Content
57%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The COBRA-specific core (sections 1-5) is accurate and highly actionable with a runnable example, but it is buried in a generic template shell: duplicated section headers, boilerplate governance sections, and unlinked reference files. Removing the template padding and wiring the references in by name would substantially improve the skill.
Suggestions
Delete the generic template sections (the first 'When to Use'/'Key Features'/'Dependencies'/'Example Usage'/'Implementation Details' variants, 'Deterministic Output Rules', 'Completion Checklist', 'Scope Reminder') and keep only the COBRA-specific sections 1-5, eliminating the duplicated headers.
Replace the vague 'reference material available in references/' mentions with explicit named links, e.g. '**Workflows**: See [workflows.md](references/workflows.md)' and '**API reference**: See [api_quick_reference.md](references/api_quick_reference.md)'.
Make the 'Recommended Workflow' concrete for COBRA tasks: e.g., load model -> check solver status -> run the requested analysis (FBA/FVA/deletion/medium) -> validate results (e.g., verify solution status is 'optimal', sanity-check growth rate) -> report, instead of the current abstract four-step boilerplate.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Roughly half the body is generic template padding: duplicated section headers (two 'When to Use', 'Key Features', 'Dependencies', 'Example Usage', 'Implementation Details'), plus 'Deterministic Output Rules', 'Completion Checklist', a 'Scope Reminder' that restates the description verbatim, and a 'Quick Validation' section that declares no validation is required. The COBRA-specific sections 1-5 themselves are lean and accurate, keeping this above a 1. | 2 / 5 |
Actionability | The example script is complete and copy-paste runnable (load_model -> optimize -> pfba -> FVA -> single_gene_deletion -> medium edit -> sampling), and section 5 documents genuinely non-obvious API behavior such as re-assigning the full medium dict and using `with model:` for reversible edits. | 5 / 5 |
Workflow Clarity | The 'Recommended Workflow' is abstract boilerplate ('select the documented execution path', 'run a final validation pass') with no COBRA-specific checkpoints, and the 'Quick Validation' section explicitly offers no verification step. Sequence is present but validation is implicit/missing, matching the anchor for a 3. | 3 / 5 |
Progressive Disclosure | The bundle provides references/api_quick_reference.md and references/workflows.md, but the body never names or links them — only vague mentions like 'Reference material available in `references/` for task-specific guidance'. References exist but are not clearly signaled, and detail that belongs in those files (section 5 API specifics) is inlined. | 3 / 5 |
Total | 13 / 20 Passed |