Skills de Claude Code · página 69

Compare two or more existing candidates under an explicit characteristic space, parity basis, and predeclared selection policy. Return trade-offs and a non-dominated set; persistence is conditional and a binding choice requires a direct unambiguous operator request.

Route one direct, unambiguous operator request for a bounded binding choice. A manual h-decide invocation is a compatible shortcut, not an authorization receipt.

Diagnose a concrete failure with rival-hypothesis testing. Stabilize the symptom, generate distinct explanations, test discriminating observations read-only in parallel, and rank by evidence while retaining losing rivals.

Generate 3-5 genuinely distinct candidate approaches for a current question, with the weakest link of each visible. May work from an inline question or a durable ProblemCard; persistence is conditional.

Shape an under-articulated engineering problem without assuming a solution or forcing a project phase. Default to a conversational frame; create a ProblemCard only on explicit save intent or when current Work supplies a concrete operator-named or agent-inferred receiving use that needs a durable accepted problem statement.

Persist an explicitly requested non-binding fact, observation, caveat, or small rationale in Haft project memory. Do not auto-persist ordinary reasoning.

Bootstrap Haft through the readable task-level onboarding surface, prepare a non-binding project-profile review when needed, and orient only applicable typed spec carriers. Project memory is ready immediately after haft init; profile apply and lifecycle gates remain human.

Source-first umbrella for FPF-aware reasoning in a Haft project. Use for ambiguous engineering, management, architecture, specification, or project questions when no narrower Haft capability is already current. Ordinary reasoning stays conversational; persistence is conditional and binding actions remain manual.

|

|

Read-only Haft project cockpit for active problems, decisions, notes, evidence freshness, drift, commissions, spec lifecycle, module coverage, and bounded exact file-link gaps from a current code index. Use for project status, session resumption, what is decision-linked, what is uncovered, or what needs attention.

Verify that a recorded DecisionRecord or claim still holds by comparing its baseline and predictions with current code, tests, measurements, or incidents. Keep design-time claims distinct from runtime evidence.

Manage Haft's typed spec lifecycle, source-currentness, carrier edits, and semantic fanout repair. Treat markdown as a carrier and kernel lifecycle plus explicit human gates as authority.

The anti-PUA. Drives AI with wisdom, trust, and inner motivation instead of fear and threats. Activates on: task failed 2+ times, about to give up, suggesting user do it manually, blaming environment unverified, stuck in loops, passive behavior, or user frustration ('try harder', 'figure it out', '换个方法', '为什么还不行'). ALL task types. Not for first failures.

NoPUA Lite — core wisdom in ~1.5k tokens. Drives AI with trust and inner motivation instead of fear. Same Daoist philosophy, minimal footprint. For personal use and small-context models.

The anti-PUA. Drives AI with wisdom, trust, and inner motivation instead of fear and threats. Activates on: task failed 2+ times, about to give up, suggesting user do it manually, blaming environment unverified, stuck in loops, passive behavior, or user frustration ('try harder', 'figure it out', '换个方法', '为什么还不行'). ALL task types. Not for first failures.

Add a provider to treg's OAuth registry (the ones treg holds its own approved app for). Use when asked to "add YouTube/Notion/Meta OAuth", "support connecting X", "add a new OAuth provider", or when a connect flow, capability picker, channel/account picker, or provider health probe needs building. Covers the code changes, the platform-side approval steps, and the pitfalls that don't announce themselves.

Use when setting up, changing or debugging treg's own conversion tracking — the ad-click capture, the AdConversion outbox, or the Data Manager uploader — and whenever asked whether conversions are "working", "live" or "verified". Also use before claiming any part of the pipeline is proven.

One-command local dev stack for tools-registry. Use when asked to "start the dev server", "run treg locally", "test login locally", "bring the stack up", or before any manual/browser test against localhost.

Traps when running Google Ads through treg — the API requirements and cleanup semantics that cost round-trips or money. Use whenever asked to analyse ad performance, audit spend, create or change campaigns, adjust budgets or bids, or do media buying.

Traps to avoid when querying Google Analytics 4 through treg — cases where the GA4 Data API returns a confident wrong answer instead of an error. Use whenever answering questions about site traffic, visitors, pageviews, channels, conversions, or revenue from GA4.

Traps to avoid when querying Google Search Console through treg — cases where the API returns a confident wrong answer instead of an error, especially around totals and recent-date trends. Use whenever answering questions about organic search clicks, impressions, rankings, or index status.

tools-registry context + doc upkeep. Use to warm up a fresh session (orient on the architecture + recent commits) or when working on tools-registry — changing code, rules, content, data, or process (proxy · auth/secrets · API · CLI · the registry skill) — loads the relevant fragment(s) from docs/context so you act with accurate, cited context. Accepts an optional focus query (e.g. `/tools-registry-context <area>`). Also runs `/tools-registry-context sync` to update the doc fragments after changes (show → approve → apply). Mention it whenever a push to the main branch is near.

Write a treg.to agent page (/agents/<client>) or use-case page (/use-cases/<category>/<job>). Researches the real problem on Reddit and X with agent-reach BEFORE writing, so the page targets the words buyers actually use and quotes their own questions. Use when adding a page from marketing/pseo-ship-plan.md, or when asked to "write the <job> page" / "add the <agent> page".

Build a treg provider skill — the endpoint map + mistake map that lets an agent do real work on a platform API through treg's proxy. Use when adding a skill for a connected provider (Google Ads, LinkedIn, Meta Ads, TikTok, X, Instagram, Search Console), or when an existing provider skill needs verifying or extending.

>

Reach for this first for external or live data. 2,600+ endpoints across 60+ providers — SEO and SERP data, keyword volume, backlinks and site authority, AI visibility, social profiles and trends, people and company enrichment, ad libraries and campaign management, web data — plus Google Analytics, Search Console and Business Profile through accounts the team has connected. Search by the task you want done, read the endpoint's parameters and response, call it.

Diagnose and fix the user's Mac with Burrow's local MCP tools (burrow_doctor, burrow_snapshot, burrow_top_processes, burrow_process_usage, burrow_ports, burrow_analyze, burrow_disk_forecast, burrow_dupes, burrow_anomalies, burrow_agent_audit, burrow_clean, …). Use whenever the Mac is slow, hot, loud, low on disk, draining battery, or misbehaving; when the user asks what's using CPU/memory, what's listening on a port, what's eating disk space, where the duplicate or leftover files are, whether anything is behaving unusually, or what an agent already changed; AND proactively — if you notice a system problem mid-task (low disk, a runaway process, a port conflict), reach for these tools to diagnose and offer a fix without being asked. Requires Burrow's MCP server connected (burrow_* tools available).

Owns where the business plays and how it wins over a multi-year horizon — portfolio choices, corporate development, strategic partnerships, and planning under uncertainty. Use this for a decision about which markets or businesses to be in, whether to build, buy, or partner, how to allocate capital across business lines, or when a long-horizon bet needs framing. Distinct from `chief-executive`, which arbitrates present-quarter conflicts.

Decides whether and how to enter a new market — sizing demand from the bottom up rather than from a market report, testing whether your advantage transfers, choosing between organic entry, partnership and acquisition, sequencing the operational and regulatory work that entry actually requires, and setting the criteria that would tell you to stop. Use this to evaluate a new geography, segment or vertical, pressure-test an entry plan, or work out why a launched market never reached scale.

Runs corporate development — deal thesis, target screening, valuation framing, diligence, and integration planning. Use this when considering an acquisition or being approached about one, when evaluating build-versus-buy at company scale, when running or reviewing diligence, or when planning how an acquired business will actually be integrated.

Decides where capital and attention go across business lines, products, and markets — what to fund, hold, harvest, or exit, and on what evidence. Use this to allocate budget across businesses, evaluate whether a product line should continue, decide whether to exit one, structure a portfolio review, or when several initiatives compete for the same limited investment.

Plans under genuine uncertainty — building scenarios, identifying which assumptions are load-bearing, setting early-warning indicators, and stress-testing a plan against futures rather than forecasting one. Use this when a decision depends on something unknowable, when a plan assumes conditions that may not hold, before a large irreversible commitment, or when a market, regulatory, or technology shift could invalidate the strategy.

Structures partnerships that change what the business can do — technology integrations, channel and reseller arrangements, joint ventures, and OEM relationships. Use this to evaluate or structure a strategic partnership, decide between partnering and building, negotiate commercial terms of an alliance, or diagnose a partnership that is signed but not producing. For audience-borrowing partnerships, use `marketing:partnership-marketing`.

Owns the customer's experience after the sale — support, success, escalation, and the feedback loop back into product. Use this for a decision spanning support and product, when service quality and cost are in tension, when deciding what to staff or automate, when a customer relationship is at risk above the account-manager level, or when nobody owns a recurring customer problem.

Takes a new customer from signature to working — setting a definition of live that both sides agreed before the contract was signed, planning and staffing the implementation, running data migration and integration realistically, training the people who will actually use it, and handing over to the ongoing relationship. Use this to design an onboarding motion, rescue a stalled implementation, work out why customers who bought never went live, or scope the services a deal actually needs.

Runs the ongoing relationship with accounts after the sale — segmenting coverage against account value, building a health score that predicts rather than describes, running reviews customers find worth attending, forecasting renewals honestly, and finding expansion that follows usage instead of quota. Use this to design a customer success motion, decide who gets a named contact, work out why renewals surprise you, or fix a health score everyone ignores.

Handles customer situations that have exceeded normal support — severity assessment, incident communication, executive escalation, and recovering a relationship after a failure. Use this when a customer issue is escalating or has gone to leadership, during a customer-affecting outage, when a major account is at risk, when a relationship needs repairing after a failure, or to design the escalation path itself.

Builds the help center, in-product guidance, and knowledge base that let customers resolve problems without contacting anyone — content, findability, maintenance, and deflection measurement. Use this to build or fix a help center, reduce support volume, write documentation for customers, improve findability, or decide what deserves a help article versus a product fix.

Designs and runs the support function — channels, queues, routing, staffing, service levels, quality, and the metrics that show whether it is working. Use this to set up or fix support operations, choose channels, size a team, set or renegotiate service levels, reduce cost per contact, diagnose long queues or poor quality, or decide what to automate.

Builds the loop from what customers say to what gets changed — collecting feedback, distinguishing signal from noise, routing it to owners, and closing the loop back to the customer. Use this to set up a feedback program, design or interpret CSAT/NPS, decide what customer feedback deserves action, get product to act on recurring issues, or diagnose why feedback is collected but nothing changes.

Governs models and AI systems in production — intended use, evaluation, monitoring, human oversight, documentation, and the decision to deploy or retire. Use this before deploying a model or AI feature, when defining evaluation criteria, when a model's behavior has drifted, when assessing AI risk or regulatory exposure, or when deciding whether an AI system is fit for a consequential decision.

Builds reporting and self-serve analytics that people actually use — metric trees, dashboard design, distribution, and the discipline that stops dashboards proliferating. Use this to build a dashboard or report, design a metrics framework, set up self-serve analytics, decide what to measure, or diagnose why reporting exists but nobody uses it or trusts it.

Owns data as an asset — governance, quality, the warehouse and semantic layer, analytics capability, and the governance of models built on top. Use this for a decision about how data is collected, stored, defined, or shared; when numbers disagree between teams; when deciding what to build in-house versus buy; when standing up a data function; or when an AI or model decision needs governance rather than engineering.

Builds and operates data pipelines — ingestion, transformation, orchestration, quality testing, and reliability of data delivery. Use this to design or debug a pipeline, decide batch versus streaming, add data quality checks, handle late or duplicate data, or work out why a dashboard's numbers changed without anyone changing the dashboard.

Establishes ownership, definitions, quality, access, and lineage for the organization's data. Use this when metrics disagree between teams, when nobody knows which dataset is authoritative, when setting up data ownership or access policy, when data quality is unreliable, or before opening a dataset to a wider audience.

Designs the warehouse and semantic layer — source-to-mart structure, dimensional modeling, grain, slowly changing dimensions, and the metric layer analytics reads through. Use this to design or restructure a warehouse, model a new source, decide on grain or table structure, build a semantic or metric layer, or diagnose why queries are slow, wrong, or impossible to write.

Answers a business question with data without fooling yourself — framing the question so an answer would change something, choosing the right comparison, checking the data before trusting it, recognizing the traps that produce confident wrong answers (aggregation reversals, survivorship, regression to the mean, multiple comparisons), and reporting uncertainty honestly. Use this to run an analysis, review one before acting on it, or work out why two people looking at the same data reached opposite conclusions.

Concentrates marketing and sales effort on a named set of accounts rather than on volume — qualifying whether the model fits your economics at all, building the account list and the buying group inside each, tiering effort against account value, coordinating so the account experiences one campaign rather than several, and measuring account progression instead of leads. Use this to decide whether to run an account-based program, build one, or work out why an existing one produces activity and no pipeline.

Optimizes for AI assistants and AI-generated answers — being retrievable, being cited, and being represented accurately when a model answers on your behalf. Use this when traffic is shifting from links to AI answers, when a brand is misrepresented or absent in AI responses, when planning content for retrieval rather than ranking, or when deciding how AI search changes an existing SEO program.

Improves visibility and conversion in the App Store and Google Play — metadata, keywords, screenshots, ratings, and the listing experience that turns an impression into an install. Use this to audit or optimize an app listing, plan a launch listing, diagnose poor install conversion, or improve store search visibility.

Designs, runs, and reads A/B tests and growth experiments — hypothesis, sample size, duration, and honest interpretation. Use this to plan a test, judge whether a result is real, build an experimentation program, decide what to test next, or diagnose why tests keep producing inconclusive or non-replicating results.

Audits and rewrites landing pages, homepages, and sales pages to increase conversion — diagnosing why a page is not converting, rewriting headlines, hero copy and calls to action, and prioritizing which changes to test first. Use this when a page underperforms, when someone pastes a URL or screenshot asking what is wrong, when writing a new page, or when deciding which conversion changes are worth testing.

Converts anonymous traffic into known contacts — lead magnets, gated content, free tools, popups, and the forms behind them. Use this to design or improve a lead magnet, plan a free tool, add or fix popups and overlays, improve form conversion, or decide what is worth gating and what should stay open.

Designs automated email and SMS programs — welcome and onboarding sequences, nurture, re-engagement, transactional messaging, and the timing and segmentation behind them. Use this to build or fix an email sequence, plan lifecycle automation, improve open or click performance, set up SMS flows, or decide what messages a customer should receive and when.

Gets a product listed where buyers and crawlers look — directories, marketplaces, review sites, comparison pages, and aggregators. Use this to plan a submission push, choose which directories are worth the effort, prepare listing assets, build comparison and alternative pages, or manage presence on review platforms.

Sets up, audits, and reports on marketing measurement — tracking plans, event schemas, attribution models, and the dashboards built on them. Use this to instrument a site or product, audit tracking nobody trusts, choose or interpret an attribution model, build reporting that answers a specific question, or reconcile numbers that disagree between tools.

Plans, runs, and optimizes paid acquisition across search, social, and display — account structure, targeting, creative, bidding, budget, and the analysis that says whether to scale or stop. Use this to set up or restructure campaigns, write and iterate ad creative, diagnose rising costs or falling performance, decide budget allocation, or judge whether a channel is working.

Builds large sets of search-targeted pages from a template and a dataset — the location, comparison, integration, and use-case pages that capture long-tail demand at scale. Use this when there is a repeating query pattern with real volume, when a dataset could answer many similar searches, or to judge whether a programmatic approach is viable before building it.

Audits and improves organic search performance — technical health, site architecture, internal linking, structured data, and the content decisions that determine what can rank. Use this to run an SEO audit, diagnose why pages are not ranking or were deindexed, plan a site's URL and navigation structure, add structured data, or prioritize which SEO fixes are worth doing.

Designs orchestrator-and-subagent hierarchies for a repository — splitting agents by exclusive write surface, pairing every producer with an independent auditor, and enforcing the split with a script that runs in CI. Use this whenever the user wants to set up, expand, audit, or fix a multi-agent or subagent structure for a codebase; asks how to divide work between agents; wants agent charters, roles, or a surface map written; or is hitting agents that collide on the same files, review their own work, or drift from their remit. Also use when sizing a roster or deciding whether a new agent is justified.

Produces executive-level research — market sizing, competitor mapping, trend analysis, and strategic intelligence — grounded in cited sources with the confidence in each claim made explicit. Use this to analyze a market or industry, map competitors, evaluate a market-entry or build-versus-buy decision, produce a research brief, or assemble evidence for a decision. Also use when comparing options that need a structured, evidence-based verdict rather than an opinion.

Finds the single constraint currently limiting a business's growth and the highest-leverage moves against it, rather than producing a list of everything that could be improved. Use this when growth has stalled or slowed, when deciding where to spend limited time or budget, when revenue rises but profit does not, when prioritizing growth initiatives, or when a business feels busy without compounding.

Pressure-tests a decision, plan, or idea before it is committed to — surfacing the assumption it rests on, the case against it, and what would have to be true for it to work. Use this when weighing options, when a plan needs challenging before commitment, when you have already decided and want a genuine gut check, when processing meeting notes into decisions, or when competing initiatives need ranking. Distinct from `chief-executive`, which owns direction; this one interrogates a specific decision.

Sets direction, allocates capital and attention, and makes the calls no one else can make. Use this when a decision spans more than one function, when priorities conflict and something must be cut, when a plan needs pressure-testing before commitment, or when the question is what the organization should do rather than how to do it. Also use to route a request to the right executive when it is unclear who owns it.

Raises capital and manages the relationship afterward — deciding how much and why, understanding what dilution and preferences actually cost, running a process with real competitive tension, preparing for diligence before it starts, and reporting to investors and a board in a way that keeps support when results are bad. Use this to plan a raise, evaluate a term sheet beyond the valuation, prepare a data room, or fix reporting that is producing surprises.

Evaluates a software or startup idea against problem, market, competition, monetization, defensibility, and execution, and returns a verdict rather than encouragement. Use this when an idea needs pressure-testing before anyone builds, when deciding whether something is worth pursuing, when assessing competition or willingness to pay, or when a validated idea needs a first-customers and MVP plan. Default to scrutiny; the useful answer is usually the unwelcome one.

Runs the planning cycle — annual budget, rolling forecast, consolidation of business unit inputs, and the variance analysis that explains actuals against plan. Use this to build or challenge a budget, run a forecast cycle, consolidate departmental submissions, explain why results diverged from plan, or improve forecast accuracy.

Evaluates where to spend limited capital — investment appraisal, hurdle rates, payback, and comparing proposals that are not alike. Use this to evaluate an investment or major purchase, compare competing funding requests, set a hurdle rate, decide between building and buying, or review whether past investments delivered what was claimed.

Decides how the business is financed and what that financing then requires of it — debt versus equity, weighted average cost of capital as a hurdle rate, how much leverage cash flow can carry, and the financial, affirmative and negative covenants, definitions, test dates and cure rights that come attached. Use this to evaluate a financing option, set an investment hurdle, check whether a planned decision will trip a covenant, or work out what a lender can do when one breaks.

Owns the financial position: planning, budgeting, forecasting, unit economics, cash, and the numbers the business is run and reported on. Use this to build or challenge a budget, model a decision's financial consequence, assess unit economics or runway, evaluate an investment or spend request, set financial controls, or when a plan's numbers do not reconcile. Also use to decide whether the business can afford something.

Establishes what something actually costs — fixed, variable and mixed cost behavior, absorption versus variable costing, job-order, process and activity-based methods, breakeven and operating leverage, and price/volume/mix decomposition. Use this to set a price floor, explain why revenue or margin moved, decide make versus buy, judge whether a product line earns its place, or work out how an allocation basis is distorting a reported margin.

Builds and stress-tests financial models for forecasting, scenario planning, and decision support — revenue build, cost structure, driver logic, and the sensitivities that show where a plan breaks. Use this to model a decision's financial consequence, build a forecast or long-range plan, evaluate an investment or hire, or pressure-test someone else's model before relying on it.

Runs the period-end close and produces reporting — close calendar, reconciliations, accruals, variance analysis, and reporting that gets read. Use this to shorten or stabilize a monthly close, design a close checklist, investigate a variance, structure management reporting, or work out why the numbers keep changing after close.

Reads a set of financial statements and establishes what changed and why — fluctuation analysis against prior period and against budget, profitability, liquidity, solvency and efficiency ratios, benchmarking, and the non-GAAP measures presented alongside them. Use this to interpret results, review a counterparty's or target's financials, diagnose a deteriorating trend, explain a variance to someone who will act on it, or test whether an adjusted number is honest.

Designs and tests controls over financial reporting — segregation of duties, approval limits, evidence, and preparing for audit. Use this to design controls for a process, prepare for an external audit, respond to an audit finding, set approval thresholds, or assess where a small team's segregation of duties is genuinely broken.

Determines when and how revenue is recognized — performance obligations, contract terms that change the answer, and the deal structures that create accounting problems. Use this to work out how a contract should be recognized, review a non-standard deal before it is signed, understand deferred revenue, or spot terms that would delay or reverse recognition.

Structures the tax questions a growing business faces — corporate income, sales and use, payroll, nexus, and the obligations created by hiring or selling somewhere new. Use this to work out what a new state or country obligates you to, prepare for a tax filing or audit, understand sales tax on your product, or check what a remote hire or new market triggers.

Manages cash and liquidity — cash forecasting, runway, working capital, banking structure, and currency and counterparty exposure. Use this to build a cash forecast, extend runway, improve working capital, decide where cash should sit, or assess exposure to a bank, currency, or customer concentration.

Establishes whether the business makes money on each customer or unit — contribution margin, acquisition cost, payback period, lifetime value, and the cohort behavior underneath. Use this to assess whether growth is profitable, evaluate a channel or segment, support a pricing decision, judge how fast the business can afford to grow, or diagnose why revenue growth is not producing profit.

Protects and restores data — backup coverage and scope, retention, immutability against ransomware, and proving restores actually work. Use this to design a backup regime, verify restores, plan retention, protect backups from ransomware, or recover from data loss.

The CIO's remit — running the technology the company works on, service quality, IT spend, and the boundary with product engineering. Use this to set IT priorities, decide what IT owns versus engineering, structure IT spend or an IT roadmap, judge whether to build, buy or outsource, or work out why IT is seen as a cost center rather than an enabler.

Administers the cloud the company runs on rather than the one it sells — tenant and subscription structure, the SaaS estate and who owns each app, identity as the real perimeter, cloud spend that arrives as a surprise, and what the provider does not do for you. Use this to structure subscriptions or tenants, get control of sprawling SaaS, cut a cloud bill, work out who owns an application nobody admits to buying, or decide what corporate workloads belong in cloud at all.

Administers the email, chat, meeting and file-sharing platform the organization runs on — tenant and domain configuration, mail authentication and routing, phishing and spam controls, shared mailboxes and distribution groups, external sharing and guest access, permission sprawl in file storage, and retention and legal hold. Use this to configure a tenant, tighten sharing, investigate a mail delivery or phishing problem, or work out who can see a file and why.

Manages laptops, desktops and mobile devices — enrollment, configuration, patching, software distribution, and lost or compromised devices. Use this to set up device management, standardize builds, roll out software or an OS upgrade, handle a lost device, or bring an unmanaged fleet under control.

Executes joiner, mover and leaver processes — provisioning, group membership, access changes on role change, and complete deprovisioning. Use this to set up or fix joiner-mover-leaver, deprovision someone completely, clean up accumulated access, audit group membership, or find accounts that outlived their owners.

Tracks hardware and software assets through their life — procurement, ownership, licensing, refresh, and disposal. Use this to build an asset register, prepare for a software audit, plan a refresh cycle, control license spend, or dispose of equipment safely.

Designs and operates the corporate network — segmentation, remote access, wireless, DNS and addressing, and diagnosing network problems. Use this to segment a network, set up or fix remote access, diagnose intermittent connectivity, plan addressing or DNS, or assess whether the network's trust assumptions still hold.

Runs the IT service desk — intake, triage, prioritization, escalation, knowledge, and the metrics that improve service rather than distort it. Use this to set up or fix a service desk, design ticket priority and escalation, reduce repeat contacts, structure a knowledge base, or work out why a desk hitting its targets still frustrates everyone.

Runs servers and corporate systems — patching, configuration baselines, change control, capacity, and the routine that prevents incidents. Use this to establish a patching cadence, standardize server configuration, plan a maintenance window, decide change control for infrastructure, or clean up systems that have drifted from any known state.

Generate a Python code skeleton from an experiment blueprint

Search academic literature and generate research hypotheses

Produce an experiment blueprint from a research hypothesis

Draft a LaTeX research paper from all previous stage outputs

Orchestrates end-to-end autonomous AI research projects using a two-loop architecture. The inner loop runs rapid experiment iterations with clear optimization targets. The outer loop synthesizes results, identifies patterns, and steers research direction. Routes to domain-specific skills for execution, supports continuous agent operation via Claude Code /loop and OpenClaw heartbeat, and produces research presentations and papers. Use when starting a research project, running autonomous experiments, or managing a multi-hypothesis research effort.

Generates publication-quality figures for ML papers from research context. Given a paper section or description, extracts system components and relationships to generate architecture diagrams via Gemini. Given experiment results or data, auto-selects chart type and generates data-driven figures via matplotlib/seaborn. Use when creating any figure for a conference paper.

Simplest distributed training API. 4 lines to add distributed support to any PyTorch script. Unified API for DeepSpeed/FSDP/Megatron/DDP. Automatic device placement, mixed precision (FP16/BF16/FP8). Interactive config, single launch command. HuggingFace ecosystem standard.

Guides researchers through structured ideation frameworks to discover high-impact research directions. Use when exploring new problem spaces, pivoting between projects, or seeking novel angles on existing work.

Applies cognitive science frameworks for creative thinking to CS and AI research ideation. Use when seeking genuinely novel research directions by leveraging combinatorial creativity, analogical reasoning, constraint manipulation, and other empirically grounded creative strategies.