rahulrajaram
- 50 skills
- 0 followers
- 9 hours ago last updated
- ▌ Test · rahulrajaramRun focused tests with available supervision or the native project runner, preserving output, timeouts and required coverage. Use when running test suites (npm test, pytest, playwright, etc.)
- ▌ Figma · rahulrajaram bundleRetrieve Figma design context, screenshots, assets, or setup guidance. Route production code implementation to figma-implement-design.
- ▌ Speak · rahulrajaram bundleRead text out loud using Kokoro TTS. Defaults to reading your last response.
- ▌ Commit · rahulrajaramCreate a well-formed commit after intelligently triaging every changed file to determine whether it should actually be committed. Filters planning artifacts, agent state, spurious docs, and secrets. Supports human-controlled overrides via .githooks/commit-allow. Use when the user says /commit, 'commit changes', 'make a commit', or asks to commit work.
- ▌ Debate · rahulrajaram bundleRun a structured multi-AI debate through gptengage. Use when the user asks for debate or multi-model deliberation, or says /debate or $debate.
- ▌ Ideate · rahulrajaram bundleGenerate a divergent idea tree from a seed through gptengage. Use for brainstorming, evolutionary ideation, or when the user says /ideate or $ideate.
- ▌ Handoff · rahulrajaram bundleWrite a verified NEXT_SHELL_PROMPT.md handoff, or activate an existing one. With extract, review the artifact, adopt it as the current request, and immediately execute its authorized work; with print, only emit it verbatim.
- ▌ Grill Me · rahulrajaramA shortcut to run the grilling skill on a plan, decision, or idea. Use when the user says /grill-me or $grill-me.
- ▌ Grilling · rahulrajaram bundleAsk dependency-aware questions about a plan, decision, or idea. Default to an interview with the user; select researched multi-agent answers, linear runtime, or a gradient lattice only when requested. Use for grilling, stress-testing assumptions, or /grilling and /grill-me.
- ▌ Humanize · rahulrajaram bundleRewrite analytical, technical, product, strategy, architecture, or agent-generated prose into clear, natural language for its intended human audience while preserving meaning, evidence, uncertainty, commitments, and important distinctions. Use after frame-goals-constraints or when the user asks to humanize text, make it readable, make it less agent-like or robotic, translate it into plain English, or turn a rigorous analysis into a customer-, executive-, or practitioner-facing narrative. Do not use to evade AI-detection systems or misrepresent authorship.
- ▌ Fp Refine · rahulrajaram bundleTransform imperative code toward explicit data flow, immutable state, typed outcomes, exhaustive dispatch, and declarative domain rules. Use for targeted FP-style refactoring; use clean-code-refine for broad clean-code review where FP may conflict with simplicity, idiom, function size, or complexity.
- ▌ Invokellm · rahulrajaram bundleInvoke one or more AI CLIs through gptengage. Use when the user asks to consult, query, or compare Claude, Codex, Gemini, or a plugin, or says /invokellm or $invokellm.
- ▌ Next Todos · rahulrajaramGenerate imperative next-step to-do lists as full sentences with clear objectives.
- ▌ Postmortem · rahulrajaram bundleGenerate Amazon COE-style 5-whys postmortem reports for incidents and failures. Use when user says "postmortem", "COE", "5-whys", "incident report", or asks to analyze why something failed.
- ▌ Privateify · rahulrajaramAdd reviewed safeguards against accidental public visibility and package publication. Adds multi-layer safeguards: CI visibility-guard workflow, pre-push hook check, publish=false in package manifests, and agent-facing CLAUDE.md directives. Use when user says /privateify, 'make this repo private forever', 'lock visibility', 'prevent publishing', or asks to ensure a repo is never made public.
- ▌ Whitepaper · rahulrajaram bundleAuthor a whitepaper, business case, investor pitch, or product value proposition for the specified reader. Use for original product/value authorship; financial modeling, naming, branding, and PDF delivery apply when requested. Preserve evidence, assumptions, and the distinction between recommendations and approved commitments.
- ▌ Archdiagram · rahulrajaram bundleGenerate an architecture diagram from the current context or codebase. Use when the user asks for an architecture diagram, system diagram, component diagram, or says /archdiagram. Supports --type for different diagram kinds, --format for output format (ascii default, mermaid, plantuml, d2, dot), and --deep-review for holistic architecture analysis.
- ▌ Max Columns · rahulrajaramKeep output within a user-specified column width. Use when the user asks for N-column, 72-column, 80-column, wrapped, terminal-width, or narrow output.
