Content
65%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 framework itself is excellent — specific, quantitative, and immediately usable, with a documentation standard that makes outputs reproducible. Weaknesses are structural: the assessment workflow is implied rather than explicitly sequenced with validation checkpoints, the single 260-line file inlines content that belongs in reference files, and there is localized redundancy (duplicated matrix, restated characteristics).
Suggestions
Add an explicit numbered assessment workflow (e.g., 1. Gather facts → 2. Rate severity against the table, citing the matching anchor → 3. Rate likelihood → 4. Compute and map the score to a level → 5. Apply that level's action list → 6. Document via the memo template) with a checkpoint that re-verifies each rating against the anchor descriptions.
Split the Documentation Standards (memo template and risk-register entry) and the When to Escalate to Outside Counsel sections into reference files (e.g., references/memo-format.md, references/escalation.md), keeping SKILL.md as a concise overview of the matrix, levels, and actions.
Trim the redundant ASCII risk matrix (the Severity x Likelihood multiplication is trivial) and collapse the per-level 'Characteristics' bullets, which restate the severity table, into the level headers or examples.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient — the severity/likelihood tables, action lists, and memo template earn their tokens and assume legal competence (no explaining what a contract or NDA is). However, the ASCII risk matrix is redundant with the trivial "Risk Score = Severity x Likelihood" formula, and each level's "Characteristics" bullets largely restate the severity table, so there is more than minor trimming to be done. Not 4 because the redundancy is noticeable rather than minor; not 2 because the padding is localized and most sections are dense and useful. | 3 / 5 |
Actionability | For an instruction-only skill the guidance is fully executable: quantified severity thresholds ("< 1% of relevant contract/deal value", "1-5%", "5-25%", "> 25%"), a copy-paste-ready memo template with ten numbered sections, a complete risk-register field table, and concrete per-level action lists plus worked example scenarios at each risk level. Meets the 5 anchor's standard of specific examples covering common cases. | 5 / 5 |
Workflow Clarity | The operational sequence (rate severity via table → rate likelihood via table → compute score → look up level → apply that level's recommended actions → document via memo template) is present but only implicit in the section ordering and memo template; no explicit step enumeration or validation checkpoint exists (e.g., re-checking a rating against the anchor descriptions). This is not a destructive/batch skill, so no cap applies, but it sits at the 3 anchor (sequence present, checkpoints implicit) rather than 4, which requires a clearly enumerated sequence with most checkpoints explicit. | 3 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are absent), so the ~260-line body is a single monolith with good internal headers. The memo template, risk-register format, and outside-counsel selection criteria are self-contained chunks that clearly belong in separate reference files. Good structure, but content that should be separate is inline — matching the 3 anchor; not 2 because organization and navigation within the file are solid. | 3 / 5 |
Total | 14 / 20 Passed |