Content
36%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 body reads as a well-written product landing page rather than an operational skill: it competently describes what the skill would do, its limits, and a sample result, but contains none of the executable material (legal criteria lists, risk matrix, Article 30 field requirements, templates) needed to actually perform the work, which it acknowledges is not in the folder. It works only as marketing documentation of a skill whose substance lives elsewhere.
Suggestions
Inline or ship as bundle files the actual operational content: the eight Belgian GBA processing types, the Article 35.3 criteria, the nine EDPB criteria, the likelihood-times-impact matrix scale, and the mandatory Article 30.1/30.2 field lists — currently all are named but none are provided.
Add a sequenced workflow with checkpoints, e.g. 1) gather processing facts, 2) run each of the three DPIA gates and record per-gate results, 3) document the negative conclusion or proceed to draft, 4) score risks before/after measures, 5) flag for DPO/legal review before delivery.
Remove the '## Questions' FAQ (it restates 'What it does') and the promotional footer links, which spend tokens without adding instructional value.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and does not re-explain GDPR concepts Claude already knows, but the '## Questions' FAQ ('How is risk scoring calculated?', 'What criteria does it use for DPIA checks?') merely repeats 'What it does', and the promotional footer ('More skills, including the paid catalog: https://...') is pure noise. That is 'some unnecessary explanation or could be tightened' — the 3 anchor — rather than the minor-trim case of 4. | 3 / 5 |
Actionability | The body names its machinery — 'the Belgian GBA list of eight mandatory processing types, Article 35.3, and the nine EDPB criteria', 'a likelihood-times-impact risk scoring matrix' — but never supplies any of it: no GBA list, no EDPB criteria, no matrix scale, no mandatory Article 30 field list, and it openly states 'Scripts, extension source, and templates from the original package are not in this folder.' The sample output gives a hint of the result format, lifting it above 'entirely vague' (1), but it is high-level description with the specific steps to execute missing — the 2 anchor. | 2 / 5 |
Workflow Clarity | Running a 3-gate obligation test and then drafting a risk-scored DPIA is inherently multi-step, yet the only sequencing is implied ('Next: Should we perform a DPIA check for the cameras?'); no ordered steps are given for either the gate test or the DPIA drafting, and there are no validation or review checkpoints even though legal-document output warrants them. This is a rough implied sequence with many gaps and no validation — the 2 anchor. | 2 / 5 |
Progressive Disclosure | The ~50-line body is cleanly sectioned (What it does, Use cases, Known limitations, Example, Questions), but the operational content that clearly belongs in reference or script files (templates, the GBA/EDPB lists, the risk matrix) is entirely absent rather than moved to a bundle, and the only pointers given lead outside the skill (the Agensi page and a paid catalog) rather than to navigable files. 'Some structure but could be better organized' — the 3 anchor — fits better than 4, where references would be present and mostly clear. | 3 / 5 |
Total | 10 / 20 Passed |