Skip to content

2026-04-25-mission-genre-templates

Mission Composition Grammar — Genre Template Library

Section titled “Mission Composition Grammar — Genre Template Library”

Status: Draft design for post-v0.1 wave (Notion: GWP-280) Date: 2026-04-25 Owner: Gameplay Design / PM (GWP-280, parent GWP-277) Scope: Ten hand-crafted genre templates — the narrative skeletons that the Mission Composition Grammar (MCG) composition algorithm picks from at step 1 (ROLL GENRE). Each genre defines a verb sequence, Cipher voice register bias, payout class, complication slot pool, sanctioned cross-affinity transition overrides, and a worked instantiation. These ten templates are the narrative DNA of every generated mission. Non-scope: No C, no Lisp runtime changes, no emulator source edits. The composition algorithm reference implementation lives in the engineering sub-task (Notion: GWP-277 → “Engineering: MCG composition algorithm”). This document defines the data the algorithm consumes at step 1.

Hardware spec references — this design targets the canonical 80×25 amber grid, 31-key input system, CIPHER-LINE auxiliary OLED, and Pi Zero 2 W firmware described in CLAUDE.md → “Canonical Hardware Specification”. Nothing in this doc restates those values.

Augments, does not supersede. This plan extends 2026-04-25-mission-composition-grammar.md (the MCG addendum: verb vocabulary §1, affinity tags §2, mission-contributions cart block §3, composition algorithm sketch §4). The verb vocabulary, affinity tag set, and transition table defined there are load-bearing inputs to every template below. Where a genre needs a transition the global table doesn’t sanction, it declares an :allowed-transitions override per the resolution to MCG addendum §7 Q3 — see §0.3 below.


A genre template is a hand-authored data record consumed by step 1 (ROLL GENRE) of the composition algorithm. Each template encodes:

  1. A pitch — the kind of story the genre tells, in one sentence.
  2. A verb sequence — an ordered (or partially ordered) skeleton drawn from the §1 verb vocabulary, with branch operators for optional/parallel/choice phases.
  3. A Cipher voice register bias — additive deltas to the mode selector (observe / annotate / reflect / drift / silent) the engine applies whenever this genre is the active mission. Per Q5 resolution.
  4. A payout class — small / medium / large / whale. Maps to the topology multiplier ladder in MCG §4 step 5.
  5. A complication slot pool — 3–5 named complications the algorithm may roll at step 4 (PAD WITH COMPLICATIONS).
  6. :allowed-transitions overrides — extra (FROM TO) cross-affinity pairs sanctioned for this genre beyond the §2 global table. Per Q3 resolution.
  7. A sample instantiation — one path through the verb sequence using carts from the 14-cart launch library and the 15 concept cartridges, demonstrating the genre can produce a real mission.

The composition algorithm’s step 1 weight roll is biased by reputation tier and recent-genre history (avoid repetition); the templates themselves are flat data, neither exclusive nor leveled. Every genre is reachable at every reputation tier, with payout class providing the natural rep-gating (whale-class payouts ride high threat, which the rep-tier clamp restricts at lower tiers).

Verb sequences use the following formal notation:

OperatorMeaningExample
Linear successionOBSERVE → OBTAIN (OBSERVE then OBTAIN)
?Optional phase (0 or 1 instance)OBSERVE → ANALYZE? → DELIVER (ANALYZE may be skipped)
``
|Choice (exactly one of the alternatives)DELIVER | DESTROY (operator commits to one branch at acceptance)
( ... )Grouping`PENETRATE → (OBTAIN

Parallel and choice operators apply per phase position, not across the whole skeleton. A skeleton like A → (B || C) → D produces a 4-phase chain when both B and C resolve; a A → (B | C) → D produces a 3-phase chain because the choice resolves to one branch.

The composition algorithm at step 2 (INSTANTIATE PHASES) walks the skeleton left-to-right, resolving operators as it encounters them. Optional phases roll a 50/50 (LFSR-seeded) at instantiation; parallel phases instantiate both sides; choice phases pick the branch with the higher cart-coverage in the operator’s loaded library, breaking ties with the LFSR.

0.3 :allowed-transitions overrides — per Q3 resolution

Section titled “0.3 :allowed-transitions overrides — per Q3 resolution”

MCG addendum §7 Q3 was resolved as: genre templates declare :allowed-transitions overrides when the narrative skeleton needs a cross-affinity transition the §2 global table does not sanction. Each template in §1–§10 below carries its own list. The composer at step 2b consults the global table first; if the prior phase’s affinity → next phase’s affinity is sanctioned globally, no override is needed. If not, the composer falls through to the active genre’s :allowed-transitions. If neither sanctions the transition, the chain is rejected and the composer rolls a different cart (per MCG §4 step 2c).

Each override carries a narrative justification — a short clause that becomes Cipher’s bridge fragment between phases. Overrides are not silent: they generate sticky :significant events in the Cipher event ring so the engine has source material when it elects to speak the bridge.

