Delegate complete coding tasks to the local OMP default model and inspect the result.
Server defines 9 tools with schemas and descriptions present, but quality is inconsistent. Tool names follow verb_noun pattern (omp_delegate, omp_status, omp_cancel) which is good. However, descriptions are generic and lack actionable context for LLM selection. Parameters have types and descriptions, but many lack constraints (enums, ranges, patterns). Error handling is minimal, no recovery guidance or categorization. Output schemas are partially documented via Zod but not exposed to the LLM. No tool annotations (readOnlyHint, destructiveHint) despite clear risk levels. Batch operations (omp_batch_delegate, omp_batch_status) are complex but descriptions don't explain when to use batch vs single-task tools. Overall: functional but below production baseline.
Delegate multiple coding tasks as a batch group with optional parallelism and dependency management
Get the status of a batch group execution
Cancel a delegated task
Cancel a batch group execution
Clean up state and artifacts for completed or failed tasks
Delegate a complete coding task to the local OMP default model
Get the current status of a delegated task
Wait for a delegated task to complete
Descriptions lack actionable context. 'Delegate a complete coding task' does not explain WHEN to use omp_delegate vs omp_batch_delegate, what 'acceptance criteria' means operationally, or what the LLM should expect in return. Descriptions should be 50-200 chars and answer: what, when, why.
No tool annotations despite clear risk levels. omp_delegate and omp_batch_delegate are WRITE operations; omp_cancel and omp_cancel_group are REVERSIBLE; omp_cleanup is DESTRUCTIVE. Annotations (readOnlyHint, destructiveHint, idempotentHint) are missing from tool definitions, preventing LLMs from reasoning about side effects.
Parameters lack constraints. 'timeout_minutes' and 'max_attempts' have no min/max bounds; 'max_parallel' is unbounded. Free-form numeric parameters invite LLMs to pass absurd values (timeout_minutes=999999, max_parallel=10000). Add ranges in descriptions and schema.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 48 | 2025-06-18+ | v2 |
Wait for a batch group to complete execution
Error handling is absent. No recovery guidance, categorization, or actionable messages. If a task fails, the LLM receives only a status string with no hint about whether to retry, ask the user, or abort. Implement error classification (retryable, user-fixable, fatal) and recovery suggestions.
Batch operations (omp_batch_delegate, omp_batch_status, omp_wait_group) are complex with nested task objects and dependency management, but descriptions do not explain the dependency model, parallelism semantics, or when batch is preferable to sequential single-task calls. Add examples or clarification.