Designing conversational flows for website chatbots and AI agents. Intent recognition architecture, branching logic, fallback handling, escalation to human, conversation analytics. Honest about scripted-bot (rigid trees, fail edge cases), hallucinating-bot (LLM without structure, makes things up), and structured-guided-conversation (LLM-powered with intent architecture and fallback discipline) patterns. Distinguishes chatbot DESIGN (this skill) from chatbot IMPLEMENTATION (engineering and platform work). Triggers on chatbot, conversational AI, AI agent, chat widget, intent design, conversational flow, bot escalation, LLM grounding. Also triggers when a chatbot is hallucinating, when a scripted bot is failing edge cases, or when a chatbot is being scoped for the first time.

Plan and run backups, set recovery objectives, and run disaster recovery drills. Use this skill when defining RPO/RTO targets, designing backup architecture, deciding what to back up and how often, planning for full-region or platform outages, or running a restoration drill. Triggers on backup, restore, RPO, RTO, disaster recovery, DR, business continuity, what if the database is gone, what if our hosting goes down, recovery drill, ransomware planning. Also triggers when an incident reveals a gap in restoration capability.

Run a comprehensive WCAG accessibility audit covering perceivable, operable, understandable, and robust principles. Use this skill whenever the user wants to audit accessibility, review WCAG compliance, fix accessibility issues, prepare for accessibility certification, address an accessibility lawsuit risk, or systematically improve a site's accessibility. Triggers on accessibility audit, WCAG audit, a11y audit, accessibility compliance, ADA compliance, screen reader test, keyboard navigation, accessibility report, fix accessibility, axe scan. Also triggers when accessibility issues have been reported and need systematic remediation.

Apply pre-composed aesthetic archetype defaults to jumpstart brand design work covering color, typography, layout, voice, and imagery direction. Use this skill whenever the user wants a starting point for brand visual and verbal direction, references a brand they want to design near (for example 'something like Stripe' or 'editorial like Linear'), needs to converge faster from many directions to one, or wants known-good defaults for a specific industry vertical. Triggers on archetype, design archetype, brand archetype, near to (brand), in the style of, similar aesthetic, design starting point, brand template, vertical-typical, industry-typical. Also triggers when the user has a brief and wants to map it to a known aesthetic family rather than design from first principles.

Manages the research/ knowledge base of verified tool and library research entries via knowledge-explorer.py. Commands — list (browse by category or SDLC layer), fetch-github (draft entry from GitHub repo), add (validate and route to category dir), update-append (append dated revision), generate-descriptions (repair missing or low-quality descriptions), migrate (convert inline-header entries to frontmatter). Use when browsing KB topics, adding or updating research entries, fetching GitHub repo metadata, generating descriptions in parallel via Haiku subagents, or migrating old-format entries to skill-spec frontmatter.

Analyze git repository history to produce a codebase risk profile before code review. Probes repo size, runs 7 parallel git analysis pipelines (hotspots, bug magnets, bus factor, contributor momentum, firefighting frequency, newly added files), cross-references hotspots with bug magnets to identify high-risk files with primary owner annotation, and writes walkthrough/recon-report.md. Use before code review to prioritize reviewer attention, at the start of linear-walkthrough Phase 1 Discovery to weight file coverage, or when onboarding to an unfamiliar codebase.

Use when writing or reviewing asyncio code in Jupyter notebooks or '#%%' cell workflows — structuring event-loop ownership, orchestrating async tasks, or choosing compatibility strategies. Also use when hitting RuntimeError: This event loop is already running, asyncio.run() failures in cells, or tasks silently never completing.

Use when turning a fuzzy idea into a refined design through dialogue — across software, writing, talks, events, naming, activities, or any creative/design task. Its adversarial lenses also stress-test a plan or proposal already on the table. Triggers: 'brainstorm', 'help me think about/through', 'figure out', 'work through', 'I have this idea', 'let's design', 'flesh this out'. Domain-agnostic. Not for: open-ended ideation with no anchor ('what should I build?'), implementing from a finalized spec, narrow binary decisions ('X or Y?'), retrospective review, mechanical refactors, or reframing whether the presenting question is the "real" one.

Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to save input tokens. Preserves all technical substance, code, URLs, and structure. Compressed version overwrites the original file. Human-readable backup saved as FILE.original.md. Trigger: /caveman-compress <filepath> or "compress memory file"

Verify, repair, backfill, and back up Pi/Claude/Codex session capture across active machines. Use when checking whether agent activity is captured, reviewing capture outbox backlogs, backing transcripts up to NAS, or scheduling capture verification.

Produces a complete 5-act pitch with speaker notes, a slide-by-slide specification, and a ready-to-paste Kimi prompt for AI slide generation. Requires validated angle and title from Stage 4. Use when you have a confirmed talk angle and need the full script, slide spec, and AI-generated presentation prompt.

Generate release notes in 3 formats (CHANGELOG.md, PR body, Slack announcement) from git commits. Automatically categorizes changes and converts technical language to user-friendly messaging. Use for releases, changelogs, version notes, what's new summaries, or ship announcements.

Use before claiming that the package can be installed or published. Verifies structure, adapters, schemas, executable scripts, and public-safety boundaries.

Analyze NIH grant portfolios and funding history using the OpenNIH FY1985-present corpus, then connect grants to investigators, institutions, publications, clinical trials, patents, targets, or drugs with ToolUniverse. Use for NIH grant discovery, topic or Institute/Center trends, PI and institution profiles, activity-code or mechanism analysis, funding growth and concentration, grant-writing landscape research, SBIR/STTR landscapes, research-policy analysis, expert discovery, funding-to-output translational impact studies, or public-facing NIH explainers for patients, advocates, local and national journalists, trainees, applicants, institutions, entrepreneurs, and taxpayers.

Functional annotation of protein variants — ProtVar structural/functional context, ClinVar clinical classifications, gnomAD population frequencies, CADD deleteriousness, ClinGen gene-disease validity, plus FAVOR one-call comprehensive GRCh38 annotation. Use for variant annotation pipelines, missense effect prediction, and protein-level variant interpretation with functional context.

Comprehensive drug-target intelligence — tissue expression (GTEx, HPA), pathways, protein interactions (STRING), variant landscape (ClinVar, gnomAD), druggability (DGIdb, ChEMBL approved drugs). 9 parallel research paths with citations. Use for full target profile reports, target characterization for drug discovery, and 'tell me about target X' queries.

Structural biology plus proteomics integration for drug target validation. Combines PDB experimental structures, AlphaFold predictions, GPCRdb, SAbDab antibody structures, ProteinsPlus binding-site prediction, BindingDB ligand-affinity data, BMRB NMR data, CATH/InterPro fold classification, intrinsic-disorder prediction (MobiDB/DisProt/IUPred3), membrane-protein topology (OPM/TopDB/PDBTM/ChannelsDB), enzyme catalytic-site data (M-CSA), PDBe ligand/compound chemistry, and MD simulation trajectory metadata (MDDB). Use for druggability assessment, binding-site characterization, ligand-pocket analysis, structural-confidence scoring (resolution, pLDDT), antibody-target interface analysis, disordered-region assessment, membrane-protein topology, catalytic-mechanism lookup, and finding published MD simulations for a target.

Single-cell RNA-seq analysis with scanpy/anndata — h5ad data loading, scRNA-seq quality control and QC gating (n_genes_by_counts, total_counts, mitochondrial percent / pct_counts_mt, pct_counts_ribo, doublet detection with Scrublet/scDblFinder, ambient RNA / SoupX awareness, empty-droplet filtering, MAD-based thresholds), normalization, dimensionality reduction (PCA, UMAP, t-SNE), clustering (Leiden, Louvain), marker gene identification, cell-type annotation, pseudotime/trajectory analysis. Use for any scRNA-seq workflow, including deciding which cells to filter, flag, or investigate before downstream analysis.

Discover novel small-molecule binders for protein targets using structure-based and ligand-based screening. Covers druggability assessment, known-ligand mining (ChEMBL, BindingDB), similarity expansion, ADMET filtering, and synthesis feasibility. Use for hit identification, virtual screening, target-to-compounds workflows, and lead-finding before commit-to-medchem.

Functional annotation of protein variants — ProtVar structural/functional context, ClinVar clinical classifications, gnomAD population frequencies, CADD deleteriousness, ClinGen gene-disease validity, plus FAVOR one-call comprehensive GRCh38 annotation. Use for variant annotation pipelines, missense effect prediction, and protein-level variant interpretation with functional context.