CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

differential-review

Performs security-focused differential review of code changes. Adapts analysis depth to codebase size, uses git blame for context, calculates blast radius by counting callers, checks test coverage of modified code, and generates a markdown report. Use when reviewing a PR, commit, or diff for security vulnerabilities, checking whether a change re-introduces a previously fixed bug, asking what else a change could break, or finding which modified code has no test covering it.

Quality

77%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./plugins/differential-review/skills/differential-review/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

56%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The SKILL.md is well-organized with a clear workflow, useful tables, and a genuinely good progressive-disclosure design — but the bundle undercuts it: all four referenced detail files are missing, so the actual instructions the skill promises cannot be loaded. Redundant sections (duplicated file lists, Do/Don't repeats, time estimates) also dilute token efficiency.

Suggestions

Add the missing bundle files (methodology.md, adversarial.md, reporting.md, patterns.md) or inline the essential per-phase steps into SKILL.md — as shipped, the decision tree and links point to files that cannot be read.

Remove the duplicated file listing ('Supporting Documentation' repeats the Decision Tree entries) and merge the 'Tips for Best Results' Do/Don't lists into the Quality Checklist and Red Flags sections.

Drop the wall-clock time estimates in Example Usage ('~30 minutes', '~6-8 hours') — they are not actionable for Claude and add tokens without guidance.

DimensionReasoningScore

Conciseness

The body is mostly efficient (compact tables for risk levels, size strategy, rationalizations), but includes several sections that could be tightened: the referenced-file list is duplicated in both the Decision Tree and 'Supporting Documentation', the Do/Don't lists largely repeat the Quality Checklist and Red Flags, and wall-clock estimates like 'Time: ~30 minutes' and 'Time: ~6-8 hours' add no actionable value. This fits the anchor of 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trimming anchor of 4.

3 / 5

Actionability

Some guidance is concrete — the risk-level trigger table, codebase-size thresholds ('SMALL (<20 files)'), red-flag escalation rules, and the executable issue-writer command — but the phase-by-phase instructions the workflow depends on are delegated to methodology.md, adversarial.md, reporting.md, and patterns.md, none of which exist in the bundle. Key execution details are therefore missing, matching the 'some concrete guidance but incomplete; missing key details' anchor rather than the mostly-executable anchor of 4.

3 / 5

Workflow Clarity

The workflow is clearly sequenced ('Pre-Analysis → Phase 0: Triage → ... → Phase 6: Report'), the decision tree routes users to the right phase, the Quality Checklist provides verification checkpoints before delivery, and Red Flags define explicit escalation triggers. It is not a 5 because per-phase steps and error-recovery loops are not in the body — they are promised in files absent from the bundle, leaving minor validation gaps.

4 / 5

Progressive Disclosure

References are clearly signaled, one level deep, and duplicated in the Decision Tree and Supporting Documentation sections — structurally a 4. However, scoring against the actual bundle, none of the four referenced files (methodology.md, adversarial.md, reporting.md, patterns.md) exist; only an unreferenced assets/trail-of-bits-mark.svg is present. The claimed file split does not exist, so the structure fails to deliver on its navigation promises.

3 / 5

Total

13

/

20

Passed

Description

88%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description: concrete capability list, explicit third-person 'what', and a rich 'Use when...' clause with multiple natural trigger phrases. Only minor improvements possible — adding a few common synonyms (e.g., 'security audit', 'regression') and sharpening distinctiveness from general code-review skills would make it a clean sweep.

DimensionReasoningScore

Specificity

The description lists five concrete, distinct actions — 'Adapts analysis depth to codebase size, uses git blame for context, calculates blast radius by counting callers, checks test coverage of modified code, and generates a markdown report' — with comprehensive coverage of the skill's capabilities. It matches the anchor for multiple specific concrete actions; a 4 would require minor coverage gaps, which are not present.

5 / 5

Completeness

It explicitly answers both questions: a clear third-person 'what' ('Performs security-focused differential review of code changes...') and an explicit 'Use when...' clause with four concrete trigger scenarios. This matches the top anchor exactly; a 4 would mean the 'when' was present but less explicit or specific.

5 / 5

Trigger Term Quality

Natural trigger phrases are well covered: 'reviewing a PR, commit, or diff for security vulnerabilities', 're-introduces a previously fixed bug', 'what else a change could break', and 'no test covering it'. A few natural synonyms users might say are missing (e.g., 'security audit', 'regression check'), so it falls just short of the comprehensive-synonyms anchor of 5.

4 / 5

Distinctiveness Conflict Risk

The security-focused differential-review niche with triggers like 're-introduces a previously fixed bug' and blast-radius analysis is mostly distinct from general review skills. There is minor overlap risk with generic code-review skills for the trigger 'reviewing a PR... or diff', which keeps it just below the minimal-conflict anchor of 5.

4 / 5

Total

18

/

20

Passed

Validation

93%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

CriteriaDescriptionResult

relative_links

Relative link issues: 5 missing

Warning

Total

15

/

16

Passed

Repository

trailofbits/skills

Path

plugins/differential-review/skills/differential-review/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.