Performance Review Writer — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited Performance Review Writer (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Performance reviews are legally significant, career-defining, and often written in 20 minutes from memory. That gap causes recency bias, grade inflation, and feedback too vague to act on. This skill closes it.
Collect raw material from four sources before writing a word: the employee's self-review, peer feedback verbatims, the role's defined expectations for their level, and a concrete list of their work from the review period (PRs merged, projects owned, incidents handled, cross-functional contributions). Do not write from memory. If the list is hard to reconstruct, that is itself a signal about visibility and documentation.
Start from the role expectations document, not a bell curve. Ask: did this person consistently meet, exceed, or fall short of the expectations defined for their level? If no level expectations document exists, write one before the review cycle, even a rough draft. Ratings applied without a standard are arbitrary.
Open with a one-sentence summary of the overall rating and why. Then structure the body around 3-4 themes drawn from the evidence, not the review form's generic categories. For each theme: state the observation, cite a specific example, and name the impact. The format is 'In [context], [name] did [specific thing], which resulted in [concrete outcome].' Avoid adjectives without evidence — 'strong communicator' means nothing; 'drove alignment between design and engineering on the auth redesign, reducing decision latency by cutting a recurring 3-way meeting to async' means something.
Every review includes one primary growth area and one specific suggested action for the next review period. Growth areas tied to a concrete next step are acted on; vague suggestions ('work on your influence') are not. Do not list more than two growth areas — that signals the manager has not prioritized.
Halo effect: one visible win or loss skewing the whole rating. Recency bias: events from the last 6 weeks overweighting 12 months of work. Leniency bias: avoiding hard ratings to sidestep a difficult conversation. If a rating requires a difficult conversation, the review is where it starts, not ends.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.