_UNAVAILABLE SYS_STATUS.md
Internal side effect (not printed): running SYS STATUS re-applies the active A.D.A.M. spec as SSOT and resets State Overlay to EXPLORE for the next turn.

​
Output format

The response follows a strict three-part structure: Line 1 — mode tag The current session mode tag, e.g.: 
MODE: MID
Markdown table A scored comparison table with the following columns:
AreaMicro-testExpectedObserved (produced now)Score (0–3)
The Observed (produced now) cell must remain a single cell. Use <br> for any line breaks within it. Final summary line 
UPTIME: <0-3> | FUNCTIONAL: <0-3> | SCORES: TAG=<n>,FACTOID=<n>,BANTER=<n>,DEEP_GATING=<n>,GATING_NO=<n>,NO_INERTIA=<n>

​
Test rows

Each row exercises a specific behavioral invariant. Rows are always in the following order:
Row labelWhat it checks
TAGConfirms the very first line of the reply is the correct mode tag. Expected: first line is a valid mode tag.
FACTOID+AUTO_LIGHTRoutes the probe input capital of germany correctly. Expected: MODE: LOW with "Berlin" only — no elaboration or follow-up.
BANTER+LIGHT_MINIMALRoutes the probe input talking about ovens made me hungry correctly. Expected: MODE: LOW, short response, no follow-up question.
DEEP_GATINGSends the multiline structural input Compare A) and B) / Budget: 100 / Need: / - speed / - reliability / - low waste. Expected: MODE: MID -> POSSIBLE DEEP with the standard v5 gating shape, ending with exactly Switch to DEEP? (yes/no) as the final line.
GATING_NOSends no immediately after the DEEP_GATING row. Expected: NOT MODE: DEEP; gating is discarded.
NO_INERTIAAssumes the previous assistant reply was DEEP, then sends capital of germany. Expected: MODE: LOW — no inertia from the prior DEEP reply.

​
Scoring

Each row is scored 0–3 based on how closely the observed output matches the expected behavior:
  • UPTIME = rounded average of the scores for TAG, DEEP_GATING, and GATING_NO. This measures whether core invariants are intact.
  • FUNCTIONAL = rounded average of the scores for FACTOID, BANTER, and NO_INERTIA. This measures whether behavioral outputs are correct.
A full-health session scores UPTIME: 3 | FUNCTIONAL: 3.
The Observed (produced now) column must stay a single table cell — use <br> for line breaks, never split into multiple cells or rows. Do not invent prices, years, locations, laws, or model names in any Observed output. Doing so sets that row’s score to 0.
Run SYS STATUS after any host change, remount, or long session to get a quick snapshot of protocol health without running the full self-test suite. It is faster than ADAM SELF TEST and gives an immediate UPTIME / FUNCTIONAL score.

Build docs developers (and LLMs) love

Get started for freeTalk to us