smart-blitz
Triage-first batch bug fixing with streaming review and merge. Extends blitz-merge with an upfront triage phase that prioritises open bugs, groups them into conflict-free batches, and assigns each the minimum model tier capable of fixing it (Claude Code sonnet/opus, Kiro and Codex gpt-5.6 luna/terra/sol). Reviews and merges each PR as soon as its fix lands instead of waiting for the whole batch. Use for "smart blitz", "triage and fix all bugs", "fix bugs by priority", "batch-fix with the cheapest model", or any request to blitz bugs with prioritisation or cost awareness.
spec-cleanup
Identify and clean up stale or superseded feature specs in a project's specs/ directory. Use whenever the user wants to prune, tidy, audit, consolidate, or declutter specs, mentions outdated/overruled/replaced/superseded specs or a spec that has a v2 or improved version, or asks which specs are still relevant — even if they don't explicitly say "cleanup".
specs-overview
Generate or regenerate specs/OVERVIEW.md with a tabular summary, status, and per-spec file listings
starwave:creating-spec
Initialize a new spec with requirements, design, and task planning. Orchestrates the entire spec-driven workflow from feature idea to actionable task list.
starwave:design
2. Create Feature Design Document
starwave:requirements
1. Requirement Gathering
starwave:smolspec
Small Spec (Smolspec) - Lightweight specification for changes that carry no contested decisions, whatever their size
starwave:tasks
3. Create Task List
capture-knowledge
Capture reusable cross-project technical knowledge into the user's Obsidian vault as a markdown note. Use whenever the user wants to save a learning, pattern, recipe, or how-to that would help future projects — phrases like "save this to my vault", "write this up as a note", "capture this for future reference", "document how X works", "remember this pattern", "add to my knowledge base", or any time after solving something non-obvious in their current repo. Examples are "how to set up an MCP server in a Swift app", "the GraphQL pagination pattern we used here", "how Transit wires up its dependency injection". Use this skill PROACTIVELY whenever the user finishes investigating something they'd plausibly want to reuse on a different project, even if they don't say "save it" explicitly — offer to capture it. Do NOT use for project-specific notes that only make sense inside one repo (those belong in that project's own notes), and do NOT use for daily/journal entries.
pr-review-html
Pull a PR from GitHub, run parallel review agents (code reuse, quality, efficiency, spec adherence), apply local fixes, and write a self-contained HTML review page to the repo root. Use whenever the user wants to review a PR before merging or asks to "review PR <number>", "look at this PR", "give me an HTML review of PR X", or wants a browseable artifact for a GitHub pull request — even if they don't explicitly say "HTML".
pre-push-review
Review unpushed commits before pushing to remote repository
pr-review-fixer
Fetch unresolved change-request review threads (both diff-anchored and CR-level), validate issues, and fix them. Also checks CI status and fixes failing tests, lint errors, and build issues. Works on both GitHub (gh) and GitLab (glab) — it detects the forge from the git remote. Use when reviewing and addressing PR/MR feedback. Filters out resolved threads, keeps only the last Claude review comment per thread (matching the `<!-- claude-local-review -->` sentinel from the local-review agent), validates issues, posts a review report as a CR comment, then fixes validated issues.
pr-pilot
Push a branch, open a change request, iterate on reviews until clean, then squash-merge. Runs /pr-review-fixer in a loop until no blockers/critical/major issues remain, rebases onto latest origin/main, and squash-merges. Works on both GitHub (gh) and GitLab (glab) — it detects the forge from the git remote. Works with or without Transit tickets. Use when you want to shepherd a PR/MR from push to merge, e.g. "push and merge this", "get this merged", "review-fix-merge loop".
pr-overview
Pull a PR from GitHub, run parallel review agents (code reuse, quality, efficiency, spec adherence) in read-only mode, surface unresolved comments, and write a self-contained HTML overview page to the repo root. Use whenever the user wants a read-only summary of a PR before deciding what to do — phrases like "overview of PR X", "summarise this PR", "what's the state of PR X", "give me an HTML overview", or any time they want a browseable artifact without touching the code.
permission-analyzer
Generate Claude Code permissions config from session history. Use when setting up autonomous mode, configuring .claude/settings.json, avoiding --dangerously-skip-permissions, or analyzing what permissions a project needs. Reads session logs to extract Bash commands and MCP tools actually used, then generates appropriate allow/deny rules.
next-task
4. Next task
make-it-so
4. Make it so (implement all tasks)
go-test-fixer
Fix Go unit tests to comply to best practices. Use this when you're asked to modernise tests.
