Content
61%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 strong, API-savvy reference: exact tool slugs, parameter formats, and genuinely non-obvious Todoist quirks (priority inversion, ID vs v2_id, due-field rules, filter gotchas) throughout. Weaknesses are redundancy between per-workflow pitfalls and the 'Known Pitfalls'/'Quick Reference' sections, missing validation/verification around the destructive and bulk workflows, and no worked payload examples for the bulk-create and filter cases.
Suggestions
Remove the 'Known Pitfalls' section's duplication of per-workflow pitfalls (or drop the per-workflow Pitfalls subsections and keep only the consolidated one) and delete the vacuous 'When to Use' closer.
Add validation checkpoints: verify batch results after BULK_CREATE_TASKS (e.g., GET_ALL_TASKS filtered to the target project), and require listing/confirming tasks before DELETE_TASK or DELETE_SECTION.
Include one worked example each for the bulk tasks array (showing the nested due object) and a complete multi-term filter string (e.g., "#ProjectName & p4 & tomorrow") to make the trickier payloads copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Most content is dense, non-obvious API knowledge Claude does not have (due-field exclusivity, priority inversion, ID vs v2_id quirks, filter syntax pitfalls), so much of it earns its tokens. However, the 'Known Pitfalls' section restates pitfalls already given per-workflow (ID formats in sections 2-3, priority inversion in section 1, filter syntax in section 4), the Quick Reference table largely re-lists the tool sequences, and the closing 'When to Use' section is filler ("This skill is applicable to execute the workflow or actions described in the overview") — matching 'mostly efficient but includes some unnecessary explanation or could be tightened'. It is above a 2 because the padding is duplication of genuinely useful content rather than explanations of concepts Claude already knows. | 3 / 5 |
Actionability | Guidance is highly concrete: exact tool slugs (TODOIST_CREATE_TASK), parameter names with formats and example values (due_string "tomorrow at 3pm", due_datetime RFC3339, filter operators "&", "|", "!"), and worked tool sequences per workflow — 'mostly executable guidance; concrete commands with minor gaps'. It is not a 5 because no complete worked example is given for the trickier payloads: a full bulk tasks array (with the nested due object) or a complete filter string, so a first attempt at those still requires assembly from prose. | 4 / 5 |
Workflow Clarity | Sequences are explicit and consistently tagged ([Prerequisite]/[Required]/[Optional]), and some verification exists (resolve project IDs before acting, list existing sections to avoid duplicates, confirm connection ACTIVE). But the batch and destructive operations — BULK_CREATE_TASKS, DELETE_TASK, DELETE_SECTION — have no validation checkpoints (no post-creation verification of batch results, no confirm-before-delete step), so per the rubric's cap for destructive/batch workflows without validation, workflow clarity is capped at 3 ('steps listed but validation gaps'). It is above a 2 because the sequencing itself is coherent and well-structured. | 3 / 5 |
Progressive Disclosure | The skill is a single well-organized file with no bundle files (no references/, scripts/, or assets/ exist), clear section headers per workflow, common patterns, and a quick-reference table — 'good structure; most content is appropriately placed; minor organization gaps'. It is not a 5 because at ~230 lines the per-tool parameter details and quick-reference table are candidates for a references/ file that would keep SKILL.md a leaner overview, and not a 3 because navigation is easy and nothing is buried or mis-signaled. | 4 / 5 |
Total | 14 / 20 Passed |