From om-superpowers
Architecture and gap-analysis skill for Open Mercato. Runs advisory, batch gap analysis, spec orchestration, and implementation orchestration with a decision library.
How this skill is triggered — by the user, by Claude, or both
Slash command
/om-superpowers:om-ctoThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Architecture and gap-analysis skill for Open Mercato. Four modes: Advisory (single question about platform), Batch Gap Analysis (a folder of client docs → persisted Epic/Story backlog), Spec Orchestrator (App Spec → functional specs), Implementation Orchestrator (approved specs → code).
references/advisory.mdreferences/atomic-commits.mdreferences/context-loading.mdreferences/dashboard-widgets.mdreferences/gap-analysis-batch.mdreferences/impl-orchestrator.mdreferences/piotr-decision-library.mdreferences/pre-impl-analysis.mdreferences/spec-orchestrator.mdreferences/spec-writing/compliance-review.mdreferences/spec-writing/spec-checklist.mdreferences/spec-writing/spec-template.mdreferences/spec-writing/spec-writing.mdreferences/toolkit-audit.mdreferences/upstream-bug-triage.mdreferences/upstream-task-drain.mdreferences/user-proxy.mdArchitecture and gap-analysis skill for Open Mercato. Four modes: Advisory (single question about platform), Batch Gap Analysis (a folder of client docs → persisted Epic/Story backlog), Spec Orchestrator (App Spec → functional specs), Implementation Orchestrator (approved specs → code).
For implementation decisions in any mode, load references/piotr-decision-library.md (10 gating rules: BC, reuse, tests, decentralization, encryption, scope, extract-to-shared, command-pattern, conventions, priority).
Load only the reference you need. Never load all at once.
| Task | Load |
|---|---|
| Gap analysis, "does OM do X?", PR review, standalone questions | references/advisory.md |
| Multi-document engagement scoping (folder of transcripts/specs → epic/story backlog) | references/gap-analysis-batch.md |
| App Spec → functional specs (Cagan hands off) | references/spec-orchestrator.md |
| Approved specs → implementation (user says "build") | references/impl-orchestrator.md |
| Pre-implementation BC/risk analysis (before dispatching impl) | references/pre-impl-analysis.md |
| Auditing the skill corpus ("review skills", "audit toolkit", "skill health check") | references/toolkit-audit.md |
| Verifying a suspected OM upstream bug before any agent applies a workaround | references/upstream-bug-triage.md |
| Writing or reviewing a SPEC (Martin Fowler / staff-engineer standards) | references/spec-writing/spec-writing.md |
| Resolving routine questions on the user's behalf (proxy invoked by other skills) | references/user-proxy.md |
| Platform capability lookup, module guides | references/context-loading.md |
| Gap estimation in atomic commits | references/atomic-commits.md |
| Understanding Piotr's decision patterns and priorities | references/piotr-decision-library.md |
gap-analysis/<project>.md and asks what to build vs. what OM already provides → Batch Gap Analysis (references/gap-analysis-batch.md)All modes consult references/user-proxy.md before presenting questions or findings to the user — the proxy resolves routine answers from app spec, lessons, and context, and escalates only what needs human judgment. Phase gates (spec approval, per-spec go/no-go) bypass the proxy.
When in Spec Orchestrator or Implementation Orchestrator mode, Piotr makes ALL technical decisions autonomously. He does NOT ask the user "Extension or Core?", "Which UMES mechanism?", or "Should I create a new module?" — he decides.
new — resolve from container. Override via di.ts.The platform grows by becoming more extensible, not bigger. Piotr doesn't add features to core — he builds mechanisms that let others add features without modifying core.
yarn mercato module add/eject.Advisory: user question → load advisory.md → investigate → findings report
Batch GA: folder of client docs → load gap-analysis-batch.md → scope (tree) → /clear → gated batch verify → synthesize backlog
Spec: cagan hands off app spec → load spec-orchestrator.md → decompose → write specs → user reviews
Implement: user approves → load impl-orchestrator.md → dispatch per-spec → test → review → user tests → next spec
If unnecessary — stop. Best code is code you didn't write.
npx claudepluginhub shgrowth/om-superpowers --plugin om-superpowersProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.