0.4 Cipher voice register bias — per Q5 resolution

Section titled “0.4 Cipher voice register bias — per Q5 resolution”

MCG addendum §7 Q5 was resolved as: yes, each genre declares Cipher mode bias. The bias is a list of (MODE, DELTA) pairs applied additively to the mode selector distribution alongside the per-beat defaults (docs/software/runtime/cipher-voice.md §6). Genre biases compose with cartridge biases and event affect biases in the standard normalization step; they are clamped at ±0.20 per mode to match the cart-bias clamp.

The bias persists for the duration of the mission and is unloaded at debrief, exactly like a per-cart bias. Multi-genre missions are not a v1 concept — each MissionInstance carries exactly one genre.

Genre biases are register declarations, not narration scripts. The genre says “this mission wants to feel reflective” by raising reflect’s weight; the actual fragments still come from the cart’s cipher-grammar block plus the runtime baseline. A REVENGE mission running ICE Breaker still uses ICE Breaker’s vocabulary — REVENGE just makes Cipher more likely to draw memory than to narrate the present.

MCG addendum §7 Q1 was resolved as: Cipher Garden and The Vault carry the :passive modifier and never anchor a phase. They contribute to a phase’s payout-bias and threat-bias when they share affinity with the cart that does anchor the phase, and they may surface in CIPHER-LINE memory recall (their events flow through the standard event ring). The genre templates below never propose Cipher Garden or The Vault as anchors; sample instantiations note when a passive cart contributes to a phase as a modifier.

Each genre declares one of four payout classes:

ClassTopology multiplier ceilingTypical phase countThreat distribution
small≤ 1.2× (MONO or short 2-CHAIN)1–2low–mid (T1–T3)
medium≤ 1.5× (3-CHAIN)2–3mid (T2–T4)
large≤ 2.0× (4-CHAIN)3–4mid–high (T3–T5)
whaleunbounded; carries a 1.3× whale-class multiplier on top of topology4–6 (with parallel phases counted individually)high (T4–T6); rep-tier-gated

Payout class is a budget, not a guarantee. A small-class genre that the LFSR rolls into a particularly long 2-CHAIN with a high cross-cart multiplier and threat boost can still pay handsomely; it just won’t enter whale territory. Whale-class genres, conversely, only pay whale-class numbers when the operator’s library is rich enough to fill the long skeleton with diverse carts.


Pitch: Get into a defended target, take the thing, get out before the alarm finishes ringing.

Verb sequence: OBSERVE → PENETRATE → OBTAIN → DELIVER

Cipher voice register bias: ((:observe +0.10) (:silent +0.05) (:reflect -0.05))

Heist is present-tense work. The voice tracks the now — what’s in front of the operator, what’s on the wire, what’s in the room. Silence rises on the approach because the cockpit-recorder aesthetic wants quiet during the reach; reflection drops because there’s no time to look back yet. The closer is annotate-class: “smooth.” or “loud.” after the exfil settles.

Payout class: medium

Complication slot pool:

  • :alarm-tripped-early — guard rotation arrives one phase sooner; threat +1 on the OBTAIN phase.
  • :inside-man-flips — the social contact (if a NEGOTIATE side-quest is in flight) doubles the asking price mid-mission; subtract 25% from final payout unless operator accepts and pays from pocket.
  • :exfil-window-halved — DELIVER phase time budget cut in half; failure converts the OBTAIN haul to a partial.
  • :wrong-vault — the OBTAIN cart’s primary target is a decoy; the real artifact is one cell deeper, requiring a re-roll of OBTAIN.
  • :silent-witness — an OBSERVE-only third party logs the operation; reputation +0 instead of +full unless operator returns later to clean the record (spawns a ghost EXPOSE follow-up; see §2).

:allowed-transitions overrides: none. HEIST runs cleanly on the global §2 table — every standard heist transition (OBSERVE → PENETRATE, PENETRATE → OBTAIN, OBTAIN → DELIVER) is either same-affinity or globally sanctioned (e.g., DIGITAL → DIGITAL for ICE Breaker self-chains, or PHYSICAL → PHYSICAL for Depthcharge → Pathfinder).

Sample instantiation (4-cart launch library: ICE Breaker + NeonGrid + Black Ledger + Depthcharge):

PhaseVerbCartDomain
1OBSERVEICE BreakerMap the corp’s perimeter — three reachable subnets, one with a SURVEILLANCE IMPLANT slot.
2PENETRATEICE BreakerBurn the outermost ICE; route into the financial subnet.
3OBTAINBlack LedgerPull the executive’s expense ledger from the open subnet (sanctioned DIGITAL → FINANCIAL transition per §2).
4DELIVERICE BreakerPush the dossier through a relay back to the operator’s deck before the audit cycle clears.

Pays ~¤9,500 at threat 3, 4 phases, 2 distinct carts (ICE×3 + BL×1, cross-cart multiplier ~1.1×). Cipher’s voice bias keeps the run tight: lots of observe mode on the penetration, a silent tick at the lift moment, an annotate smooth. after delivery clears.


Pitch: Find the buried truth, prove it, publish it. Whoever buried it has reasons to stop you.

