AGENT SKILL

diff-first-review

Use when reviewing a change, pull request, patch, or diff for defects, regressions, risks, or missing tests.

License 许可
GPL-3.0
最近更新
2026-11-15
Stars
★ 0
Forks
0

01 前言 Front Matter

name
diff-first-review
description
Use when reviewing a change, pull request, patch, or diff for defects, regressions, risks, or missing tests.

03 SKILL.md 正文

Review the change surface before repository history.

  • Start with changed-file names, diff statistics, then relevant hunks.
  • Trace only changed symbols to their definitions, callers, contracts, and focused tests.
  • Read unchanged files only when a hunk exposes a concrete dependency, invariant, or compatibility risk.
  • Prioritise correctness, security, behavioural regression, and missing tests; ignore unrelated style.
  • Report findings as path:line - severity: issue; explain impact and smallest remedy.
  • No finding: state this and name residual test gaps or uncertainty.

04 同仓库更多 Skills

hmlendea
hmlendea/github-copilot-memories

ui-design-review

Use when designing, building, reviewing, or repairing user interfaces, or when UI work must be verified before delivery. Covers accessibility, touch and interaction, performance, style selection, layout and responsive behaviour, typography and colour, animation, forms and feedback, navigation, and charts, with a prioritised rule set, a native and mobile pre-delivery checklist, and a browser-driven audit procedure that reports evidence-ranked findings.

★ 0
hmlendea
hmlendea/github-copilot-memories

release-verification

Use after a release, publication, or deployment to verify the delivered artefact and observe the affected surfaces until the observation window is clean. Covers release identity, surface evidence, finding classification, observation cadence, and partial-verification reporting.

★ 0
hmlendea
hmlendea/github-copilot-memories

pr-babysit

Use when monitoring a pull request or review cycle until it is ready to merge. Covers check observation, unresolved review threads, verification prior to resolution, and evidence-based completion.

★ 0
hmlendea
hmlendea/github-copilot-memories

supply-chain-security

Use when assessing software supply chain exposure from dependency manifests, lockfiles, SBOMs, VEX records, licence declarations, package names, or CVE identifiers. Covers dependency and transitive risk, SBOM validation against the NTIA minimum elements in CycloneDX and SPDX, VEX status interpretation, licence compliance, typosquatting and dependency confusion, and CVE triage through CVSS 4.0, EPSS, SSVC 2.1, and the CISA KEV catalogue. Produces a prioritised, SLA-bound remediation assessment.

★ 0
hmlendea
hmlendea/github-copilot-memories

secure-code-review

Use when performing a security-focused code review of a pull request, diff, file, or module, or when code touching authentication, authorisation, cryptography, input handling, deserialisation, file handling, or logging requires verification. Performs a structured review against OWASP ASVS 4.0.3 requirements and the CWE Top 25 (2024), producing findings mapped to ASVS controls and CWE identifiers with severity ratings and specific remediation guidance.

★ 0
hmlendea
hmlendea/github-copilot-memories

secrets-management

Use when examining secret handling, credential exposure, vault or cloud secrets manager integration, environment-file hygiene, rotation policy, or credential access for agents and CI/CD pipelines. Performs a structured secrets management assessment against the OWASP Secrets Management Cheat Sheet and NIST SP 800-57 Part 1 Rev 5, covering detection tooling, exposure routes, rotation automation, and centralised storage, without ever reproducing a secret value.

★ 0