- ▌ Metabuilder · rahulrajaramUse MetaBuilder to design, agree, compile, inspect, run, qualify, recover, or improve a governed engineering harness. This is the global entry point for building a new harness in an arbitrary target repository and routes fresh design through metabuilder-harness-design before metabuilder-consumer-qualification.
- ▌ Repo Topics · rahulrajaramAnalyze a GitHub repository and apply relevant topic labels under repository details. Use when user says /repo-topics, 'add topics', 'add labels to repo', 'tag the repo', or asks to set GitHub repository topics.
- ▌ Pr Lifecycle · rahulrajaram bundlePrepare and open a GitHub pull request and follow required CI through success. Use for an end-to-end PR workflow; select commit, history cleanup, documentation and release steps only when needed and authorized.
- ▌ Autonomy Loop · rahulrajaram bundleDrive an epic as a principal-architect loop: rank tasks, execute bounded slices, verify, checkpoint, and optionally chain safely.
- ▌ Clear Writing · rahulrajaram bundleTurn dense, awkward, repetitive, or AI-generated writing into clear, direct, readable prose for a broad audience. Act as an editor, not a ghostwriter: preserve the author's meaning, technical precision, structure, examples, and voice. Preserve meaning, precision, logical validity, and voice before improving flow, simplifying language, and compressing. Use when the user asks to edit, tighten, clean up, clarify, or rewrite a draft for readability, or to remove AI-style repetition, theatrical language, throat-clearing, weasel words, abstraction, and manufactured transitions in any prose they write. Supports default mode (edit prose) and grill mode (edit plus flag evidence, logic, assumption, and verification gaps with structured placeholders). Do not add facts, invent citations, or strengthen writing beyond what the draft supports.
- ▌ Readme Doctor · rahulrajaram bundleCorrect or audit README and CLI help against actual project behavior. Use for narrow documentation/help fixes or an explicitly requested full documentation audit; preserve the requested scope.
- ▌ Skill Creator · rahulrajaram bundleCreate or update a Codex skill with appropriately scoped instructions and any needed supporting resources.
- ▌ Squash Commits · rahulrajaram bundleAnalyze Git history and safely squash contiguous thematic commits with recoverable backups and clean conventional messages. Use when the user asks to squash, consolidate, compress, or tidy commit history.
- ▌ Markdown To PDF · rahulrajaram bundleConvert markdown to PDF via pandoc + weasyprint. Use when the user asks to generate a PDF from markdown or says /markdown-to-pdf.
- ▌ Clean Code Refine · rahulrajaram bundleReview or refactor code across multiple clean-code dimensions: behavior preservation, language idiom, function size, complexity, cohesion, FP/dataflow opportunities, testability, and simplicity. Use for clean-code review/refactoring or when fp-refine may conflict with broader readability.
- ▌ Git Status Report · rahulrajaramReport local git working-tree and tracking-ref status for the current repository and submodules. Refresh remotes only within existing explicit authorization.
- ▌ Reference Cleaner · rahulrajaram bundleRemove references to blocklisted terms from git history and source files. Use when user says /reference-cleaner, 'clean references', 'remove mentions of X', 'scrub project names from history', or asks to sanitize a repo before publishing.
- ▌ Check Antipatterns · rahulrajaram bundleCheck an in-progress conversation and active code changes for anti-patterns, evidence-backed review findings, and immediate course corrections. Use for live checks, code review during active work, before risky operations, or when work feels stuck. Do not use for completed-session postmortems; use analyze-conversation.
- ▌ Install Commithooks · rahulrajaram bundleInstall shared commithooks framework into a project. Copies dispatchers and lib into .git/, scaffolds .githooks/ stubs, and wires hook installation into the project's dev setup path. Use when user says /install-commithooks, 'install hooks', 'setup git hooks', 'add commit hooks', or asks to wire up shared commithooks.
- ▌ Memleak Investigate · rahulrajaramInvestigate longitudinal memory growth in a specific Linux process using /proc and, with approval, attach/tracing tools. For whole-system OOM, swap, or tuning without a target process, use system-memory-audit first.
- ▌ Pythonpackagesevere · rahulrajaram bundleDecompose a Python package into independent projects. Use when the user says 'pythonpackagesevere', 'split package', 'decompose python package', or asks to break a monolith into separate projects.
- ▌ System Memory Audit · rahulrajaramAudit Linux system-wide memory health, swap activity, PSI, kernel tunables, and top consumers. Use for system memory health or tuning; for one named process with longitudinal growth, use memleak-investigate.