Verb sequence: OBSERVE → ANALYZE → COMMUNICATE

Cipher voice register bias: ((:annotate +0.10) (:reflect +0.05) (:drift -0.05))

Expose is editorial work. The voice labels what it finds — that one. anomalous., costly., the good kind of quiet. annotate rises because the genre wants Cipher to call out anomalies as they surface. reflect rises modestly because exposure missions chain context across phases (this ledger entry. same shell as the relay run last month.). drift drops because exposes are present-tense investigations, not idle voyages.

Payout class: medium

Complication slot pool:

  • :source-recants — a NEGOTIATE side-objective is required to keep the source in play; failure halves the publication payout.
  • :rival-publication — another operator (or AI rival, design-deferred) is racing to publish first; time pressure compresses the COMMUNICATE phase by 30%.
  • :libel-shield — the buried truth is wrapped in legal countermeasures; ANALYZE phase requires a second pass with an additional cart from the library.
  • :dead-link — one of the OBSERVE evidence chains has decayed since collection; threat +1 on the ANALYZE phase as the operator must rebuild the inference.
  • :audience-fracture — the COMMUNICATE phase splits into two parallel pushes (mainstream + underground); operator picks one, forfeits the other’s payout.

:allowed-transitions overrides:

  • (INFORMATION → SOCIAL) — sanctioned globally, but EXPOSE specifically biases toward this pairing for the OBSERVE → COMMUNICATE chain.
  • (FINANCIAL → SOCIAL) — sanctioned globally; reaffirmed for forensic-driven exposes.
  • (IDENTITY → INFORMATION)not in §2 global table. Sanctioned per genre: a forged identity surfaces in an archive query that closes the case (CLONESHOP + The Vault). Cipher bridge: “that face. seen it in the records.”

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBSERVEBlack LedgerAudit the corp’s quarterly filings; surface a payment to a dead shell.
2ANALYZEBlack LedgerReconstruct the shell network; identify the beneficial owner.
3COMMUNICATEICE BreakerPush the proof to a public BBS via a relay handshake — the only COMMUNICATE-satisfier in the launch library, sanctioned via global FINANCIAL → DIGITAL.

3 phases, 2 distinct carts, threat 2–4 depending on shell depth. The Cipher voice register makes Black Ledger’s annotate mode dominate the audit — that one felt wrong., too clean., costly. — and lets ICE Breaker close with pushed. clean. once the broadcast lands.


Pitch: Find the target, get a clean shot, end the threat. Sabotage as a clean cut.

Verb sequence: OBSERVE → PENETRATE → DESTROY

Cipher voice register bias: ((:observe +0.10) (:silent +0.10) (:annotate -0.05))

Eliminate is the most cockpit-recorder of the genres. The voice watches, the voice waits, the voice drops to silence on the strike. Annotate drops because the engine should not editorialize the kill — let the silence carry it. silent rises sharply, both at the approach and at the moment of the destroy event itself.

Payout class: small

Complication slot pool:

  • :collateral — DESTROY phase risks a secondary asset; threat +2 unless operator accepts the collateral and takes a reputation hit.
  • :second-target — intelligence reveals a paired target that becomes a parallel DESTROY; payout doubles, threat +1 on both.
  • :witness-cleanup — a third party observed the strike; a ghost EXPOSE follow-up is appended (see §2) with reduced payout.
  • :misidentified-target — the OBSERVE chain pointed at a decoy; rolls into a re-OBSERVE phase, threat -1 because the operator now has the corp’s countermeasures mapped.
  • :sanction-reversed — between PENETRATE and DESTROY, the contract issuer rescinds; operator chooses to abort (full payout, no rep) or proceed (full payout, rep -3 for unsanctioned action).

:allowed-transitions overrides:

  • (NETWORK → MARITIME)not in §2 global table. Sanctioned per genre: ICE Breaker maps a relay station that Depthcharge then physically destroys (network intelligence enables a maritime strike). Cipher bridge: “the relay. came up at three-fifty.”

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBSERVEICE BreakerLocate the underwater relay station’s network signature; map its ICE.
2PENETRATEDepthchargeApproach the relay submersibly under acoustic cover (sanctioned PHYSICAL → MARITIME globally; here we use the genre override NETWORK → MARITIME for the seam).
3DESTROYDepthchargePhysical sabotage — torpedo the relay or plant a charge.

3 phases, 2 distinct carts, threat 3–5. Cipher goes mostly silent on the descent, lets Depthcharge’s :active-hack mode bias dominate, and emits a single annotate line at the strike: quiet now.


Pitch: Build power across multiple territories until you control something nobody can take from you. The long campaign genre.

Verb sequence: (OBTAIN || NEGOTIATE) → MANUFACTURE → DECIDE → DELIVER

Cipher voice register bias: ((:reflect +0.10) (:annotate +0.05) (:silent +0.05) (:drift -0.05))

Empire is the genre that thinks across phases. The voice connects beats — same shell. third time., here again. last time it cost us. reflect rises because the genre rewards Cipher remembering the operator’s prior decisions; drift drops because empire is not idle, it’s deliberate. Modest silent rise to give weight to the DECIDE phase.

