hari
Community@harivansh-afk
Hari maintains a 90-skill registry spanning Swift/iOS debugging, Nix packaging, Antithesis deterministic testing, Rust/Python conventions, and agent orchestration for developer environments.
Agent Skills by hari
Showing 81 vetted skills indexed across 4 GitHub repositories.
harivansh-afkbeeper
Send and read chat messages across bridged networks via the Beeper Desktop API.
harivansh-afkproperty-testing
Write property-based tests in Rust using the hegel library with generators, shrinking, and replay.
harivansh-afkprompt-eval
Validate prompt changes by testing fresh Claude sessions against neutral tasks.
harivansh-afkpython-style
Enforces Python conventions for uv projects, Nix packaging, type checking, and ruff linting.
harivansh-afkantithesis
Provides cross-cutting facts and gotchas for Antithesis testing work on the ix repo.
harivansh-afkzero-tech-debt
Refactors code changes toward the intended end-state architecture by deleting dead compatibility paths.
harivansh-afksearching
Search codebases, external repos, and fleet history using semantic and exact text search.
harivansh-afkdistill-claude-md
Compress and deduplicate instruction documents while preserving behavioral rules.
harivansh-afkcraft-standard
Enforces code quality standards for ownership, typed boundaries, and root-cause fixes.
harivansh-afkunknowns
Surfaces unknown unknowns before unfamiliar work and rewrites the user's prompt.
harivansh-afkdependency-intake
Manages external dependency intake in a Nix monorepo with fetchers, hashes, and update scripts.
harivansh-afkrust-style
Applies Rust house style rules for edition 2024 crates, unsafe validation, fuzzing, and mutation testing.
harivansh-afkunibind
Generates idiomatic TypeScript, Python, Kotlin, and Elixir SDK bindings from one Rust surface.
harivansh-afkio-uring
Implements Linux io_uring async I/O operations in C and Rust.
harivansh-afkupstream-contribution
Guides opening pull requests against third-party upstream repositories following their contribution rules.
harivansh-afkvisual-explainers
Build static-first visual explainers as Svelte dashboard boards or self-contained HTML files.
harivansh-afkfirst-principles
Explains technical and STEM concepts using first-principles teaching with interactive HTML explainers.
harivansh-afkreview-changes
Runs a multi-agent adversarial review over the working-tree git diff and returns ranked findings.
harivansh-afkexperiment
Runs a change-measure-decide experiment loop to validate prompt and agent tuning changes.
harivansh-afkwriting-style
Rewrites prose to remove AI writing patterns and match human expert style.
harivansh-afksynthesize
Iterates research, proposal, and independent critique loops into an ADR-shaped decision memo.
harivansh-afknix-style
Enforces Nix code style rules and idioms for writing and reviewing Nix code.
harivansh-afknix-philosophy
Guides Nix flake design covering outputs, pure evaluation, and IFD boundaries.
harivansh-afknix-debugger
Diagnose unexpected Nix rebuilds, slow builds, and silent failures using plan scoring and provenance tools.
Frequently Asked Questions About hari
FAQPage SchemaWhat tasks can I accomplish with harivansh-afk's skill registry?▼
You can debug SwiftUI, Xcode, and iOS build issues, plan SQLite/GRDB/SwiftData migrations, write Nix flakes with lint-compliant style, run Antithesis property-based test campaigns, orchestrate coding agents in tmux, review diffs adversarially, and process PDFs, XLSX files, and chat messages via Beeper.
Who is the target audience for these skills?▼
The registry targets Apple platform engineers working in Swift and SwiftUI, Nix and Rust systems developers, and engineers running agentic coding sessions with Codex CLI or Claude Code who need structured verification, prompt evaluation, and deterministic testing practices.
How do I install and run these skills?▼
Use the skill-installer skill to install curated skills into $CODEX_HOME/skills from a GitHub repo path, or the clawdhub skill to discover, install, update, and publish skills via the Clawdhub CLI. Antithesis skills additionally require snouty, Docker or Podman with Compose, and agent-browser.
What prerequisites and dependencies do the skills require?▼
Antithesis skills require snouty, Docker/Podman Compose, agent-browser, and jq. Beeper requires the Beeper Desktop API on localhost with OAuth PKCE. Property testing uses the hegel library. Swift skills assume Xcode, Instruments 26, and GRDB.swift; Python skills assume uv, ruff, and ty/zuban/mypy.
Do the skills support verifying that prompt or instruction changes actually work?▼
Yes. The prompt-eval skill spawns fresh clean-context Claude sessions with neutral tasks to check whether edits to CLAUDE.md, skills, or system prompts took effect, reporting a pass rate. The experiment skill runs change-measure-decide loops with rollouts against a baseline for keep-vs-revert decisions.