avifenesh/awesome-slash
28 skills54 installs26.1k starsGitHub
Install
npx skills add https://github.com/avifenesh/awesome-slashSkills in this repo
1Consultconsult runs cross-tool AI consultation by querying another AI CLI and returning its response. A developer uses it to get a second opinion or cross-check a decision with Gemini, Codex, Claude, OpenCode, or Copilot. It maps effort levels to specific models, can attach a diff or file as context, and supports resuming prior sessions.2installs2Debatedebate provides prompt templates, context-assembly rules, and a synthesis format for structured multi-round debates between AI tools. A developer uses it to stress-test an idea by pitting a proposer against a challenger across configurable rounds, with every claim required to cite evidence. It manages context growth by summarizing earlier rounds and outputs a reasoned verdict.2installs3Deslopdeslop cleans AI slop from code by scanning for debug statements, ghost code, dead code, and other repo-hygiene issues. A developer runs it during review to get certainty-ranked findings (HIGH/MEDIUM/LOW), with auto-fixable ones flagged. It returns structured JSON for an orchestrator to apply fixes and can use a repo-map to find orphaned infrastructure and unused exports.2installs4Discover Tasksdiscover-tasks pulls open work items from GitHub Issues, GitHub Projects v2, GitLab, and local markdown task files, then filters out claimed tasks and issues that already have an open PR. It scores the remaining tasks with a priority filter using label maps for bugs, security, and features, and presents them for the developer to pick. It runs standalone or as Phase 2 of the /next-task workflow.2installs5Drift Analysisdrift-analysis compares documented plans (PLAN.md, roadmaps, READMEs, issues) against the actual codebase to detect four kinds of drift: plan, documentation, issue, and scope. It collects state with pure JavaScript and runs a single semantic analysis pass, then scores each gap with a priority formula and confidence indicators. The output is a prioritized reconstruction plan bucketed into Immediate, Short-term, Medium-term, and Backlog.2installs6Enhance Agent Promptsenhance-agent-prompts analyzes agent definition files (name, description, tools, model frontmatter plus body) against prompt-engineering best practices. It runs 30 pattern checks covering frontmatter, structure, instruction effectiveness, tool configuration, and security, reporting issues by certainty. With a --fix flag it applies auto-fixes such as scoping unrestricted Bash or upgrading weak constraint words to MUST.2installs7Enhance Claude Memoryenhance-claude-memory analyzes project memory files (CLAUDE.md, AGENTS.md) for structure, instruction effectiveness, content positioning, and efficiency. It validates that file and command references actually exist, estimates token counts, and flags duplication with README. The goal is memory files that AI assistants follow more reliably, with cross-tool support for Claude Code, OpenCode, and Codex.2installs8Enhance Cross Fileenhance-cross-file analyzes semantic consistency that spans multiple agent, skill, and workflow files, catching issues single-file analysis misses. It flags tools used in a body but missing from frontmatter, references to non-existent agents, duplicate instructions across files, and contradictory ALWAYS-vs-NEVER rules. All 8 patterns are medium certainty with no auto-fix, since cross-file changes need human review.2installs9Enhance Docsenhance-docs analyzes markdown documentation for readability, structure, and RAG optimization. It validates links and heading hierarchy, estimates tokens, replaces verbose phrases, and checks chunk sizes and semantic boundaries for retrieval. An --ai mode tunes docs for agent/RAG consumption while the default balances human and AI readability, with auto-fixes available via --fix.2installs10Enhance Hooksenhance-hooks analyzes Claude Code hook definitions and scripts (.md, .sh, .json) for safety, correctness, and best practices. It classifies each hook by type and event against the full lifecycle reference, checks matchers, timeouts, exit codes, and output schemas, and flags dangerous shell patterns. With --fix it applies auto-fixes for HIGH-certainty issues.2installs11Enhance Orchestratorenhance-orchestrator coordinates all of the enhance-* analyzers in parallel and merges their findings into one unified report. It discovers which artifact types exist (plugins, agents, CLAUDE.md, docs, prompts, hooks, skills, cross-file), spawns the matching enhancer subagents with Task, and aggregates results ordered HIGH to LOW certainty. It never auto-fixes without an explicit --apply flag and never touches MEDIUM or LOW issues.2installs12Enhance Pluginsenhance-plugins analyzes plugin structures, MCP tool schemas, and security patterns against function-calling best practices. It validates plugin.json fields, tool and parameter descriptions, schema shape, and version sync, and flags security risks like unrestricted Bash, command injection, path traversal, and hardcoded secrets. Auto-fixes cover HIGH-certainty schema issues via --fix.2installs13Enhance Promptsenhance-prompts analyzes general and system prompts for clarity, structure, examples, and output reliability. It applies research-backed prompt-engineering rules covering XML tags, few-shot example counts, positive versus negative constraints, chain-of-thought use, and schema-based structured output. It complements enhance-agent-prompts, which focuses on agent frontmatter files instead of general prompts.2installs14Enhance