Continuous-Claude-v3

The Code reporting discipline AND the reporting-system operations command. Discipline half — the "one spine" rule (every reportable unit of work lands one Weekly Work Tracker row, Source-tagged), field mapping, Bridge auto-mirror tie-in. Operations half (proposal 07) — /reporting status (registry tail, trust metrics, drift, next scheduled runs, log locations), /reporting run <pipeline>, /reporting heal. Use when finishing reportable work, when asked "how should I report this" / "log this for the report", at session end to capture unreported work, or for pipeline ops — "/reporting", "reporting status", "is reporting healthy", "run the health check", "why is a report row missing/stale".

Generate the weekly FourthOS sponsor update package for Carly (VP) and Christian (CTO). Pulls live portfolio data from the FourthOS Notion cockpit, refreshes the Notion Update Package page, renders a progressive-disclosure HTML artifact set (Tier 1 briefing dashboard, Tier 2 teaching deep-dive), stages it UNLISTED to the ai-enablement-decks GitHub Pages site, and notifies Dave to review before promoting it live. Use when asked to generate, build, preview, or promote the FourthOS weekly sponsor update, or when the CCv3-FourthOS-Weekly scheduled task runs. Triggers on "fourthos weekly", "sponsor update", "Carly update", "Christian update", "weekly portfolio update".

Run a feature through the full tri-model Game Plan pipeline — rostered roles (Claude hub, Grok builder/researcher, Codex reviewer/fixer), a workroom disk bus, and human gates. Use when the user types /game-plan [feature], says "run the game plan on X", "full crew on this", "roster this feature", or wants a multi-milestone feature built with cross-model build/review separation. Opt-in orchestrator on top of /workroom; for one-shot tasks use /grok or /codex directly, for trivial edits use neither.

Audit and curate the memory system -- identify signal vs noise, score entries, and archive low-quality learnings.

Detect drift between the continuous-claude repo and the active ~/.claude/ directory

Initialize Continuous Claude v3 for a new project with full toolset activation. Creates project CLAUDE.md (interview-driven, 8 sections), ROADMAP.md, knowledge tree, Serena code intelligence, and project registry entry. Use when opening a new project folder for the first time, starting a new project, setting up CCv3 in an existing repo, or the user says "init project", "setup project", "new project", "initialize", "start new project".

Query and manage the centralized project registry for port assignments, stack info, dev commands, and project status

Helps a salesperson turn a repeated piece of their work into a reusable Skill, by asking a few questions and writing the skill for them. Use this whenever someone says "turn this into a skill", "make a skill for [task]", "can I save this workflow", "I do this every week", "build me a reusable [research/email/prep] skill", or wants Claude to do a recurring sales task the same way every time — even if they don't say the word "skill".

Create handoff document for transferring work to another session

Enterprise-tier account-planning orchestrator + doc-writer for Fourth sales reps. A compressed 3-gate phase machine that fuses Salesforce data with account-research web findings and produces the 8-section living account-plan doc. Use when an Enterprise or Emerging-advanced rep is building a full account plan for a hospitality prospect — or says "build an account plan for [prospect]", "plan this account", "turn my research into an account plan", "run the Cockpit on [account]", or works the Account Cockpit breakout. Delegates all web research to the account-research skill; never re-derives research logic.

The workhorse account-research skill for Fourth sales reps. Turns a hospitality prospect name plus minimal context into a structured research brief using Claude's built-in web search. Use when a rep is preparing to prospect, qualify, or plan an account — or asks to "research this company", "build a research brief on [prospect]", "who do I call at [company]", "what are they using today", "give me a cold-call opener for [account]", or any request for outside-in business context on a hospitality prospect. Installed in every tier; every downstream artifact clones its output shape.

Identify failure modes before they occur using structured risk analysis, with a cross-model adversarial pass (Codex, Grok, or both)

Comprehensive code review workflow - parallel specialized reviews → synthesis (includes cross-model Codex adversarial pass by default)

Root cause analysis before fixes - NO fixes without investigation

Hand a task to the OpenAI Codex harness (gpt-5.5, on the ChatGPT subscription — no API key) to WORK THROUGH — answer it, or actually implement it with write access — not just review. Cross-model execution, complementary to Claude. Use when the user types /codex, says "hand this to Codex", "have Codex implement/do X", "delegate to Codex", "codex, work through this", or wants GPT-5.5 to execute a well-specified task. For adversarial code/plan REVIEW use /review or /premortem (those already call Codex read-only); this skill is for getting work DONE.

