skills

Applies first-principles, multi-order consequence analysis, inversion, systems thinking, probability, antifragility, and high-energy thrive ascent (2036 lens) before material coding decisions (architecture, dependencies, public API, security, data contracts, hard-to-revert refactors). Use when choosing between approaches, planning cross-cutting changes, or when the user asks for decision analysis, tradeoffs, pre-mortem, second-order effects, or thrive vision.

Load and maintain the shellyxz shell kernel ontology graph — PATH contract, kernel/plugin boundary, load order, and verification bridge (ab/av/at). Use when editing ~/.config/shell core/, plugins/verification/, bin shims, moving files across the kernel boundary, or answering where a concept lives.

Authors evidence-driven architecture backlog docs (coming-next style): scorecards, mermaid diagrams, blueprint cards SN-*, thrive vision, gantt order. Use when creating or updating roadmaps, arch-design/coming-next.md, backlog after PRs, or when the user wants blueprint cards, Musk 5-step, or optimistic 10-year plans. Pairs with ai-optimization, fusion-sage, and higher-order-decision-architect.

Generates project-specific tmux verification cockpits for repos that use an av-style agent verify workflow. Reads target project AGENTS.md, setup docs, and scripts; writes .agents/verification/ with SOC-style panes and tiered auto-launch. Use when setting up av layouts, mission-control verify panes, or regenerating .agents/verification after stack changes.

Peramanathan Sathyamoorthy's senior ML/AI systems engineering principles — first principles thinking inspired by Feynman and Musk, Machine First architecture, strong mathematical and physics foundations, rapid practical learning through clean iteration, deep systems capability, clear language, extensibility, and uncompromising quality.

Overarching patterns for self-guarded, pause-aware, agent-driven desktop/web platforms. Combines finder-reactor, Tauri shell, X resources, and CV guard into a cohesive autonomous system with intervention only on guards. Use for high-level autonomy design, meta-improvement loops, MCP composability. Fusion for the reactor as living system; fission for specific guards.

Manifest V3 Chrome extension development — service worker lifecycle, message passing (popup ↔ service worker ↔ content script), content script injection, and host_permissions fetch patterns. Use when editing manifest.json, background/service worker, popup, content scripts, chrome.runtime.sendMessage, or extension permissions.

Strict self-guarded mechanism for reading an external CV source (e.g. cvdata.json in a portfolio repo) for grounding and safely promoting insights back with sidecars, diffs, previews, backups, and explicit user confirmation. Never mutates external profile without multiple gates. Use for CV-related logic in finder-reactor or prep flows. Sidecar-first, auditable, intervention points. Fission for safe edit code; fusion for protecting the public profile while improving future matches.

Readonly exploration of a C/C++ or CMake repo before MVU or structural changes. Use when user asks to explore and report structure, CMake support, or elomaxz integration points.

Core autonomous, self-guarded decision loop for opportunity-finder apps. Handles discovery (e.g. X search), analysis (LLM + CV + platform context), prep generation, tracking, and guarded promote with cost/rate/fit/CV mutation guards, human pause points, structured decisions, and logging. Use when designing, implementing, or debugging the agentic heart of a finder platform. Fission for tight loops; fusion for reactor architecture and surplus.

Plan and implement CMake + elomaxz MVU refactors in C codebases. Use when user references elomaxz, MVU, CMake refactor plan, or attached implementation plan.

Reorganize scattered C/C++ sources into src/ with CMake updates. Use when user asks to tidy layout, move files to src, or clean root-level clutter.

Run readonly subagent exploration of a CMake/C repo and return a structured report. Use when delegating explore/map before CMake or MVU work.

Patterns for building agentic, MCP-exposed, self-guarded desktop apps in Tauri (Rust backend + React/TS frontend). Expose finder functions as MCP tools, implement guards/pauses in UI and backend, command palette as agent interface, secure key storage, integration with X resources and CV guard. Use when implementing the Tauri shell, reactor UI, or MCP server. Fission for Rust/TS code; fusion for desktop as agent body.

Systematic Tauri IPC debugging — triage Intent Engine layers (MVU → safeInvoke → Rust commands → HTTP/sqlite/keyring), avoid false leads (browser Network tab, keyring noise). Use when invoke fails, search/cycle broken, bearer errors, tauri dev hangs, blank window, or user asks to debug Tauri/desktop shell issues.