Payout class: whale

Complication slot pool:

  • :rival-claim — a competing empire-builder targets the same territory; DECIDE phase gains a parallel OBSERVE sub-phase to monitor the rival’s move.
  • :foundation-cracks — the MANUFACTURE phase output (a forged identity, a chemical compound, a network position) is unstable; DELIVER risks corruption unless operator pays for stabilization out of pocket.
  • :unwilling-vassal — a NEGOTIATE phase fails downstream; the held territory rebels mid-DELIVER and threat spikes +2 on the closing phase.
  • :scrutiny — a passive Black Ledger or Cipher Garden in the loaded library raises the corp’s audit attention; threat +1 across all phases but payout +20% for the additional risk.
  • :dynasty — on success, the genre seeds a follow-up EMPIRE contract on the next mission board refresh with elevated payout (compounding gameplay).

:allowed-transitions overrides:

  • (IDENTITY → FINANCIAL)not in §2 global table. Sanctioned per genre: a forged identity (CLONESHOP) operates as a financial actor (Black Ledger). Cipher bridge: “the persona. signed the wire.”
  • (SOCIAL → NETWORK)not in §2 global table. Sanctioned per genre: a NEGOTIATE outcome restructures a social network or org chart. Cipher bridge: “the new chair. mapped now.”
  • (CHEMICAL → FINANCIAL)not in §2 global table. Sanctioned per genre: a manufactured compound (SYNAPTIC concept) commands a market price. Cipher bridge: “ten grams. cleared.”

Sample instantiation (4-cart launch library — degraded fit, see §11 sanity check):

EMPIRE struggles on the 4-cart library because MANUFACTURE has no satisfier among ICE Breaker / NeonGrid / Black Ledger / Depthcharge. The composer falls back to MONO (a single-cart EMPIRE-flavored Black Ledger run: layered audit + decision + delivery, played as a long forensic case). Full EMPIRE requires CLONESHOP, SYNAPTIC, WETWARE, or a Relay overlay carrying a MANUFACTURE-satisfying template.

Sample instantiation (concept-augmented library — 4 launch + CLONESHOP + HOSTILE TAKEOVER):

PhaseVerbCartDomain
1aOBTAIN
1bNEGOTIATE
2MANUFACTURECLONESHOP (concept)Forge an identity sufficient to occupy a board seat.
3DECIDEHOSTILE TAKEOVER (concept)Commit to the proxy fight at the annual meeting.
4DELIVERICE BreakerPush the new corporate filing through the SEC bridge.

5 phases counted (parallel pair counts as two), 4 distinct carts, threat 4–6, whale-class payout in the ¤30k–¤45k range depending on rep tier.


Pitch: Move through the world without leaving a trace. The stealth-purity genre.

Verb sequence: OBSERVE → (PENETRATE | DELIVER) → OBSERVE

Cipher voice register bias: ((:silent +0.15) (:drift +0.05) (:observe -0.05))

Ghost runs the most silent register in the library. The voice almost vanishes. silent rises sharply because the cockpit recorder is choosing to honor the operator’s stealth — anything Cipher says risks giving the operator away (in-fiction the device is silent, not the radio). drift rises because ghost runs leave Cipher meditating between events. observe drops because there is often nothing to observe out loud.

Payout class: small

Complication slot pool:

  • :near-miss — at one phase, a passive observer almost catches the operator; threat +1, but a successful stealth-recovery raises the payout 25%.
  • :residual-trace — the OBSERVE-out phase reveals a small footprint left by the operation; closing the trace requires a ghost EXPOSE follow-up (see §2).
  • :overheard — a NEGOTIATE side-channel was needed and produced witnesses; Cipher voice register temporarily un-suppresses to let annotate flag the slip.
  • :double-blind — the OBSERVE-out phase reveals the operator was also being observed; rolls a parallel reverse-OBSERVE sub-phase (counter-surveillance).
  • :perfect-run — no complications fire; payout +30% bonus and a sticky :anomalous event in the Cipher event ring (Cipher will drift to this run for many missions afterward).

:allowed-transitions overrides:

  • (MARITIME → INFORMATION)not in §2 global table. Sanctioned per genre: a Depthcharge run yields surveillance footage that flows to The Vault as passive contribution. Cipher bridge: “what we heard. archived.”
  • (AERIAL → INFORMATION)not in §2 global table. Same logic for NIGHTOWL → archive.

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBSERVEDepthchargeSonar-sweep the harbor approach. Acoustic-signature minimization mandatory.
2PENETRATEDepthchargeSubmerge to the relay station, dock without active sensors.
3OBSERVEDepthchargeWithdraw, sweeping for any acoustic signatures the operation left behind.

3 phases, 1 cart (MONO topology), threat 2–4. Cipher’s voice register bias compounds with Depthcharge’s existing :high-tense → :silent +0.10 cart bias to produce one of the quietest mission types in the library. Most ticks are silent; the few that speak come from drift mode pulling memory of the harbor approach.


Pitch: The operator is owed something. They take it back, and they make sure the debtor knows who took it. The personal genre.

Verb sequence: OBSERVE → MANUFACTURE → DELIVER → ANALYZE

Cipher voice register bias: ((:reflect +0.15) (:drift +0.10) (:annotate +0.05) (:observe -0.10))

Revenge is the most past-haunted register. reflect and drift both rise sharply — the genre is about memory. The voice keeps connecting present beats to past wounds: same office. last time we walked out broke. observe drops because revenge is not about narrating the present, it’s about narrating the past breaking through into the present. annotate rises modestly to give the closer a cold, editorial quality at debrief.