Hand a task to the xAI Grok Build CLI (grok-4.5, on the X Premium+ subscription — no API key) to WORK THROUGH — answer it, or actually implement it with write access — not just review. Cross-model execution, a third training family distinct from both Claude and GPT-5.5/Codex. Use when the user types /grok, says "hand this to Grok", "have Grok implement/audit X", "delegate to Grok", "grok, work through this", or wants Grok to execute a well-specified task. For adversarial code/plan REVIEW use /review or /premortem (those can already call Codex read-only; this skill's --review flag adds Grok); use this skill directly for getting work DONE.

Safely update a cross-model harness (Codex, Grok, or a future one) — upgrade its CLI, re-verify the capability surface, live-probe new model ids, and propagate the verified allowlist to every registered edit point. Use when the user says "/harness-update", "update codex", "update grok", "enable model X", "new codex/grok version", "add a model to the allowlist", or when a worker run errors with "requires a newer version". Never add a model id from public docs without a live probe on this account.

Manage tri-model Game Plan workrooms — the disk bus for multi-model feature work (Claude hub, Grok builder, Codex reviewer/fixer). Use when the user types /workroom (new/status/post/advance/resume), says "open a workroom", "room status", "advance the room", "post to the room", or when resuming multi-model feature work from CONTRACT.md + status.json. Opt-in: rooms are created deliberately, never auto-created. Protocol source of truth: .workroom/PROTOCOL.md.

Interactive workspace discovery - learn what tools, workflows, agents, and hooks are available

Notion CLI (ntn) operator reference — direct API requests, Markdown page I/O, data-source queries, file uploads, and the parked Workers charter. Use when scripting Notion reads/writes from Bash or PowerShell, refitting scheduled jobs away from claude -p + MCP, publishing HTML attachments via embed blocks, querying or upserting Notion databases from the command line, or when the user mentions "ntn", "notion cli", "notion workers", or "notion api from terminal". For semantic search, page-section edits with selection strings, views/DB creation, or Bridge operations, prefer the Notion MCP (see decision table).

This skill governs the three-layer Bridge communication system between Claude.ai (Eve), Claude.ai (Donna), and Claude Code for Dave Hayes. Use when reading or writing to the Bridge HQ page, updating handoff queues between AI instances, logging decisions or implementation notes, syncing sprint state, or performing any operation on the Bridge Life Bucket, its Projects, or Tasks. Triggers on phrases like "update the bridge", "handoff to Code", "what did Code build", "bridge sync", "log this to the bridge", "pass this to Code", or "what's in the queue".

Build and maintain machine-refreshed Notion dashboard pages that combine interactive HTML embeds, native DB views, and Notion-AI-readable markdown sections. Use when creating a new auto-refreshed Notion page/dashboard, wiring a page into the project-cards sweep, refreshing the FourthOS Mobile Cockpit, or when the user says "notion dashboard", "mobile cockpit", "refresh mobile cockpit", or "machine-maintained page". Codifies the two-phase publish pattern (deterministic ntn vs MCP-only embed bind) with its 200KiB/1-hour constraints and section-ownership contract.

Log a row to the AI Enablement Weekly Work Tracker (Notion) when work doesn't go through a formal Bridge writeback. Complements notion-bridge auto-mirror. Use when the user says "log work", "log this ship", "add to tracker", "tracker entry", "/log-work", or describes a design decision, meeting, research spike, or small fix worth recording for the weekly report.

This skill governs the three-layer Bridge communication system between Claude.ai (Eve), Claude.ai (Donna), and Claude Code for Dave Hayes. Use when reading or writing to the Bridge HQ page, updating handoff queues between AI instances, logging decisions or implementation notes, syncing sprint state, or performing any operation on the Bridge Life Bucket, its Projects, or Tasks. Triggers on phrases like "update the bridge", "handoff to Code", "what did Code build", "bridge sync", "log this to the bridge", "pass this to Code", or "what's in the queue".