- ▌ Analyze Conversation · rahulrajaram bundleAnalyze a completed conversation retrospectively for anti-patterns, tooling gaps, and durable learnings, then generate a markdown report. Use for postmortems of finished sessions or when the user explicitly says /analyze-conversation. Do not use for live, in-progress checks; use check-antipatterns instead.
- ▌ Pi Defects Harvester · rahulrajaramScavenge local shell history and pi/agent artifacts at the end of an interactive pi session, extracting actionable defect signals into a short redacted markdown digest. Use to capture errors, repeated tool failures, truncations, cost spikes, or unfinished shells without manually opening every history file. Source-preserving; writes one digest under ~/.pi/agent/reports/.
- ▌ Diagram Review Viewer · rahulrajaram bundleCreate hand-authored Mermaid diagrams with an interactive HTML review viewer (pan/zoom viewport, zoom toolbar, resizable info rail, digest badge) for operator review. Use when asked for an architecture, state-transition, or design diagram meant for human review in a browser, or when asked to package a diagram as a reviewable artifact. Not for throwaway ASCII sketches or images embedded in docs.
- ▌ Ecosystem Borrow Audit · rahulrajaram bundleAudit a workspace listing and all depth-1 git repos for borrowable components, missed architectural opportunities, and integration gaps; then run independent gptengage ideate sweeps across multiple sigma values. Use when the user asks for ecosystem review, cross-repo borrowing analysis, missed-opportunity checks, or multi-sigma ideation after repo analysis.
- ▌ Figma Implement Design · rahulrajaram bundleTranslate Figma nodes into production-ready code with 1:1 visual fidelity using the Figma MCP workflow (design context, screenshots, assets, and project-convention translation). Trigger when the user provides Figma URLs or node IDs, or asks to implement designs or components that must match Figma specs. Requires a working Figma MCP server connection.
- ▌ Frame Goals Constraints · rahulrajaram bundleFrame a complex product, platform, or system direction into a living product thesis before solutioning. Work backward from the future change and customer job; connect an ambitious design horizon to a credible first wedge; separate facts, bets, preferences, constraints, actors, trust boundaries, tradeoffs, and unknowns; and produce aligned customer, product, and technical language with measurable success, stop, and revision rules. Use when the user asks for the big picture, north star, product worldview, positioning, strategic direction, goals and constraints, an approachable expression of a technical vision, or a shared operating model before planning or implementation.
- ▌ Rewrite Commit Messages · rahulrajaramUse when the user wants to rewrite existing git commit messages with git filter-repo, rename commit messages in bulk, normalize noisy auto-generated messages, or safely plan a history-wide message-only rewrite.
- ▌ Yore Vocabulary Harvest · rahulrajaramExtract candidate vocabulary terms from a Yore index for stopword curation and domain filtering.
- ▌ Define Operating Charter · rahulrajaram bundleDefine and ratify a durable operating charter for a long-running agentic system by separating goals, tenets, environment, actors, authority, delegation, constraints, unknowns, evidence, lifecycle, recovery, success, and stop conditions. Use when the user wants to establish how an orchestrator, supervisor, harness, or worker system should operate before autonomous execution, or asks to define an execution envelope or operating charter.
- ▌ Metabuilder Harness Design · rahulrajaram bundleDesign a new MetaBuilder harness from repository evidence and a user objective through thorough grilling, an agreed brief, and typed intent/module/workflow commitments. Use before fresh harness construction; use consumer qualification later to run and assess an already designed harness.
- ▌ Yore Vocabulary LLM Filter · rahulrajaram bundleClassify harvested Yore terms for Whisper vocabulary using a selected gptengage backend, validate strict JSON, and present a human-gated merge preview.
- ▌ Tui Web Design Orchestrator · rahulrajaram bundleGenerate structured design prompt packets for terminal UIs and web UIs from natural-language briefs. Use when the user requests a design prompt/packet for a TUI or web UI, or explicitly invokes this skill. A UI implementation request does not by itself select prompt-packet generation.
- ▌ Autonomous Execution Contract · rahulrajaramExecute agreed long-running engineering work autonomously from a bounded objective, using explicit stop rules, verification targets, and low-reasoning defaults.
- ▌ Objective To Dag Decomposition · rahulrajaramDecompose vague or complex objectives into a MECE issue tree, typed reasoning nodes, an execution DAG, and a phased implementation plan with verification.
- ▌ Metabuilder Consumer Qualification · rahulrajaramRun and assess an already designed Linux-local consumer harness with MetaBuilder while keeping controller evidence separate from consumer semantic judgment. Use after harness design when a target project is testing whether the resulting harness is suitable.