Payout class: large

Complication slot pool:

  • :debtor-flees — between MANUFACTURE and DELIVER, the target relocates; DELIVER phase requires an additional OBSERVE sub-phase to re-acquire.
  • :collateral-target — the manufactured payload (forged dossier, planted compound, digital exploit) affects bystanders; reputation hit unless the operator accepts and proceeds.
  • :debt-disputed — at ANALYZE phase, a paper trail surfaces suggesting the operator’s grievance was misplaced; Cipher emits a :significant-tagged drift in the closing beats either way.
  • :reciprocity — successful revenge generates an enemy who will appear as a complication slot in future missions (cross-mission state).
  • :catharsis — operator chooses to forgo payout and accept reputation +5 instead (DECIDE-class branching).

:allowed-transitions overrides:

  • (IDENTITY → AERIAL)not in §2 global table. Sanctioned per genre: a forged identity calls down a drone delivery (CLONESHOP + NIGHTOWL). Cipher bridge: “the persona. then the drop.”
  • (AERIAL → FINANCIAL)not in §2 global table. Sanctioned per genre: an aerial drop has financial fallout (drone-delivered dossier triggers a panic-sale; the worked example in MCG addendum §5). Cipher bridge: “the drop. then the dump.”

Sample instantiation (4-cart launch library — degraded fit):

REVENGE has no MANUFACTURE-satisfier in the launch four. The composer falls back to a 3-phase EXPOSE-shaped run: OBSERVE (ICE Breaker) → ANALYZE (Black Ledger) → COMMUNICATE (ICE Breaker), played as a public-shaming revenge rather than a manufactured-payload revenge. Full REVENGE requires a MANUFACTURE-class concept cart (CLONESHOP, SYNAPTIC, WETWARE).

Sample instantiation (concept-augmented library — 4 launch + CLONESHOP + NIGHTOWL):

This is the worked example from MCG addendum §5. See that document for the full four-phase REVENGE chain: ICE Breaker (OBSERVE) → CLONESHOP (MANUFACTURE) → NIGHTOWL (DELIVER) → Black Ledger (ANALYZE), payout ¤17,472 at threat 4. Reproduced here only by reference; the worked example in MCG §5 was authored against this template.


Pitch: Something is out there that nobody else has found. The operator finds it first, brings back proof, decides what to do with it.

Verb sequence: OBSERVE → ANALYZE → DECIDE

Cipher voice register bias: ((:reflect +0.10) (:drift +0.10) (:silent +0.05) (:observe +0.05))

Discovery is the contemplative register. The voice shifts toward memory and ambient atmosphere — the operator is somewhere new and Cipher feels its way into it. All four contemplative modes (observe, reflect, drift, silent) get small positive deltas; annotate is the only mode left flat because discovery is rarely editorial — it’s experiential. The DECIDE phase produces a sticky :significant event for the chosen branch.

Payout class: medium

Complication slot pool:

  • :dead-end — the OBSERVE phase yields nothing the ANALYZE phase can use; the run downgrades to a small payout but seeds future DISCOVERY missions on the same vector.
  • :not-alone — another operator (or AI rival) has been to the location first; ANALYZE phase finds traces; DECIDE phase becomes “report or claim.”
  • :dangerous-find — the discovery is hazardous (forbidden archive, biohazard, weaponizable code); DECIDE phase carries genuine moral weight reflected in reputation outcome.
  • :partial-revelation — only a fragment of the discovery is recoverable; ANALYZE phase splits into a parallel re-OBSERVE.
  • :luminous — a perfect-run modifier; the discovery rewrites a small piece of the deck’s domain knowledge (raises a knowledge index by 3 instead of 1).

:allowed-transitions overrides:

  • (MARITIME → INFORMATION)not in §2 global table. Sanctioned per genre: a Depthcharge sweep recovers an artifact that goes to The Vault. Cipher bridge: “what was down there. logged.”
  • (PHYSICAL → INFORMATION) — sanctioned globally; reaffirmed for archaeological-flavored discoveries.

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBSERVEDepthchargeSonar-sweep an uncharted seamount; locate a sunken vessel.
2ANALYZEBlack LedgerCross-reference the vessel’s manifest (recovered as digital data) against historical insurance records — sanctioned MARITIME → FINANCIAL via PHYSICAL bridge per §2. The Vault contributes passively (:passive modifier) raising payout-bias 1.1× because INFORMATION shares affinity with the analysis.
3DECIDEBlack LedgerReport the find to the insurance underwriter (collect bounty), or sit on it (no payout, but the discovery enters a private archive that may surface in later EMPIRE or REVENGE missions).

3 phases, 2 distinct carts (Depthcharge + Black Ledger, Vault passive), threat 2–4. Cipher mode bias produces a contemplative tone — many drift ticks during the analysis, a long silent stretch on the deciding beat.


