zhixuan312
Community@zhixuan312
zhixuan312 maintains the ZZ platform skill suite: gated SDLC delivery flows, plugin evaluation pipelines, and multi-model delegation for Claude Code runtimes.
Agent Skills by zhixuan312
Showing 36 vetted skills indexed across 2 GitHub repositories.
zhixuan312zz-plugin-eval
Evaluates installed plugins through a five-stage locate, profile, define, judge, and report workflow.
zhixuan312zz-plugin-profile
Compute evidence blocks and conformance verdicts from plugin run traces without model calls.
zhixuan312zz-plugin-locate
Resolves a plugin's released version, content digest, skill versions, and declared tools for evaluation.
zhixuan312zz-plugin-define
Derives plugin-specific evaluation rubrics and gates scoring on human approval.
zhixuan312zz-migrate
Migrate an mma repository's journal and historical documents into the ZZ platform as knowledge nodes and sources.
zhixuan312zz-update
Updates all installed ZZ plugins by refreshing the marketplace and reporting version changes.
zhixuan312zz-access
Issues, replaces, and revokes personal platform access tokens and configures client setup.
zhixuan312zz-admin
Administer principals, teams, tokens, and enrolment links on the zz-stack platform.
zhixuan312sdlc-audit-criteria
Applies eleven prose failure modes to audit SDLC documents and returns structured JSON findings.
zhixuan312sdlc-spec
Facilitate decision interviews and write approved specification documents for delivery initiatives.
zhixuan312sdlc-plan-audit
Audits plan.md against prose failure modes and the plan's execution contract.
zhixuan312sdlc-research
Answers one external research question with cited sources across five perspectives.
zhixuan312sdlc-explore
Grounds raw ideas by dispatching parallel investigation, research, and recall workers into one explore.md.
zhixuan312sdlc-review
Review a finished change against ten release-safety failure modes and emit cited findings as JSON.
zhixuan312sdlc-flow
Routes software delivery initiatives through explore, spec, audit, plan, execute, review, and close stages.
zhixuan312sdlc-spec-audit
Audit spec.md against eleven prose failure modes and a seven-point specification contract.
zhixuan312sdlc-recall
Search the ZZ knowledge base and synthesize prior team decisions into a plain-English briefing.
zhixuan312sdlc-execute
Implements an approved plan by dispatching one subagent per task in plan order.
zhixuan312zz-router
Routes ZZ platform delivery requests to the correct installed flow and loads it.
zhixuan312zz-platform
Governs initiative and document lifecycle rules for every flow on the ZZ platform.
zhixuan312zz-handover
Converts a closed initiative's documents and telemetry into gated knowledge nodes and a handover document.
zhixuan312zz-deck
Convert existing prose documents into argument-driven HTML slide decks on a fixed visual system.
zhixuan312mma-research
Compile cited research sources from arxiv, semantic_scholar, github_search, and brave-search into a bibliographic report.
zhixuan312mma-context-blocks
Register large documents once and reference them by ID in mma-* calls.
Frequently Asked Questions About zhixuan312
FAQPage SchemaWhat tasks can I accomplish with zhixuan312's ZZ platform skills?▼
You can run gated software delivery (explore, spec, plan, execute, review), evaluate plugins against pre-agreed rulers, delegate codebase investigations and code reviews to workers, migrate mma repository history into knowledge nodes, and manage platform tokens, teams, and principals.
Who are these skills designed for?▼
Engineering teams running Claude Code locally who need structured delivery flows with human approval gates, platform administrators managing principals, teams, and PATs, and reviewers who want evidence-based plugin grading rather than ad-hoc opinions.
How do I install and run these skills?▼
Skills run inside local Claude Code runtimes as installed plugins. Use zz-update to refresh the marketplace and update installed plugins, zz-doctor to verify token reachability and plugin integrity, and invoke entry points like sdlc-flow, zz-plugin-eval, or zz-router.
What prerequisites do the SDLC skills require?▼
SDLC stage skills require a runtime that can dispatch subagents and reach the working tree directly. The flow is sequential: explore grounds the idea, spec and plan each require human approval and pass audits before execution, and zz-handover closes every initiative.
How does plugin evaluation prevent biased judgements?▼
zz-plugin-eval enforces five stages: locate pins the plugin version, profile computes run traces, define writes rulers.md that a person must approve before scoring, judge scores with a pinned external model, and report writes gated findings.md.