bankielewicz

Independently validate a pull request from another AI session before a human merge. Enforce CI/head-SHA triage, an isolated review worktree, repository gates, negative-path proof for guards, reproduced findings, a three-tier verdict, cold fix handoff, delta review, and an explicit merge-or-hold outcome. Use for delegated builder PR review, fix delta review, or agent-work validation. Do not use for your own uncommitted diff or a plain human-authored PR without delegation context.

Independently validate a pull request from another AI session before a human merge. Enforce CI/head-SHA triage, an isolated review worktree, repository gates, negative-path proof for guards, reproduced findings, a three-tier verdict, cold fix handoff, delta review, and an explicit merge-or-hold outcome. Use for delegated builder PR review, fix delta review, or agent-work validation. Do not use for your own uncommitted diff or a plain human-authored PR without delegation context.

Execute a rejected pull request's fix round as a cold implementer session. A reviewer found defects and wrote a handoff comment with numbered findings and C0–C7 checkpoints; work that punch list under invariants enforced by a PreToolUse hook and gate scripts — reproduce the reviewer's red before touching the tree, stay inside the handoff's fence, one `F<n>:` commit per finding, record the output of every acceptance criterion, post a closing report, and STOP without merging. Use whenever a review comes back with changes requested and you are handed the PR plus its findings. Do not use it to author the original plan (that is `delegate`), to build a slice from a plan with no prior review (that is `implement-workflow`), or to review someone else's PR (that is `review-workflow`).

Execute a rejected pull request's fix round as a cold implementer session. A reviewer found defects and wrote a handoff comment with numbered findings and C0–C7 checkpoints; work that punch list under invariants enforced by a PreToolUse hook and gate scripts — reproduce the reviewer's red before touching the tree, stay inside the handoff's fence, one `F<n>:` commit per finding, record the output of every acceptance criterion, post a closing report, and STOP without merging. Use whenever a review comes back with changes requested and you are handed the PR plus its findings. Do not use it to author the original plan (that is `delegate`), to build a slice from a plan with no prior review (that is `implement-workflow`), or to review someone else's PR (that is `review-workflow`).

Create Web, GUI, or Terminal design specifications in story, standalone, or extract mode from a customizable brief, brand, and visual direction; generate design.md, implementation artifacts, and web artifacts when the selected target is Web. Use through /create-design when a project needs a design source of truth or an existing mockup must be extracted.

Create Web, GUI, or Terminal design specifications in story, standalone, or extract mode from a customizable brief, brand, and visual direction; generate design.md, implementation artifacts, and web artifacts when the selected target is Web. Invoke explicitly with $spec-driven-design.

Run the delegation loop: act as a standing architect/reviewer who surveys ideas into merit-ranked plans, pins design decisions with the maintainer (AskUserQuestion), records maintainer-pinned plans in evidence docs, generates full-fidelity zero-ambiguity build prompts for cold AI builder sessions (Claude Code / Codex / other), resolves builder HALTs, reviews the resulting PRs against watchpoints declared at prompt time, and records outcomes. Language- and project-agnostic; per-project specifics live in .delegate/config.md. Also covers greenfield bootstrap (idea → repo → first slice) and the design-session leg (mockup prompts for Claude Design, mockup review with numbered rulings). Invoke via `/delegate PHASE`, or when the user asks to "generate a build prompt for another session", "pin a design", "generate a mockup prompt", "review the mockup", "review the builder's PR", or run any part of this loop.

Execute a cold implementation session from an externally authored plan, handoff, task-register entry, or build prompt. Enforce slice arming, consultation, AskUserQuestion HALTs, test-first commit order, repository gates, one worktree/branch/PR, cold-review handoff, and stop-without-merge. Use when the user asks to implement or execute a plan written elsewhere. Do not use to author plans or resolve builder HALTs (`delegate`), review completed PRs (`review-workflow`), or apply a review punch list (`fix-workflow`).

Re-audits a consumer DNA export's existing analysis documents into a deterministic, allow-list-cited, HelixyAI-branded HTML report. Use when the user wants to re-audit, rebuild, or verify a prior genome-analysis document set against their raw chip genotypes — e.g. "redo my genome report", "rebuild my health analysis from my 23andMe file", "verify what my prior genome docs claimed". It builds a genotype-truth INDEX from the raw export, audits the prior docs for unsupported or aspirational claims, rebuilds them with rsID + allow-list citations and a provenance block, and renders an interlinked HTML report. Requires a genome-reader install for parsing. Not a diagnostic tool — it restates authoritative-source guidance (CPIC/FDA/peer-reviewed) only and every recommendation ends with "consult your prescribing clinician".

Parses, summarizes, queries, and converts DNA and genome data files locally. Use whenever the user shares a FASTA, FASTQ, VCF, BAM/SAM/CRAM, BED, GFF/GTF, or a 23andMe / AncestryDNA / MyHeritage raw export — even if they don't name the format. Use for sequence inspection, variant counts, genotype lookups by rsID, region extraction, FASTQ QC, VCF Excel reports, format conversion, and DNA→protein translation. Do not use for de novo assembly, read alignment, variant calling, phylogenetic trees, or anything requiring an external paid API or uploading user data.

Specify, capture, scaffold, build, compare, secure, migrate, and verify authorized clean-room MVP reimplementations and evidence-grounded brownfield changes. Use when Codex is asked to clone, recreate, replicate, or reverse-specify software; enhance, extend, modernize, refactor, upgrade, migrate, or harden an existing repository; generate an AI-optimized clone or enhancement specification; produce or implement gaps_analysis.md; migrate a clone-pack/v1 pack; prove target-versus-clone and before-versus-after behavior; or add GUI, Playwright, end-to-end, full-stack QA, or CI verification without guessing. Covers websites, SaaS products, APIs, services, servers, clients, libraries, SDKs, CLIs, browser extensions, AI/ML systems, data pipelines, storage systems, distributed or realtime systems, games, and embedded or IoT software.

Professional-grade personal color analysis matching the rigor of $300-$400 human consultant services. Use whenever the user asks about personal color analysis, seasonal color typing, "what colors suit me," "what's my color season," "am I a warm or cool tone," color palette recommendations, wardrobe color planning, or uploads selfies asking which colors flatter them. Also trigger for specific season names (Soft Autumn, Bright Winter, Cool Summer, etc.), undertone determination, draping, or the 12-season system. Also for closet audits ("is this item on my palette?"), single-item checks, and palette refinement for users already typed. Enforces strict photo intake, runs four-dimension analysis with mandatory mid-analysis checkpoint, performs adjacent-season draping comparison, handles deep-skin protocols and olive complexions, and produces a full deliverable with hex codes, metals, grooming guidance (tailored to user-selected categories), neutrals, and starter capsule. Optionally generates PDF swatch cards.