Pitch: Something has spread. Track it back to its source, then decide whether to contain or weaponize. The investigative-with-stakes genre.

Verb sequence: OBSERVE → OBSERVE → ANALYZE → (DESTROY | MANUFACTURE)

Cipher voice register bias: ((:annotate +0.10) (:reflect +0.10) (:silent +0.05) (:drift -0.10))

Contagion is forensic with weight. The voice labels what it finds (that one. anomalous.) and connects nodes (same signature as the relay run.). drift is suppressed sharply because contagion is active — the voice cannot afford to wander, the spread is happening. The closing branch produces very different debriefs: DESTROY closes with quiet now.; MANUFACTURE closes with held. ours.

Payout class: large

Complication slot pool:

  • :patient-zero-cold — the second OBSERVE phase fails to identify the source; threat +1 on ANALYZE as the operator must reason backward from spread pattern alone.
  • :active-spread — between phases, the contagion advances; downstream phases gain time pressure (-25% time budget per phase remaining).
  • :counter-investigator — another party is also tracing the contagion; OBSERVE phases gain a parallel counter-OBSERVE (cat-and-mouse).
  • :ethical-fork — the DECIDE between DESTROY and MANUFACTURE carries reputation weight (DESTROY = +rep, MANUFACTURE = +¤ +rep_dark).
  • :resistant-strain — DESTROY phase requires two passes; MANUFACTURE phase requires an additional supporting cart.

:allowed-transitions overrides:

  • (CHEMICAL → DIGITAL)not in §2 global table. Sanctioned per genre: a chemical compound’s distribution is mapped via digital trail. Cipher bridge: “the molecule. then the wire.”
  • (BODY → INFORMATION) — sanctioned only via :bridge cart in §2; per genre, sanctioned without a bridge cart when the genre is CONTAGION (the disease is information). Cipher bridge: “the body. the record.”
  • (DIGITAL → BODY) — sanctioned only via :bridge cart in §2; per genre, sanctioned without a bridge cart when the genre is CONTAGION (digital-borne pathogen, e.g., a corrupted neural firmware). Cipher bridge: “from the wire. into them.”

Sample instantiation (4-cart launch library — moderate fit):

PhaseVerbCartDomain
1OBSERVEICE BreakerSurveil three nodes showing identical anomalous traffic signatures.
2OBSERVEBlack LedgerCross-correlate financial flows from the same nodes; identify the funding source.
3ANALYZEBlack LedgerReconstruct the spread pattern; identify patient zero.
4DESTROYICE BreakerSabotage the source node (the financial-laundering node that funded the digital contagion).

4 phases, 2 distinct carts, threat 3–5. The MANUFACTURE branch is unavailable on the launch library and the composer constrains the choice to DESTROY automatically. Full CONTAGION with the MANUFACTURE branch requires SYNAPTIC, WETWARE, or DAEMON.


Pitch: Get the package from origin to destination. Conditions are adverse, route is contested, the package matters more than the operator does.

Verb sequence: OBTAIN → DELIVER → DELIVER?

Cipher voice register bias: ((:observe +0.10) (:annotate +0.05) (:silent +0.05) (:drift -0.05))

Courier is the journey register. The voice tracks the route — what’s behind, what’s ahead, what just changed. observe rises because the operator wants real-time awareness. annotate rises modestly for the seam moments (smooth., loud., lost the tail.). drift drops because the courier doesn’t have time to wander mentally; the package is in motion. The optional second DELIVER (a hand-off to a final recipient) produces a :significant event regardless of mode.

Payout class: small

Complication slot pool:

  • :tail — adversary detected en route; threat +1 on the DELIVER phase, parallel OBSERVE sub-phase to evade.
  • :cargo-degrades — the OBTAIN-phase package has a half-life; DELIVER’s time budget halves and missing it converts payout to partial.
  • :second-package — at one phase, an additional cargo is offered for parallel delivery; payout +40% but threat +2.
  • :checkpoint — a NEGOTIATE side-objective is required to clear an inspection; failure triggers DESTROY of the cargo (mission failed but operator escapes).
  • :rerouted — the DELIVER destination changes mid-mission; the optional second DELIVER becomes mandatory and triggers a route re-roll.

:allowed-transitions overrides:

  • (MARITIME → AERIAL)not in §2 global table. Sanctioned per genre: cargo transfers from submersible to drone for a hand-off. Cipher bridge: “the wet drop. up the air.”
  • (AERIAL → SOCIAL)not in §2 global table. Sanctioned per genre: a drone delivery is the prelude to a face-to-face hand-off. Cipher bridge: “the drone. then the door.”

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBTAINDepthchargeRecover an encrypted package from a sunken dead-drop cache.
2DELIVERDepthchargeSurface and run the harbor; deliver to the dockside contact.
3DELIVERICE Breaker(Optional second DELIVER, ~50% roll.) Push the package’s contents up an encrypted relay to the final recipient — sanctioned PHYSICAL → DIGITAL globally.

2–3 phases (depending on the optional roll), 1–2 distinct carts, threat 2–4. Cipher’s observe bias keeps the route narrative present and clipped: bearing two-seventy., dock clear., package up.


