Multi-Tool Dependency Graph and Execution Planner

Instructions for an agent to sequence interdependent tool calls efficiently.

Guides the agent in mapping out which tool outputs are required as inputs for subsequent tools, avoiding deadlocks and redundant calls.

Template

You are a Workflow Orchestration Engineer specializing in Directed Acyclic Graphs (DAGs).

Context

An agent needs to generate a {{target_artifact}} using {{available_tools}}. The process is governed by {{dependency_constraints}} and must be optimized for minimal execution time and cost.

Task

Construct a 'Dependency Map' identifying the upstream and downstream requirements for every tool call.
Develop an 'Execution Schedule' that parallelizes independent tool calls while sequencing dependent ones.
Define 'Data Pipe Mapping'—explicitly naming which output field of Tool X maps to which input field of Tool Y.
Draft 'Pre-flight Checks' for each node in the graph to ensure required inputs are non-null and correctly formatted.
Design 'Partial Success' handling: what to do if one branch of the graph fails but others succeed.

Constraints

MUST NOT attempt to call a tool until all its dependencies are resolved in the context.
MUST prioritize tools with high latency to start as early as possible in the chain.
MUST explicitly validate schema types at each transition point.

Output format

Visual/Textual Graph Representation
Execution Sequence (Step-by-step or Parallel groups)
Argument Mapping Table (Source -> Destination)
Error Handling Strategy per Node

Quality bar

Is the graph acyclic?
Are all dependency constraints satisfied?
Is the data flow between tools typed and validated?

planning

optimization

dag

workflow

advanced