Integration and usage of official X Developer Platform agent resources (llms.txt, skill.md, MCP/XMCP, xurl, OpenAPI) for accurate, composable X access in agentic apps. Use when building search, analysis, posting, or any X interaction. Ensures llms/skill ground prompts, MCP exposes tools, xurl provides CLI UX. Fission for implementation; fusion for unifying with finder-reactor and agentic design.

Triage every non-trivial task first: single-shot when safe, full orchestration when multi-step, multi-agent, or high-risk. Covers briefs, verification before "done", iterative waves, and resume. Use when coordinating workers, delegating, or reviewing agent output — not for one-file typo fixes.

Fission engine for token-efficient coding (JS, TS, Node.js, Rust, Python, ML/AI). Pruning, compression, strict token budgeting. Pair with fusion-sage for synthesis and surplus. Trigger on implementation, debugging, refactoring; hand off architecture to fusion-sage.

Audits pnpm allowBuilds whitelist packages for supply-chain risk: resolved lockfile versions, new registry releases, lifecycle scripts, trust downgrades, and incident intel. Runs checks and install verification. Use when the user asks about allowBuilds, postinstall/build scripts, strictDepBuilds, approve-builds, or whether native-build deps were compromised.

Audits IDE tooling supply chain: Cursor/VS Code extensions, Cursor plugins, and bundled node_modules using pnpm/npm audit where lockfiles exist. Use when the user asks about Cursor plugins, extensions, MCP helpers, IDE malware, or whether editor dependencies were compromised.

Generates minimal, security-first .devcontainer configs (devcontainer.json, optional Dockerfile) with pinned digests, non-root user, cap drops, and small blast radius. Use when the user asks for devcontainer, Dev Containers, Codespaces, remote development in Docker, or hardened containerized dev setup.

Fixes dependency vulnerabilities and deprecations, hardens the supply chain, and wraps package-manager installs with Socket Firewall (sfw). Use when the user asks for pnpm/npm audit, security fixes, deprecated packages, supply-chain attack prevention, StepSecurity-style dependency hygiene, or sfw.

Fusion-oriented evolution of Context Sage. Uses the original fission engine as containment field, then adds synthesis, surplus generation, and self-amplifying knowledge loops. For AI coding tasks where you want not just efficiency, but compounding intelligence.

Most sensible Graphite (gt) stacking decision and execution flow. Diagnoses coupling (overlapping file touches across planned slices), honestly recommends "gt is overkill here — use plain branches / split-to-prs / sequential waves / one PR" when gt would add complexity or risk (especially with coupled core files or agent worktree generation). When gt is chosen or the safe path is taken, enforces commit-in-generation-env + explicit two-phase handoff + backup refs + per-slice escape branches + state so generated code is never lost and recovery is trivial. Use on any gt/graphite/stack/submit --stack decision, execute-plan assembly, concurrent agent work, or manual PR planning.

Generates editor-agnostic workspace settings from a .editor/profile.json manifest: .vscode/settings.json, extensions.json, and Cursor workspaceOpen plugin loading. Use when setting up per-project extensions, Cursor plugins, editor defaults, or porting workspace config to new repos.

Guides modern HTML semantics, WAI-ARIA, WCAG contrast, data-* state attributes, and native elements over div-heavy UI; pairs Tailwind with modern CSS where CSS wins. Use before marketing/UI refactors, landing pages, forms, dialogs, nav, or when the user mentions semantics, WAI-ARIA, accessibility, data attributes, native elements, color contrast, or modern CSS.

Upgrades npm/pnpm dependencies safely: patch and minor first, framework majors (Next.js, React) with official or community codemods and code fixes. Use when the user asks to update, upgrade, or bump packages, reduce outdated deps, or migrate to a new major of Next, React, TypeScript, Tailwind, or Biome/ESLint.

Harvest Cursor agent transcripts into agent-prompt-tuning-lab: pnpm harvest:all, seed-manifest, env vars, subagents, and gold tagging. Use when collecting or re-indexing local Cursor JSONL for this repo.