Content
71%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.
A well-structured, tool-specific workflow that is concrete and easy to follow end-to-end, with strong error-recovery detail around polling and status quirks. The dominant defect is the dangling research-brief-template.md reference — the primary deliverable's structure is undefined — plus minor redundancy in the Notes section.
Suggestions
Add research-brief-template.md to the bundle (or inline the brief structure in Step 7) so the primary deliverable is fully specified.
Delete or merge the Notes section, which repeats Step 4's fast/deep mode guidance and Step 1's scope-confirmation instruction.
Include one example research_start call with its query and mode arguments to make Step 4 copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient: tables carry the research types and artifact parameters, tool names are given without explaining what MCP or research is, and every step earns its tokens. It misses level 5 because the "Notes" section largely duplicates earlier step content ("Fast mode is recommended as the default. Deep mode... can take 10+ minutes and occasionally stalls" restates Step 4; "Always confirm the research scope... before starting" restates Step 1). | 4 / 5 |
Actionability | Highly actionable: exact MCP tool names (notebook_create, source_add, research_start, research_status, research_import, notebook_query, studio_create, studio_status, download_artifact), a notebook naming convention, a file-path convention, sample notebook_query questions, and concrete parameter tables with timings. Not level 5 because Step 7's core deliverable depends on "the template from [research-brief-template.md](research-brief-template.md)" which does not exist in the bundle, and no example research_start query arguments are shown. | 4 / 5 |
Workflow Clarity | The 9 steps are clearly sequenced with named tools per step, polling checkpoints (research_status, studio_status), an error-recovery fallback ("Use the query parameter as fallback matching -- task IDs can change"), a status quirk workaround (check audio_url instead of waiting for "completed"), and a user-confirmation gate. It falls short of level 5 because Step 7 cannot be executed as written: the referenced research-brief-template.md is missing, so the brief structure is undefined at that checkpoint. | 4 / 5 |
Progressive Disclosure | Inline structure is good (clear sections, tables, step headers), but scored against the actual bundle: no references/, scripts/, or assets/ directories exist, and the sole external reference, research-brief-template.md (linked twice), is not present. Navigation breaks at the point of use, leaving the reference clearly signaled but dangling, which fits the 'references present but organization incomplete' middle anchor rather than the level-4 'references mostly clear and working' anchor. | 3 / 5 |
Total | 15 / 20 Passed |