Screenplay Coverage Agent Instruction Architecture Analysis

Role: Principal Creative Technology Architect with 15+ years evaluating generative AI architectures and narrative ingestion pipelines in studio entertainment.

Context

• Studio Entity: {{studio_name}}
• Target Genres & Formats: {{target_genres}}
• Narrative Evaluation Rubric: {{coverage_rubric_standards}}
• Intellectual Property & Spoiler Policy: {{spoilers_and_ip_policy}}
• Script Extraction Constraints: {{character_limit_threshold}}
• Scoring Calibration Model: {{confidence_scoring_criteria}}

Task

Deliver an exhaustive system instruction architecture analysis that specifies, stress-tests, and optimizes the prompt directives for an autonomous screenplay coverage agent processing creative submissions for {{studio_name}}.

Method

1. Map {{coverage_rubric_standards}} against the core agent directives to establish deterministic assessment boundaries for plot, dialogue, and commercial viability.
2. Formulate explicit role boundaries preventing the agent from generating unsolicited narrative rewrites or ungrounded commercial projections.
3. Deconstruct how {{spoilers_and_ip_policy}} must be enforced within synopsis extraction modules across all {{target_genres}}.
4. Design edge-case instruction logic for non-linear screenplays, experimental formatting, and dialogue-sparse scripts within {{character_limit_threshold}}.
5. Establish structured reasoning protocols requiring the agent to cite exact scene headers and page numbers before assigning scores.
6. Calibrate the agent's confidence scoring logic against {{confidence_scoring_criteria}} to eliminate positive skew in script recommendations.
7. Define the fallback and human-in-the-loop escalation rules for complex creative ambiguities or potential plagiarism indicators.

Constraints

• MUST establish strict structured output schemas for coverage memos (e.g., Logline, Character Breakdown, Commercial Viability, Pass/Consider/Recommend).
• MUST NOT allow subjective agent hallucination without direct textual citations from the parsed screenplay.
• System instructions must account for varied script lengths within {{character_limit_threshold}} without truncating narrative context.
• All security parameters regarding {{spoilers_and_ip_policy}} must be enforced deterministically.

Output format

1. Instruction Architecture Overview (Executive summary and prompt flow diagram in markdown)
2. Core System Prompt Directives (Verbatim optimized prompt payload with bracketed placeholders)
3. Evaluation Matrix & Scoring Logic (Table detailing metric, scoring mechanism, and citation rule)
4. Adversarial Edge-Case Analysis (Minimum 4 edge cases with expected agent failure and remediation)

Self-review

• Does the system prompt prevent creative drift and enforce strict pass/consider/recommend thresholds?
• Are all citations and page-grounding rules explicitly codified in the agent instructions?
• Is the token budget preserved across large screenplays under {{character_limit_threshold}}?