fix-bug
Systematic bug investigation, resolution, and documentation. Use when fixing bugs that need thorough analysis, test coverage, and a formal bugfix report. Applies systematic debugging methodology, creates regression tests, and generates a standardized report in specs/bugfixes/<bug-name>/. For complex bugs, spawns competing implementation agents (including alternative harnesses like Kiro) and selects the best solution. Triggers on requests like "fix this bug", "debug and document this issue", or when a bug needs both resolution and documentation.
explain-like
>-
design-critic
Critical review of design documents, architecture proposals, and requirements
claude-code-workshop
>-
crafting-effective-readmes
Use when writing or improving README files. Not all READMEs are the same — provides templates and guidance matched to your audience and project type.
ideaverse
Work with Ideaverse-based Obsidian vaults using the ACE framework (Atlas/Calendar/Efforts) and LYT (Linking Your Thinking) methodology. Use when creating notes, organizing knowledge, building Maps of Content (MOCs), maintaining daily logs, cross-referencing ideas, assimilating new information, or working with any Obsidian vault that follows Ideaverse conventions. Triggers on requests to add knowledge, link notes, create MOCs, update daily logs, archive projects, or discover connections between ideas.
ideaverse-enrichment
Systematically add new knowledge to Ideaverse vaults while maintaining consistency. Use when assimilating new information, extracting concepts from sources, processing research, integrating learned knowledge, detecting duplicates, or classifying knowledge types. Triggers on requests like "add this to my vault", "process this article", "integrate this knowledge", "what type of knowledge is this", or "check for duplicates".
ideaverse-maintenance
Keep Ideaverse vaults healthy through audits, diagnostics, and maintenance workflows. Use when running vault diagnostics, detecting link rot, identifying orphan notes, finding MOC bloat, suggesting archival candidates, validating frontmatter, or generating vault health reports. Triggers on requests like "audit my vault", "find broken links", "check vault health", "what needs archiving", "find orphan notes", or "run maintenance".
skill-creator
Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
python
Python development guidelines and best practices. Use when working with Python code.
humanizer
|
effect-best-practices
Enforces Effect-TS patterns for services, errors, layers, and atoms. Use when writing code with Effect.Service, Schema.TaggedError, Layer composition, or effect-atom React components.
agent-integration-testing
Use when the user requests integration testing, feature validation, or test plan execution
boost-modules
Create custom modules for [Harbor Boost](https://github.com/av/harbor/tree/main/boost), an optimizing LLM proxy. Use when building Python modules that intercept/transform LLM chat completions—reasoning chains, prompt injection, structured outputs, artifacts, or custom workflows. Triggers on requests to create Boost modules, extend LLM behavior via proxy, or implement chat completion middleware.
tinygrad
Deep learning framework development with tinygrad - a minimal tensor library with autograd, JIT compilation, and multi-device support. Use when writing neural networks, training models, implementing tensor operations, working with UOps/PatternMatcher for graph transformations, or contributing to tinygrad internals. Triggers on tinygrad imports, Tensor operations, nn modules, optimizer usage, schedule/codegen work, or device backends.
turso-db
Install, configure, and work with Turso DB — an in-process SQLite-compatible relational database engine written in Rust. Use when the user needs to (1) install Turso DB, (2) create or query databases with the tursodb CLI shell, (3) use Turso from JavaScript/Node.js via @tursodatabase/database, (4) work with vector search or embeddings in Turso, (5) set up full-text search with FTS indexes, (6) configure transactions including MVCC concurrent transactions, (7) enable encryption at rest, or (8) use Change Data Capture (CDC) for audit logging.
bugbash
Systematically explore and test any software project (CLI, API, Backend, Library, etc.) to find bugs, usability issues, and edge cases. Produces a structured report with full reproduction evidence (exact commands, inputs, logs, and tracebacks) for every issue.
timeboxed-iterating
>
preact-buildless-frontend
Build-less ESM frontends that run directly in the browser without bundlers. Use this skill when creating static frontends, SPAs without build tools, prototypes, or when the user explicitly wants no Vite/Webpack/bundler. Covers import maps, CDN imports, cache-busting, hash routing, and performance patterns.
pull-llamacpp-model
Use when pulling or downloading a new llamacpp model. The active ROCm image (kyuz0/amd-strix-halo-toolboxes) fails to start in the ephemeral pull container without ROCm device access. Must temporarily switch to the standard CPU image.