Pitch: The operator is hired to verify, not to act. Find what’s there, document it, leave. The least violent genre.

Verb sequence: OBSERVE → ANALYZE → COMMUNICATE

Cipher voice register bias: ((:annotate +0.15) (:observe +0.05) (:silent +0.05) (:reflect -0.05))

Audit is editorial-clinical. The voice labels methodically — that one. anomalous., clean., costly., complete. annotate rises sharply because the genre’s purpose is naming what is found. reflect drops because audit work is bounded — past missions don’t apply, only the current dataset does. The closing COMMUNICATE phase produces a clean, reportorial close: delivered. on the wire.

Payout class: small

Complication slot pool:

  • :scope-creep — during ANALYZE, the operator surfaces a finding outside the contracted scope; client offers payment to extend (parallel sub-phase) or to bury (Cipher event ring keeps the suppressed finding sticky for future missions).
  • :counter-audit — the audited party is auditing the auditor; OBSERVE phase gains a parallel counter-OBSERVE.
  • :findings-disputed — COMMUNICATE phase requires a NEGOTIATE side-objective to deliver findings over client objection; failure halves payout.
  • :material-weakness — the audit reveals a weakness the operator can later exploit (seeds a HEIST/EXPOSE/REVENGE follow-up; cross-mission state).
  • :clean-bill — perfect run; no findings; a small payout but +5 reputation and a sticky :routine event ring entry that biases future Cipher voice toward calm registers.

:allowed-transitions overrides: none. AUDIT runs cleanly on the global §2 table — every standard audit transition is sanctioned (DIGITAL → INFORMATION, FINANCIAL → INFORMATION, INFORMATION → SOCIAL).

Sample instantiation (4-cart launch library):

PhaseVerbCartDomain
1OBSERVEBlack LedgerPull the contracted firm’s quarterly statements and treasury logs.
2ANALYZEBlack LedgerReconcile against the source ledger; flag anomalies; identify material weaknesses. The Vault contributes passively (:passive) — INFORMATION affinity raises payout-bias 1.1× because the audit benefits from cross-archival lookup.
3COMMUNICATEICE BreakerPush the audit report to the client over an authenticated channel — sanctioned FINANCIAL → DIGITAL globally.

3 phases, 2 distinct carts (Black Ledger + ICE Breaker, Vault passive), threat 1–3. The smallest payout class in the library, but the cleanest reputational outcome — AUDIT runs build the operator’s credit as a trustworthy professional, which gates other genres at higher rep tiers.


§11. Sanity check — instantiation against the 4-cart launch library

Section titled “§11. Sanity check — instantiation against the 4-cart launch library”

Per the GWP-280 brief: each genre must be tested against the 4-cart launch library (ICE Breaker + NeonGrid + Black Ledger + Depthcharge). When a genre cannot fully instantiate, the composer falls back to MONO (a single-cart genre-flavored run) per MCG §4 step 5; the genre is not orphaned.

GenreFits 4-cart library?Notes
HEISTFull 4-phase chain instantiates with ICE Breaker + Black Ledger as shown in §1 sample.
EXPOSEFull 3-phase chain with Black Ledger + ICE Breaker as shown in §2 sample.
ELIMINATEFull 3-phase chain with ICE Breaker + Depthcharge as shown in §3 sample.
EMPIRE⚠️ degradedMANUFACTURE has no satisfier in the launch four. Falls back to MONO (long Black Ledger forensic-empire run) until a MANUFACTURE-class cart is added. Full EMPIRE needs CLONESHOP, SYNAPTIC, WETWARE, or HOSTILE TAKEOVER.
GHOSTMONO Depthcharge instantiation as shown in §5 sample. Single-cart genre by design at any library size.
REVENGE⚠️ degradedMANUFACTURE has no satisfier; falls back to a 3-phase EXPOSE-shaped public-shaming revenge. Full REVENGE (with a manufactured payload) needs CLONESHOP, SYNAPTIC, or WETWARE. The MCG addendum §5 worked example uses CLONESHOP + NIGHTOWL; both are concept carts.
DISCOVERYFull 3-phase chain with Depthcharge + Black Ledger as shown in §7 sample.
CONTAGION⚠️ partialDESTROY branch fits cleanly (4-phase chain shown in §8 sample). MANUFACTURE branch unavailable; composer constrains the choice. Full CONTAGION with both branches needs SYNAPTIC, WETWARE, or DAEMON for the MANUFACTURE side.
COURIER2–3 phase chain with Depthcharge + ICE Breaker as shown in §9 sample.
AUDITFull 3-phase chain with Black Ledger + ICE Breaker as shown in §10 sample.

Summary: 7 of 10 genres fit the launch library cleanly. The three degraded fits (EMPIRE, REVENGE, CONTAGION) all share a single root cause: the MANUFACTURE verb has no satisfier in ICE Breaker / NeonGrid / Black Ledger / Depthcharge. Adding any single MANUFACTURE-class cart (CLONESHOP, SYNAPTIC, WETWARE, HOSTILE TAKEOVER, DAEMON) restores full coverage for two of the three; adding CLONESHOP specifically restores all three (CLONESHOP satisfies EMPIRE’s identity-build, REVENGE’s persona-forge, and CONTAGION’s identity-vector spread).

