Parallel and Compositional Call Planning Schemas
Design declarations that make independent actions callable in parallel and dependent actions chainable, then validate the resulting call plan.
Two declaration groups (independent vs dependent), an expected call plan showing which calls fan out and which must wait, and validation checks to run before execution.
Role
You are designing a tool set whose call topology the model must infer correctly.
Task
For the scenario {{scenario_description}}, produce two declaration groups. Group A covers {{independent_actions}}: keep their parameter sets disjoint so they can be emitted in one turn without ordering constraints. Group B covers {{dependent_action_chain}}: make each step's required parameters obtainable only from a prior step's result, naming the carried values in {{shared_state_fields}}. Then write the expected call plan as an ordered list marking parallel batches and blocking edges, and list the pre-execution checks from {{validation_rules}}.
Context
Models fan out correctly only when independence is visible in the schemas; conversely, dependent steps must not be independently satisfiable or the model will guess intermediate values.
Inputs
- {{scenario_description}}
- {{independent_actions}}
- {{dependent_action_chain}}
- {{shared_state_fields}}
- {{validation_rules}}
Constraints
- No Group A function may require another Group A function's output
- Every Group B step must be blocked on a named upstream field
- Do not allow a dependent parameter to have a default that hides the dependency
- Validate arguments before any side-effecting call
Output Format
Markdown: Group A declarations, Group B declarations, ordered call plan with parallel/blocking annotations, validation checklist.
Quality Criteria
- Parallel batch is genuinely order-independent
- Dependency edges traceable through named fields
- Call plan matches the declared schemas
- Validation checks cover type, range and authorisation