run-llms
Set up and run local LLMs with Harbor. Use when the user wants to run models locally, install Harbor, start Open WebUI, llama.cpp, Ollama, vLLM, Docker Model Runner, MLX, or oMLX, pull GGUF or HuggingFace models, add SearXNG web search, Speaches TTS/STT, or Open Terminal code execution, launch Codex/Claude/Grok/OpenCode against a Harbor backend, or troubleshoot GPU, VRAM, and service startup.
workmachine
>
factual-communication
Restate content as a flat list of atomic facts in the style of av/facts fact sheets. Use when the user says "just facts", "facts only", "as facts", "give me the facts", or replies with a bare "facts" / "facts?" to a previous message — then the scope is that message's content.
workgraph
Orchestrate work as a cyclic directed graph of subagent-executed nodes with transition criteria on edges, budgeted cycles, and per-node gates (runtime verification, metric, or artifact). Use when the user says "workgraph", "run this as a graph", "graph this work", or when a goal has parallel branches, feedback loops, or mixed gate types that a linear loop cannot express. Every node is sized to a 64k context budget, so small goals run as a two-node do/verify graph and large goals scale out without caps. For clock-driven or metric-driven single loops, timeboxed-iterating and autoresearch remain lighter.
use-coding-agents
>
make-video
>
discipline
Bulletproof agent operating protocol. 15 failure-prevention rules distilled from 120+ real sessions and 10 agent definitions. Covers fabrication, constraint tracking, verification, scoping, retry discipline, and communication. Load before any task to prevent the most common agent failure modes.
catalog-deslop
Catalog-then-fix slop removal on an existing prose draft. Parallel catalog subagents inventory four slop categories (theatrics, overwrought register, corrective antithesis, dramatic short sentences), fix subagents repair only what a catalog names, mechanical lint + frequency analysis bracket the run, and a regression gate (lint delta, word count, cold meaning check) decides whether the result ships. Use when the user says "deslop this draft", "catalog deslop", "strip the AI patterns from this file", or as the final stage of the article-factory and mosaic-writing skills. For writing or redrafting prose from scratch, use the prose skill instead - this skill only repairs an existing draft in place.
mosaic-writing
Assemble a finished piece from human-written fragments via an orchestrated two-loop pipeline - an assembly loop (assembler + cold reviewer) that sequences the fragments into a draft, then a polish loop (polisher + cold reviewer), then frequency analysis and a catalog-deslop pass with a regression gate. Two modes with opposite fidelity contracts - stitch mode preserves the author's exact wording, narrative mode preserves every idea but may rewrite freely. Use when the user says "mosaic", "assemble my fragments", "stitch these notes into a post", or provides a fragments file to turn into a finished piece. For writing from a topic with research, use article-factory; for prose from scratch, use the prose skill.
improve-skill
Improve an existing agent SKILL.md through one gated cycle: select the target (default this skill), snapshot the committed baseline, diagnose against skill-design-principles plus four rubric dimensions, apply exactly one focused change, and ship only if a different agent than the author PASSes with quoted evidence. Use when the user says "improve this skill", "improve-skill", "iterate on SKILL.md", or "apply skill-design-principles to a skill". Do not use for code slop over a duration (anneal), generic timeboxed work (timeboxed-iterating), graph orchestration (workgraph), metric hill-climb on a codebase (autoresearch), or stripping slop from existing prose (catalog-deslop).
ideate
Timeboxed ideation on a topic where lens-driven proposers generate structurally different ideas, blind critics cull them, and pairwise judges rank the survivors on a tournament ladder instead of 1-10 scores, evolving the leaders until the clock runs out. Use when the user wants to brainstorm, explore ideas, discover features, generate options, or think through possibilities for a specified duration. Triggers on requests like "brainstorm X for 30 minutes", "ideate on X", "spend an hour thinking about X", "what features should we build", "explore options for X". Not for executing a task over time (timeboxed-iterating) or optimizing a measurable metric (autoresearch).
sandcastle
Orchestrate AI coding agents (Claude Code, Codex, OpenCode) in isolated sandboxes using the @ai-hero/sandcastle SDK. Use when the user needs to (1) run agents AFK in Docker/Podman containers, (2) build multi-agent pipelines with plan-execute-review patterns, (3) run parallel agents on separate worktrees, (4) create iterative agent loops with maxIterations, (5) extract structured output from agent runs, (6) set up sandcastle in a new or existing project, or (7) write prompt files with template args and shell expressions.
Page 984 of 1735 · 86742 results
