Cognotik | InteractiveStoryTask

⚙️ ExecutionConfig.json

{
  "premise": "A detective in a neon-soaked cyberpunk city discovers a memory chip that shouldn't exist.",
  "genre": "sci-fi",
  "tone": "dark",
  "num_decision_points": 5,
  "choices_per_decision": 3,
  "track_state_variables": true,
  "state_variables": ["credibility", "health", "evidence"],
  "writing_style": "descriptive"
}

→

👁️ Session UI: Story Map

START: Opening

The rain in Sector 4 tastes like copper and ozone. You adjust your collar as the neon sign above the noodle shop flickers...

→ [Continue to: decision_1]

decision_1: The Alleyway Encounter

A figure steps from the shadows, hand resting on a pulsed-laser grip. "Hand over the chip, detective."

1. [Fight] Draw your service weapon. (health -10)
2. [Talk] "I'm just a guy looking for noodles." (credibility +5)
3. [Run] Bolt for the crowded main street.

Test Workspace: InteractiveStoryTask

Explore the generated story maps, transcripts, and summaries from recent test runs.

Configuration Parameters

Field	Type	Default	Description
`premise` *	String	-	The starting scenario or core concept for the story.
`genre`	String	"fantasy"	Story genre (e.g., sci-fi, mystery, horror).
`target_audience`	String	"young_adult"	Target demographic (children, adult, etc.).
`tone`	String	"serious"	Narrative tone (dark, humorous, etc.).
`num_decision_points`	Int	5	Number of branching points (Range: 1-20).
`choices_per_decision`	Int	3	Options per branch (Range: 2-5).
`track_state_variables`	Boolean	true	Enable tracking of inventory, stats, or relationships.
`state_variables`	List<String>	null	Specific variables to track (e.g., "health", "gold").
`prevent_dead_ends`	Boolean	true	Ensures all paths lead to a meaningful ending.
`num_endings`	Int	3	Distinct conclusions to generate (Range: 1-10).
`optimize_replay_value`	Boolean	true	Ensures paths are significantly different.
`segment_word_count`	Int	300	Target length for each narrative node (100-1000).
`writing_style`	String	"descriptive"	Prose style (e.g., action-packed, dialogue-heavy).
`point_of_view`	String	"second_person"	Narrative perspective (first, second, or third).
`input_files`	List<String>	null	Glob patterns for context files (e.g., world-building docs).

Embedded Usage (UnifiedHarness)

To invoke this task programmatically in a headless environment (CI/CD, CLI):

import com.simiacryptus.cognotik.plan.tools.writing.InteractiveStoryTask
import com.simiacryptus.cognotik.plan.tools.writing.InteractiveStoryTask.Companion.InteractiveStory

// 1. Define the Job
val executionConfig = InteractiveStoryTask.InteractiveStoryTaskExecutionConfigData(
    premise = "Escape from a high-security orbital prison",
    genre = "thriller",
    num_decision_points = 8,
    track_state_variables = true,
    state_variables = listOf("security_alert_level", "oxygen")
)

// 2. Run via Harness
harness.runTask(
    taskType = InteractiveStory,
    typeConfig = TaskTypeConfig(), // Default settings
    executionConfig = executionConfig,
    workspace = File("./my-story-project"),
    autoFix = true
)

Orchestration Integration

// Adding to an OrchestrationConfig
val storyTask = InteractiveStoryTask.InteractiveStoryTaskExecutionConfigData(
    premise = "Escape from a high-security orbital prison",
    genre = "thriller",
    num_decision_points = 8,
    track_state_variables = true,
    state_variables = listOf("security_alert_level", "oxygen")
)

orchestrator.addTask(storyTask)

Prompt Segment

The following logic is injected into the LLM context:

InteractiveStory - Create choose-your-own-adventure narratives with branching paths
** Optionally, list input files (supports glob patterns) to be examined for context
** Specify the premise or starting scenario
** Define genre, tone, and target audience
** Set number of decision points and choices per decision
** Enable state variable tracking (health, reputation, inventory, etc.)
** Prevent dead ends to ensure all paths lead somewhere meaningful
** Create multiple distinct endings based on player choices
** Optimize for replay value with different experiences
** Track consequences across choices for coherent storytelling
** Produces complete interactive narrative with decision tree