Gameplay Framework Walkthrough (IV) — NEONGRID, a grid run (and the borrow mechanic)

• runtime/mission-control.md (board ↔ UDS) · runtime/mission-objectives.md (objective graph) · runtime/currency-and-economy.md (currencies)
• cartridges/design-bibles/mission-type-catalog.md (A) · cartridges/authoring/verb-partition.md + cartridges/design-bibles/verb-taxonomy-map.md (B) · runtime/recon-mode.md (C) · programs/enrichment-contract.md (D) · cartridges/authoring/capability-shapes.md (E) · programs/knsalk.md (F) · cartridges/authoring/lisp-paradigm.md (Tier-1 as play) · runtime/baselines/grid.md (the :grid baseline) · cartridges/modules/neon-grid.md (the cart)

Scope: the smallest job in the series — one cart (NEONGRID), training-tier, a single PENETRATION. The verb is PENETRATE, the affinity PHYSICAL. The point isn’t a new mechanic; it’s that the framework’s floor — a cart that borrows verbs and barely leaves Tier-1 — still rides the same board → objective → capability → resolve spine as ICE BREAKER did.

---

The contract: DEAD ANGLE — cross a patrolled grid, tag the checkpoints in order, leave no trace

0 · The board

Operator HOSHIKO boots. Mission Control synthesizes the board seeded by Universal Deck State — reputation sets the threat tier, cartridge_history shows NEONGRID registered (bit 0x02), cipher_seed is the entropy. NEONGRID is inserted, so it supersedes the System-tier :grid baseline (the bare calibration loop) and the board leans physical. NEONGRID is training-tier (threat-cap 3), so the contracts run low. One surfaces: PENETRATION · PHYSICAL · T2 · ¤200. → foundations: Board ↔ UDS (mission-control.md §3); the supersedes mechanic (§2.3) over the :grid baseline (baselines/grid.md)

1 · The type

The board prints the class PENETRATION = the Tier-3 verb PENETRATE × the affinity PHYSICAL — the player-legible “get through the grid” — instantiated by NEONGRID’s TRAVERSE template (visit the checkpoints in sequence). Two of NEONGRID’s other verbs ride along: OBSERVE (reading the patrols → SURVEILLANCE) and DECIDE (committing at a checkpoint under pressure → COMMITMENT). → A (mission-type-catalog.md — PENETRATION / SURVEILLANCE / COMMITMENT)

1½ · The borrow — what NEONGRID puts on the verb layer

Here is what makes this cart different. The first three carts each owned a Tier-2 spine (INTRUSION, REMOTE OPS, FORENSICS). NEONGRID owns none — it is a borrower. Its action vocabulary is a thin sentry-evasion slice, THREAD / SCATTER / SHADE, lifted from the INTRUSION + SIGNALS spines (ICE BREAKER’s and DRIFT/SHELLFIRE’s). The Tier-2 layer is a shared vocabulary, not a per-cart silo, and the partition rule (B) treats an owned verb and a borrowed verb identically:

Tier	What NEONGRID contributes
Tier-1 primitives	the whole movement surface — CAR drill-forward, CDR step-sideways, BACK retrace, ATOM test-dead-end, EQ compare two marked cells, numpad N/S/W/E. NEONGRID is where these are taught.
Tier-2 verb-tools	none native — borrows the evasion slice THREAD (thread between patrols) / SCATTER (break a coordinated pattern) / SHADE (slip detection), all CORE, from INTRUSION + SIGNALS
Tier-3 verbs	PENETRATE · OBSERVE · DECIDE

Because it is the onboarding cart, most of DEAD ANGLE is played in Tier-1: the Lisp keys themselves, as embodied spatial movement. NEONGRID is where the operator’s hands first learn the grammar the rest of the deck is built on. → B the borrow (verb-taxonomy-map.md §3, NEONGRID borrower row; verb-partition.md) · Tier-1 as gameplay (lisp-paradigm.md)