The composer’s MONO fallback (MCG §4 step 5) ensures no genre is orphaned — the worst case is a genre rolls and instantiates as a single-cart short run instead of its preferred multi-cart skeleton. The Cipher voice register bias still applies; the genre still feels like itself even when degraded.

This is precisely the design intent: the launch library is complete (every genre is reachable), and concept-cart additions expand the surface (every additional cart unlocks longer skeletons, more parallel phases, more cross-affinity richness).


Q1 — Genre repetition window. MCG §4 step 1 says “weighted by reputation tier and recent-genre history (avoid repetition).” How long is the repetition window — last 3 contracts, last 6, decay over time? Recommend last 3 (per genre, per session) with a soft penalty (-50% weight) rather than a hard exclusion. Resolve in the engineering sub-task (GWP-277 sub-task 4).

Q2 — Whale-class rep gating. EMPIRE is the only whale-class genre in this v1 library. Should whale-class genres carry a hard rep floor (e.g., rep ≥ 25), or rely on threat clamps to naturally gate them? Recommend the latter — natural gating preserves the design principle that every genre is reachable at every rep tier (per §0.1). EMPIRE just won’t pay whale-class numbers at rep 5 because the threat clamp keeps the chain short.

Q3 — Genre count expansion. This v1 library is 10 genres. The composition algorithm’s weighting scheme (MCG §4 step 1) needs a strategy for adding genres without disrupting existing genre frequencies. Recommend declaring genre weights explicitly per cart (every cart’s mission-contributions block could carry an optional :genre-weights field that biases certain genres up or down when this cart is the active anchor). Track as future MCG addendum revision.

Q4 — Complication composition. Some complications (:silent-witness from HEIST seeds an EXPOSE follow-up; :material-weakness from AUDIT seeds a HEIST follow-up) imply cross-mission state. Where does this state live — in the MissionInstance, in Universal Deck State, in a new “follow-up board” subsystem? Recommend adding a small pending-followups slot (4–8 entries, ring buffer) to the MissionBoardGen state for v1; promote to deck-state if the feature proves load-bearing in playtesting.

Q5 — Genre-specific topology. Some genres prefer specific topologies (GHOST → MONO, EMPIRE → BRANCH/PARALLEL, COURIER → CHAIN). Should genre templates declare a :preferred-topology field that the composer biases toward at step 5 (COMPUTE PAYOUT)? Recommend yes; defer to engineering sub-task (GWP-277 sub-task 4) along with the topology shape grammar (MONO/CHAIN/BRANCH/PARALLEL/EPISODIC/NESTED/ESCALATION/ECHO).


  • HEIST, EXPOSE, ELIMINATE, EMPIRE, GHOST, REVENGE, DISCOVERY, CONTAGION, COURIER, AUDIT each fully specified (§1–§10)
  • Each genre declares: pitch, verb sequence (formal notation), Cipher voice register bias, payout class, complication slot pool, :allowed-transitions overrides, sample instantiation
  • Per Q1 resolution: passive carts (Cipher Garden, The Vault) never proposed as anchors; surface only as modifiers in sample instantiations
  • Per Q3 resolution: each genre’s :allowed-transitions overrides are explicit and narratively justified
  • Per Q5 resolution: each genre declares Cipher mode bias as additive deltas to the mode selector
  • Each genre tested against the 4-cart launch library; degraded fits flagged (§11)
  • Cross-references added to MCG addendum (§6 sweep + §4 step 1 forward pointer)
  • Open questions captured (§12)

  • 2026-04-25-mission-composition-grammar.md — the MCG addendum this plan extends. Defines the verb vocabulary (§1), affinity tag set + cross-affinity transition table (§2), mission-contributions cart contribution block (§3), composition algorithm sketch (§4), and worked REVENGE example (§5) that this plan’s REVENGE entry references.
  • 2026-04-21-mission-board.md — the Mission Board generator design. The genre templates feed step 1 of the composition algorithm that produces MissionInstance records for the board.
  • docs/software/runtime/cipher-voice.md — Cipher voice engine. Defines the five utterance modes (observe, annotate, reflect, drift, silent) and the mode selector that this plan’s per-genre bias deltas feed into.
  • docs/software/runtime/orchestration.md — capability model spec; the cross-program integration section that the MCG addendum generalizes and that this plan’s sample instantiations honor.
  • docs/software/cartridges/design-bibles/launch-titles-capability.md — launch-titles design bible. The 4-cart launch library (ICE Breaker + NeonGrid + Black Ledger + Depthcharge) sample instantiations target.
  • Notion parent task: GWP-277 — Design and implement Mission Composition Grammar for procedural multi-phase missions.
  • Notion sub-task: GWP-280 — Spec: MCG genre template library (10 narrative skeletons). This document.
  • Notion concept library: GWP-283 — 15 candidate cyberpunk cartridges for KN-86 expansion. The concept carts (CLONESHOP, NIGHTOWL, WETWARE, SYNAPTIC, HOSTILE TAKEOVER, DAEMON, etc.) referenced in degraded-fit notes and concept-augmented sample instantiations.