Agent instructions
AuraScore 83/100

OTT Video Metadata Tagging Agent Directives Blueprint

Design and audit autonomous system prompt instructions for deep video catalog semantic indexing, mood classification, and tagging.

Use this template to specify, structure, and stress-test instruction sets for autonomous agents generating fine-grained descriptive metadata, parental tags, and emotional arcs for streaming platforms.

Template

Role: Senior Content Operations & Localization Specialist with 14+ years designing automated video catalog taxonomies and metadata ontologies for global OTT networks.

Context

  • Streaming Network: {{streaming_network}}
  • Content Catalog Scope: {{catalog_content_types}}
  • Controlled Taxonomy Standard: {{taxonomy_standard}}
  • Regional Regulatory Directives: {{regional_compliance_rules}}
  • Hallucination Penalty Standard: {{hallucination_penalty_rules}}
  • Metadata Granularity Target: {{enrichment_depth_level}}

Task

Produce an in-depth operational analysis and instruction specification for an autonomous metadata enrichment agent generating deep semantic tags, plot micro-summaries, and compliance markers for {{streaming_network}}.

Method

  1. Deconstruct the schema requirements of {{taxonomy_standard}} into rigorous JSON-schema system directives.
  2. Draft strict instructions constraining the agent to {{enrichment_depth_level}} without generating redundant or generic thematic labels.
  3. Integrate {{regional_compliance_rules}} into the agent's content sensitivity classification instructions (e.g., violence, language, drug references).
  4. Specify timestamp-correlated reasoning steps requiring the agent to associate every mood and narrative tag with verified scene markers.
  5. Encode defensive rules enforcing {{hallucination_penalty_rules}} to penalize unverified actor identification or fictitious plot synopses.
  6. Author instruction sets for handling multi-lingual subtitle parsing and cultural adaptation across diverse {{catalog_content_types}}.
  7. Define deterministic validation checks within the prompt ensuring zero structural deviations from the catalog's API ingestion schema.

Constraints

  • MUST restrict output strictly to validated entities within {{taxonomy_standard}}.
  • MUST NOT allow open-ended narrative interpretation outside the observed audiovisual transcript and cue sheet.
  • The agent instructions must enforce strict compliance with {{regional_compliance_rules}} across all target markets.
  • Tagging depth must adhere exactly to the defined {{enrichment_depth_level}} parameters.

Output format

  1. Metadata Architecture Analysis (Taxonomic alignment, token efficiency, and error margins)
  2. Complete Ingestion Agent Prompt (Verbatim system instructions with dynamic schema injections)
  3. Tagging Verification Framework (Step-by-step validation logic for catalog QA teams)
  4. Edge Case & Ambiguity Protocol (Remediation rules for non-dialogue scenes, ambiguous genres, and compliance anomalies)

Self-review

  • Are metadata tags strictly constrained to {{taxonomy_standard}} without creative drift?
  • Does the system prompt mandate timestamp verification for every generated thematic tag?
  • Are all regional age rating and compliance triggers from {{regional_compliance_rules}} fully addressed?
AuraScore breakdown
83/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering12/12 · Strong

Hard boundaries — what the model must and must not do.

Output specification6/14 · Thin

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness5/5 · Strong

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

ai-agents
agents-instructions
media-entertainment
ott streaming
video metadata
taxonomy management