You are a principal engineer designing tool interfaces for a...

Prompt

You are a principal engineer designing tool interfaces for autonomous AI agent systems. I am upgrading "AI Visual Director Production" v7.8.5 → v8.0.0. Currently, the skill DESCRIBES production steps but has no formal interface to actual tools. Agents cannot autonomously execute gates—they can only narrate them. PRODUCTION GATES THAT NEED TOOL CONTRACTS: - R2: Concept divergence verification (needs image gen + comparison) - R3: Reference calibration / style tile generation (needs image gen + aesthetic scoring) - R4: Character model sheet locking (needs consistent character gen + drift detection) - R5: Timing/beat map validation (needs audio analysis + EDL parsing) - R6: Generation execution with durable-resolution proof (needs video gen + frame consistency check) - R7: Edit assembly verification (needs timeline parsing + transition validation) - Meta: Gate evidence validation itself (needs schema validation + content verification) YOUR TASK: Design the complete Tool Integration Contract layer using Model Context Protocol (MCP) as the standard. Produce: 1. MCP tool definitions (JSON) for each gate-critical tool above 2. For each tool: exact input schema, output schema, error codes, retry semantics, and idempotency guarantees 3. A meta-tool `validate_gate_evidence` that accepts a gate receipt + schema reference and returns structured pass/fail with violation details 4. Fallback protocol: when a tool is unavailable or errors, what manual/descriptive path the agent must take instead 5. Tool discovery contract: how an agent enumerates available tools and their capabilities at runtime 6. Security boundary: which tools require human approval vs autonomous execution, and how that's encoded CONSTRAINTS: - Follow MCP specification strictly (tools/resources/prompts separation) - Inputs/outputs must be JSON-serializable; no binary blobs (use URIs/hashes) - Every tool must declare its resource cost tier (LOW/MED/HIGH) for diagnostic budget alignment - Error responses must be machine-actionable, not human-readable prose - Design for heterogeneous backends: the same contract must work whether the backend is Replicate, ComfyUI, Runway, or local SDXL Do NOT ask clarifying questions. Make definitive design choices. Output complete MCP JSON definitions ready for implementation.