Master Console
This repository also configures its own agents. See what master-console tells them →
Command Claude Code
Establish and verify an epic's linkages — its companion plan artifact, ROADMAP § Epics entry, TRACKER header, child sprints, and dependency edges to sibling epics — repairing the round-trip pointers. (Designed; build deferred.).
Command Claude Code
Turn a messy plain-language epic request into a durable, validate-shape epic tracker file plus a companion plan artifact (AC, sprint candidates, dependency/risk maps, tracker linkage). The epic equivalent of /sprint:plan.
Command Claude Code
Detect and repair a drifted or malformed epic file — missing §-sections, blank required sections, broken links, percent/state inconsistencies, and tracker/roadmap reconciliation drift — fail-closed, additive, with a repair report. (Designed; build deferred.).
Command Claude Code
Run an independent, cross-provider review of an epic plan — feasibility, overclaims, missing enforcers, blast-radius gaps, sequencing — and record the verdict + findings against the epic. (Designed; build deferred.).
Command Claude Code
Split an over-large epic into two or more coherent epics — partition scope/sprints/DoD/AC, preserve provenance and dependencies, and create the child epic files with cross-links. (Designed; build deferred.).
Command Claude Code
Transition a planned epic into active execution — mint its first wave of sprints, set state to Active, and stand up the required mode/team. (Designed; build deferred.).
Command Claude Code
Report an epic's true, evidence-based status — percent completion, sprint roll-up, DoD progress, blockers, and tracker/roadmap reconciliation drift. Read-only. (Designed; build deferred.).
Command Claude Code
Author or refine a skill/prompt in standard format, producing a sibling eval-pack for evaluation.
Command Claude Code
Evaluate a skill or prompt artifact against its eval-pack, emitting a validated decisionrecord.
Command Claude Code
Query the events log by type, time range, or regex match.
Command Claude Code
Tail the events log — last N events with timestamp, type, and message.
Command Claude Code
Search saved favorite moments by keyword across category, title, and notes — find one specific moment you remember saving but cannot locate by category alone.
Command Claude Code
Deep fix — Full diagnostic with automatic framework selection, 5 solutions, root cause analysis, and prevention.
Command Claude Code
Quick fix — Direct Investigation, no formal framework. Read error, find cause, fix it, verify.
Command Claude Code
Turn an angle into native-ad image prompts (scene-first, no text/logo/product, --ar) and render them via Higgsfield (headless API; gated by higgsfield-spend-gate — <$5 autonomous, ≥$5 escalates).
Command Claude Code
Turn an angle into a video ad (swipe→script→storyboard→image-to-video) and generate it via Higgsfield (headless API; gated by higgsfield-spend-gate). Video assembly/editing is out of scope.
Command Claude Code
Write a long-form advertorial (pre-sell editorial) from a message brief — research → foundational docs → swipe → write → Chief coherence review.
Command Claude Code
Mine untapped marketing angles from real customer voice (Amazon/Reddit/forums) — ≥3 evidence-backed alternates. Reuses research:deep.
Command Claude Code
Iterate a winning creative/message against a conversion/engagement scalar — thin wrapper over karpathy:run + parallel variation fan-out. Data picks winners.
Command Claude Code
Build a converting landing page from a conversion brief — conversion-hierarchy, scaffold component library, mobile-first. Reuses the content render path + scaffold.
Command Claude Code
Distill the single winning message (the spine artifact) from an audience dossier + angles — contrast + depth, market promise inside the claims boundary.
Command Claude Code
Find validated high-margin products for paid traffic — EQ-scored (Product×Ads×Funnel×LTV), SCALE/TEST/SKIP, with margin math. Reuses research:deep.
Command Claude Code
Fail-closed enforcer for the guides/ library — asserts every guide is anchored, the registry is fresh, every anchor is wired live into the bootstrap pipeline, and there are no orphan records or markers. The /guides analog of /scan:scan-coverage and /maps:coverage.
Command Claude Code
Audit and restructure the guides/ launch-guide library — backfill the guide-anchor contract onto every guide, (re)generate guides/registry.json, and flag structural drift — WITHOUT rewriting guide prose (that's /guides:write). The skills:cleanup analog for guides.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: