Latest Agent Skills

Generates a CONTEXT.md routing file for a specified project subdirectory, following the canonical anatomy and ensuring it meets size and content constraints.

Launches long-running evaluation batches in isolated tmux sessions with pre-flight verification, monitoring, and post-flight analysis for unattended runs.

Interrogates research plans before evals or published claims, forcing explicit null hypotheses, confound checks, and data verification across four review waves.

Verifies every bibliographic entry in a paper for existence, metadata accuracy, and contextual support using web/DBLP/arXiv lookups, flagging hallucinated authors, wrong years, fabricated venues, and misattributed citations.

Stress-tests a plan against the project's domain model and documented decisions, sharpening terminology and updating DOMAIN.md and ADRs inline as decisions crystallize.

Launches four parallel adversarial review agents covering security, scope, fidelity, and test gaps, then synthesizes findings into a ranked verdict for plans or feature branches.

Runs a data augmentation pipeline on a given input and reports pass/fail per transform, diagnosing failures against known issues.

Sync Claude Code assets between a project and its project-template buffer, with subcommands for status, diff, pull, promote, trial, and sync-from-buffer. Promotion is always opt-in and never automatic.

Drafts a LaTeX paper section by section from an outline, with venue-specific templates and citation management.

Grades and classifies evaluation batch results, applying exclusions, diagnosing failure modes, computing pass rates, and generating summary tables for papers.

Runs a structured code review using four parallel subagents covering style, correctness, security, and performance, then synthesizes findings by severity and gives an approve/needs-changes recommendation.

Guides the creation of well-structured Claude Code skills and slash commands, covering skill types, scope, frontmatter settings, and file organization.

Rewrites free-form draft prompts into the conventions Claude Opus 4.6 or 4.8 responds best to, either inline or by overwriting a draft file.

Orchestrates structured experiments end-to-end: initialize, plan, run, and evolve with reproducibility tracking and hypothesis validation.

Audits a paper draft by tracing every numeric claim and assertion back to result files on disk, flagging untraced, mismatched, stale, uncited, and phantom references.

Runs a test-driven rendering loop for Copier templates, writing E2E assertions first and iterating until every flavor renders cleanly before committing.

Guides a structured feature-development workflow: explore the codebase with subagents, draft a plan for approval, implement step by step, verify with parallel checks, and record conventions.

Runs a post-batch analysis pipeline after an eval completes: verifies results, executes analysis scripts, refreshes dashboards, and writes a summary report.

Conducts multi-source web research on any topic, cross-referencing claims and delivering a structured report with sources and recommendations.

Designs and iterates production-grade frontend interfaces, covering UX review, visual hierarchy, accessibility, performance, responsive behavior, theming, typography, layout, color, motion, and design systems. Handles websites, landing pages, dashboards, product UI, components, forms, and more.

Executes multi-stage release plans by reading a PHASE-PLAN.md, implementing each stage, running validation and critique gates, committing with clean scope, and advancing automatically, halting on repeated failures or unresolvable issues.

Launches coordinated agent teams for tasks requiring persistent cross-talk, shared state, and lifecycle management, with team design, advisor coordination, and quality gates.

Parses external reviews, enforces coverage and grounding, and drafts a text-only rebuttal under venue limits, managing follow-up rounds.

Launches a model evaluation batch with parameter collection, pre-flight checks, execution, and post-run analysis for interactive or foreground runs.

Runs a two-wave validation pipeline (deterministic checks then rule-based checks) before every git commit, writing a sentinel file that a pre-commit hook enforces. Includes optional quick mode (first wave only) and a fix loop with up to three iterations.

Scans recently changed code for tech debt: duplicated logic, dead code, magic numbers, deep nesting, and TODOs, then proposes fixes.

Generate a structured specification document before writing implementation code, validating scope, drafting specs, and registering them in the project.

Generates diagrams and visuals from code or descriptions, routing to specialized tools for hand-drawn, formal, academic, or architectural output.

Writes or updates a HANDOFF.md document to preserve context for the next agent, including goal, progress, what worked, what didn't, and next steps.

Structured workflow to reproduce, diagnose, plan, fix, verify, and record a bug fix, requiring a deterministic reproduction before diagnosis and minimal changes only.

Quickly refreshes project state when resuming work by reporting git status, recent commits, environment, memory staleness, pending tasks, and red flags.

Creates a conventional commit from staged changes, using an optional user-provided subject.

Consolidates auto-memory files in ~/.claude/projects by auditing, planning, and executing a 4-phase cleanup with user approval.

Orchestrates a strict four-stage shipping pipeline: session critique, validation, commit, and pull request, halting on failure. Supports a critique-only mode for mid-session quality checks.

Pushes the current branch and opens a GitHub pull request with a structured title and body.

Assesses genome, transcriptome, and protein completeness with BUSCO v6, automatically resolving the correct lineage from an organism description and generating reproducible reports.

Performs deterministic image operations with Python Pillow: resizing, cropping, compositing, format conversion, watermarks, brightness/contrast adjustments, and web optimization.

Returns structured prediction-market data for Polymarket, Kalshi, Limitless, Opinion, Predict.Fun, dFlow, and UMA oracle via a local API, covering markets, cross-venue search, leaderboards, smart money, wallet analytics, identity clustering, and resolution status.

Generate or edit images through ClawRouter's local API, with automatic payment via x402 and support for multiple models.

Audits an Obsidian wiki for structural issues like orphaned pages, broken wikilinks, stale content, and provenance drift, with optional automated fixes.

Adapts unfamiliar public websites generically, testing search inputs, buttons, openers, and popups before deciding whether a dedicated site profile is needed.

Provides a structured checklist of design principles and rules to produce grounded, non-generic UI/HTML/slide outputs by gathering context, offering multiple variants, and avoiding AI-slop patterns.

Retrieves current weather conditions and forecasts for any location using free, keyless services.

Search, install, update, and list agent skills from the ClawHub public registry using natural-language queries.

Updates the installed Aegis Method Pack on the current AI coding host, including registration, dry-run previews, and post-update verification.

Frames tasks with explicit goals, success evidence, stop conditions, and boundaries before execution, then routes work to the appropriate workflow.

Opens a GitHub pull request targeting the main branch using the gh CLI and the repository's PR template, after verifying local checks pass.

Routes app-store optimization requests to the right specialist skill, loading up to three matching skills and asking one clarifying question when intent is ambiguous.

Plans a comprehensive app launch strategy, covering pre-launch ASO, press outreach, beta testing, launch-day activities, and post-launch optimization.

Mounts a fetch-native CopilotRuntime on any JS server, wires middleware, selects an AgentRunner, and integrates built-in or external agent frameworks.

Annotates VCF variants using Ensembl VEP, ClinVar, and gnomAD, ranks them by predicted impact, and generates a reproducible report.

Searches UK Biobank's 12,000+ data fields and publications by natural language query, returning ranked field IDs and descriptions for research questions.

Scaffolds a new ClawBio skill from a JSON/YAML spec or interactively, generating SKILL.md, Python skeleton, tests, and updating the catalog.

Performs differential expression analysis on label-free quantitative (LFQ) proteomics data from MaxQuant and DIA-NN outputs, including preprocessing, imputation, statistical testing, and visualization.

Downloads genomes, genes, virus sequences, and taxonomy data from NCBI using the datasets and dataformat CLI tools.

Automates genome-wide association studies from genotype files to publication-ready results, running PLINK2 QC and REGENIE regression with Manhattan and QQ plots.

Predicts tissue or cell-type gene expression (log TPM and TPM) from a TSS-centered DNA sequence using the hosted Genomic Intelligence G0 Expression model, conditioned on a free-text cell-type description.

Predicts gene and transcript structure from a DNA sequence using the hosted Genomic Intelligence API, producing a report and JSON output.

Discovers and executes bioinformatics tools from the Galaxy ecosystem via natural language, with multi-signal scoring, workflow templates, and reproducibility bundles.

Computes HEIM diversity and equity metrics from VCF or ancestry data, generating heterozygosity, FST, PCA plots, and a composite HEIM Equity Score with markdown reports.