2 · Accept → the objective graph loads

○ P1  Tag checkpoint α                     ¤ 60        primary · briefed · on-complete
○ P2  Tag β then γ, in sequence            ¤140        primary · briefed · :requires P1 · on-resolve
+ O1  Ghost run — :hold (alert ≤ CAUTIOUS)  +2 REP      optional · briefed · on-resolve (escrow)
◌ O2  [latent] Side cache off the patrol    +MASTERY    optional · revealed @ dead-end spur · on-resolve

Mission succeeds when P1 + P2 are :done (checkpoints in order). The time limit and the alert ceiling are not new subsystems: the engine watches them as the same :hold-and-TTL machinery the objective model owns — if a sentry reaches her cell the goal goes :failed (respawn at the last checkpoint; Rep −2). → foundations: objective graph (mission-objectives.md §3, ADR-0043)

3 · Recon — in-cart OBSERVE (no program)

Unlike I–III, NEONGRID hands no world to a program: there is no spatial-grid lens in the program roster, and the pack-in cart is self-contained. Her recon is the cart’s OBSERVE phase — she presses INFO for the sonar/heatmap view and reads the patrol intelligence: Sentry A’s 3-turn east-west cycle, the dead angle just north of its loop. Case the patterns first (measured) or sprint in cold (rip-and-run) — recon-mode’s same axis, run inside the capability. No DOSSIER key is deposited; the intel lives in her head and the run. → C recon mode — the measured/rip axis, here in-cart (recon-mode.md) · D not fired — no program-enrichment (no spatial lens; enrichment-contract.md §7)

4 · No DOSSIER — this cart trades in muscle, not knowledge

And so step 4 of the other walkthroughs — the fact-node landing in DOSSIER — doesn’t happen here. NEONGRID is a mastery cart, not a knowledge cart: what a clean run leaves behind is proficiency, not a typed key in the library. (Where COLD LEDGER fed the DOSSIER, DEAD ANGLE feeds the hands.) → foundations: the four currencies, and that knowledge is only one of them (currency-and-economy.md)

5 · Jack in — (load-capability :neongrid) — the grammar made physical

NEONGRID’s capability is a mini-game — the real-time spatial maze, OBSERVE-LEARN-EXECUTE-VERIFY (down to ~0.5 s cycles in flow state). She moves on Tier-1 primitives: CAR to drill forward through cleared cells, the numpad to step N/S/W/E, BACK to retrace, ATOM to test a dead-end, EQ to compare two marked cells. Every keypress is heard — audio detection raises a sentry’s alert by +1; silent turns let it decay — so precision is stealth. When Sentry A’s cycle pins her, she reaches for the borrowed Tier-2 slice: THREAD through the A/B patrol overlap, SHADE into a dead-end spur to bleed the alert back down. She tags α → (goal-complete 'P1) → ¤60 banked (on-complete). → E capability shape: mini-game (capability-shapes.md) · B Tier-1 movement + borrowed THREAD/SHADE · the audio-detection heat (neon-grid.md §3)

6 · The fork

Threading toward β she spots a spur off Sentry B’s loop → (goal-reveal 'O2) (a side cache, +MASTERY). Taking it means extra keypresses deeper in the patrol — alert climbs toward ALERT, threatening the O1 ghost. She goes for it: SCATTER to break B’s intercept, then THREAD back onto the β→γ line. Alert spikes to the top of CAUTIOUS. → foundations: latent reveal + risk-coupled escrow (mission-objectives.md §3); the cart’s sentry-prediction collision (neon-grid.md §6)

7 · Nobody hacks the deck — and that is the framework, again

This is the F beat, and it parts from I/II and from III. NEONGRID has a live heat meter — the sentry alert — and she even spends borrowed evasion verbs against it. But the threat is in-world and training-tier: if a sentry reaches her cell the run fails (respawn at the last checkpoint, Rep −2) — the grid never traces her deck. Heat here is local to the capability; it never routes to a deck-level counter-intrusion, so knSALK stays dormant. Two carts now, two reasons F is dark: COLD LEDGER raised no heat (cold analysis); DEAD ANGLE raises heat but contains it. knSALK answers offense the target can trace home — a maze guard isn’t that. → F knSALK (knsalk.md) — not fired, by design; contrast §7 of I / II