Refresh and publish a living HTML status card for a FourthOS project. Zero-LLM deterministic assembler (scripts/project-cards/) emits a publish manifest; this skill does the one Notion MCP embed-bind step. Use when the user says "/project-card", "refresh card", "living status card", "project card", or after reportable FourthOS work on a project.

Unified code-intelligence facade. Routes a query to the right L3 specialist (codegraph/Serena/TLDR/ast-grep/archival memory/context bus) per the boundary contract, degrading honestly when a specialist is MCP-only or absent. Use when you want "who calls X", "find symbol X", "control/data flow in X", "have we solved this before", an ast-grep rewrite preview, or to read the session context bus -- and you want one typed entry point instead of remembering which tool owns which query. Triggers on "/code-intel", "code intel", "who calls", "find symbol", "code context", "rename preview", "read the bus".

Token-efficient code analysis via 5-layer stack (AST, Call Graph, CFG, DFG, PDG). 95% token savings.

Create, edit, and export live Excalidraw diagrams using mcp-excalidraw-server (MCP tools + canvas REST API). Use when an agent needs to draw/lay out diagrams, convert Mermaid to Excalidraw, query/update/delete elements, or export/import elements from a running canvas server (EXPRESS_SERVER_URL, default http://localhost:3100).

Vet external skills installed from skills.sh through structural validation, trigger accuracy evaluation, optimization, and automatic registration into skill-rules.json. Use when a new skill is installed via npx skills add, when re-vetting an existing skill, or when bulk-vetting all unvetted skills.

Canonical reference for querying and storing learnings in the persistent memory system. Use when users request recall of prior work, want to store a learning, ask "what did we do before with X", or when starting implementation similar to past sessions. Covers CLI commands, backend architecture, score interpretation, quality bar, learning types, scope tagging, and when sub-agents should recall. Consolidates content previously split across dynamic-recall, proactive-learning, memory-usage-guidelines, and agent-memory-recall rules.

Expert guidance for creating Fourth-branded content across all formats (excluding PowerPoint). Covers brand voice, messaging, visual identity, Fourth iQ branding, tone principles, color systems, typography, and photography for marketing materials, documents, web content, and all communications.

Produce a structured coverage/quality audit report (with CSVs, named users, callouts, and Rally-story-ready dev gaps) on any iQ360 event for any customer. Use whenever the user asks for "metrics", "an audit", "coverage", "EBR/QBR numbers", "what's broken with event X", or "who's doing X" on iQ360 event data — even if they don't say the word "audit". Differs from `iq360-events`: that skill answers single ad-hoc KQL questions; this skill is for multi-step analyses that end in a report. If the user just wants a single count or one-liner query, defer to `iq360-events` instead.

Report hit rate, score distribution, and quality of the proactive memory-recall surface. Use when you want to audit how well memory-awareness is performing — what intents match, what doesn't, what scores typically look like. Reads .claude/logs/memory-recall.jsonl.

Mark a memory entry as superseded by a newer entry. Use when a previously-correct learning is replaced (e.g., the API changed, the architecture was refactored, the failed approach is now the right one). The old entry stays in archival_memory for audit but is excluded from default recall (its valid_until is set to NOW()).

Query optimization and migration best practices

Microsoft Kusto.Cli for KQL queries against Azure Data Explorer and Microsoft Fabric Eventhouse — connection strings, headless execution, CSV export

Query and update project knowledge tree for navigation

Meta-skill workflow orchestrator for bug investigation and resolution. Routes to debug, implement, test, and commit based on scope.

Full development cycle browser automation for viewing, debugging, testing, and visual inspection of web apps. Five-tier strategy: @playwright/mcp (interaction), @playwright/cli (AI agent disk-based), CDP CLI (`scripts/cdp.mjs`) (performance), Playwright-core (scripting), and Playwright CLI (e2e test suites). Use when users need to browse, take screenshots, debug network/performance, run e2e tests, record tests with codegen, or do visual QA. Triggers on "browser", "screenshot", "viewport", "performance trace", "network debug", "visual QA", "responsive test", "e2e test", "playwright test", "codegen", "test suite".

Show full session token usage, costs, TLDR savings, and hook activity