Nichalas Barnes

colorpulse6-skills

AI-powered PR review with contract verification, frontend and backend rules, inbox mode, and incremental re-review. Auto-detects project type and applies relevant rules.

Quality-gated push pipeline. Auto-detects package manager and runs type check, lint, tests, build in parallel. Auto-fixes failures, creates PR, polls CI.

Interactive walkthrough for addressing review comments on your own PRs. Fix, push back, defer, or skip each thread with batched implementation and auto-reply.

Generates a browser-viewable HTML timeline of recent Claude Code sessions with key decisions, PRs, commits, files, and active time. Daily or weekly recaps from ~/.claude/projects transcripts.

Independent separate-model (Codex / gpt-5.5) code review to complement same-model Claude audits. Writes a neutral brief (independent findings first, then confirm/refute), runs `codex exec --sandbox read-only`, and reconciles — a different model catches what same-model reviewers miss.

Template plugin: an orchestrator skill that fans work out to parallel specialist agents and aggregates their findings into one report. Copy it to start a new skill.

Multi-lens, multi-model code review conductor. Runs scoped Claude review agents in parallel (security, concurrency, performance, contract) while a separate model (Codex) reviews independently, then reconciles into one cross-model verdict. Pairs with codex-second-opinion.

Triage a diff by change risk and escalate only the risky parts to deeper / separate-model review. A cheap deterministic pass scores each changed file (danger paths, churn, blast radius); only HIGH-risk files get the expensive review (review-board / codex-second-opinion).

Adversarially review a design doc / spec / RFC before you build it. Runs a pre-mortem plus parallel design-review lenses (edge cases, failure handling, data integrity, rollout/rollback, observability, scaling, security/STRIDE), then reconciles into a prioritized hardening report with an APPROVED/REVISE verdict.

Audit dependency-manifest / lockfile changes in a diff. A zero-network offline pass scores the diff (semver deltas, pin loosening, lockfile-integrity changes, install scripts, typosquatting); then per-ecosystem scanner agents run the real tools (osv-scanner, npm audit, pip-audit, cargo audit, govulncheck) and maintainer-health checks.

Find the most important MISSING tests in a change. Fans out gap-finder agents over the changed files; each ranks untested code by risk = impact x likelihood (untested branches, error paths, boundaries, weak assertions), not raw coverage %. Produces a prioritized list of tests to add with proposed cases.

Meta-skill for authoring new skills/plugins in this repo. Interviews for intent, picks the right shape (inline vs orchestrator + parallel agents), scaffolds the files, writes them against a distilled best-practices guide, self-reviews against an anti-pattern checklist, and registers the plugin.

Validate every plugin in this marketplace for structural + registration consistency: valid plugin.json, name matches directory, registered in marketplace.json + README, SKILL.md frontmatter present, scripts executable + compilable, and references over 100 lines carrying a Contents TOC.