8 · Resolve — (complete-mission)

Both primaries :done (checkpoints in order). She reaches the exit clean. Settlement:

Goal	Reward	Outcome
P1	¤60	kept (banked on-complete)
P2	¤140	paid (mission succeeded)
O1	+2 REP	forfeited — the cache detour pushed alert past CAUTIOUS
O2	+MASTERY	paid — cache grabbed, run survived

One sanctioned UDS write (ADR-0040 §6): credit_balance += 200 (training-cheap, by design), reputation += the success delta (no ghost bonus), and — the part that matters for this cart — NEONGRID mastery (the proficiency vector) climbs hard: a tight run sharpens hand-eye + decision speed, which biases every future cart’s difficulty grade. cipher_seed advances; phase_chain clears. No DOSSIER write — nothing to bank but the muscle. → foundations: economy — weighted to mastery, the per-cart proficiency currency (currency-and-economy.md §7)

---

Coverage — what fired, and what didn’t

Piece	Where
A — mission-type catalog	PENETRATION = PENETRATE × PHYSICAL (§1)
B — verb partition	the borrow — Tier-1 movement + borrowed THREAD/SCATTER/SHADE (§1½, §5) — the centerpiece
C — recon mode	in-cart OBSERVE of sentry patrols, measured vs rip (§3)
D — program enrichment	not fired — no spatial-grid lens; the cart is self-contained (§3)
E — capability shapes	NEONGRID mini-game (§5)
F — knSALK	not fired, by design — sentry alert is in-world heat, not a deck trace (§7)
objective model	goal graph + escrow forfeit + time/alert as :hold/TTL (§2, §6, §8)
economy	weighted to mastery (proficiency); ¤ + reputation move, knowledge does not (§8)
DOSSIER	not written — muscle, not knowledge (§4)
Board ↔ UDS	seeded board (+ :grid supersede), sanctioned write (§0, §8)

Why this proves coherence — the floor test

Three things this run shows that the others couldn’t:

1. The Tier-2 layer is shared, not siloed. A cart can own a spine (I/II/III) or borrow a slice — NEONGRID lifts THREAD/SCATTER/SHADE from INTRUSION + SIGNALS and the partition rule (B) doesn’t blink. Verbs are a common vocabulary the whole roster draws from; NEONGRID proves the borrow path runs end to end.
2. The framework scales down to the pack-in cart. NEONGRID is the minimal cart — barely more than the bare :grid deck it supersedes. It rides the full spine (board → objective graph → capability → resolve → one UDS write) while touching neither program-enrichment (D) nor DOSSIER nor knSALK (F). The smallest cart and the largest (ICE BREAKER, all three tiers) run on the same machine: the framework is a floor as much as a ceiling.
3. Heat is a property of the job, not a fixture. Four carts, three heat stories: ICE BREAKER and DEPTHCHARGE route heat to the deck (knSALK fires); BLACK LEDGER raises none (cold analysis); NEONGRID raises heat but contains it in-world (caught, not counter-hacked). The system predicted all three.

And the quiet one under all of it: NEONGRID is where the operator’s hands first learn CAR / CDR / EVAL as movement — the very keys that, three carts on, breach a subnet, lift a sunk core, and CONS forensic evidence into a verdict. The grammar the framework is built on is taught here, as play.

A rip-and-run variant — skip the §3 casing, sprint the checkpoints, eat the Rep hit if caught — exercises a different subset; and NEONGRID’s Versus/Sysop linked-play (a human hunting you through the grid, neon-grid.md §10) is its own defender dynamic, distinct from knSALK. Same machine, more playstyles.