Skillsenhance-skills analyzes SKILL.md files for trigger quality, structure, and discoverability. It validates frontmatter fields, checks that descriptions contain 'Use when' trigger phrases, verifies side-effect skills are protected from auto-invocation, and confirms tools are scoped. With --fix it applies auto-fixes for detected issues. It ships a complete frontmatter and invocation-control reference.2installs15Learnlearn researches a topic online by gathering and scoring web sources, then synthesizes a structured learning guide. It uses a progressive broad-to-deep query architecture, scores each source on authority, recency, depth, examples, and uniqueness, and fetches only the top sources to avoid context rot. It writes a topic guide and a master index under agent-knowledge/ that agents can reuse later.2installs16Orchestrate ReviewThis skill orchestrates a multi-pass code review by spawning parallel review agents, each focused on a different concern. It runs four core passes for code quality, security, performance, and test coverage, and adds conditional specialists like database or frontend based on detected file signals. A developer uses it for a thorough or deep review, and it aggregates findings by severity and iterates until the code is clean.2installs17Perf AnalyzerThis skill synthesizes the results of a performance investigation into clear, evidence-backed recommendations. A developer uses it at the end of a perf study, feeding it baseline data, experiment results, profiling evidence, and tested hypotheses. It outputs a summary, recommendations, abandoned hypotheses, and next steps, and it only cites evidence that exists in logs or code.2installs18Perf Baseline ManagerThis skill manages performance baselines by storing one JSON file per version and comparing versions. A developer uses it to keep a consistent record of benchmark metrics and environment metadata across releases. It enforces a single baseline per version and overwrites older baselines for the same version to avoid duplicates.2installs19Perf BenchmarkerThis skill runs sequential performance benchmarks under strict duration rules to establish baselines and validate regressions. A developer uses it when they need reliable, non-parallel benchmark measurements. It enforces a 60 second minimum run, a 10 second warmup, re-running of anomalies, and emits metrics as a JSON block between markers.2installs20Perf Code PathsThis skill identifies the likely implementation paths for a performance scenario before profiling begins. A developer uses it to narrow a codebase to the entrypoints, handlers, and hot files worth investigating. It leans on repo-intel data when available or falls back to grep, and it lists candidate files with symbols and short evidence.2installs21Perf Investigation LoggerThis skill appends structured notes to a performance investigation log. A developer uses it to keep a durable, evidence-backed record of each phase, including verbatim user quotes, decisions with rationale, and pointers to files, metrics, and commands. It writes to a per-investigation markdown file and never paraphrases user quotes.2installs22Perf ProfilerThis skill runs profiling tools to capture CPU and memory hot paths with concrete evidence. A developer uses it to find hotspots, generate flame graphs, and capture JFR or perf output tied to a specific scenario. It verifies debug symbols first, records file and line for each hotspot, and keeps outputs minimal and evidence-backed.2installs23Perf Theory GathererThis skill generates performance hypotheses for a specific scenario, grounded in git history and code paths. A developer uses it at the start of a perf investigation to enumerate what might be slow and why. It reviews recent git history, identifies involved code paths, and produces up to five hypotheses each with evidence and a confidence level, without suggesting optimizations.2installs24Perf Theory TesterThis skill runs controlled experiments to validate performance hypotheses. A developer uses it to test one hypothesis at a time by applying a single change, measuring the delta against a clean baseline, and reverting before the next experiment. It runs at least two validation passes, forbids parallel benchmarks, and records evidence to reach an accept, reject, or inconclusive verdict.2installs25Sync Docssync-docs scans a repository to find documentation that has drifted from the current code. It compares README, CHANGELOG and docs files against changed files, flagging version mismatches, removed exports, outdated examples and undocumented symbols. A developer runs it after code changes or before a PR to keep docs accurate. It runs in report mode by default and can optionally apply fixes.2installs26Validate Deliveryvalidate-delivery checks whether a coding task is complete and ready to ship. It runs the project's tests and build, verifies extracted requirements are implemented, and compares test counts before and after to detect regressions. A developer or agent runs it before shipping to get a pass or fail decision. It outputs structured JSON with specific fix instructions when checks fail.2installs27Maintain Cross PlatformThis skill holds the knowledge needed to maintain AgentSys's cross-platform architecture across Claude Code, OpenCode, and Codex CLI. A developer uses it when preparing a release, validating platform compatibility, or updating installation infrastructure. It documents config formats, install locations, frontmatter transformation rules, and the validation scripts that gate every push.1installs28Repo IntelThis skill performs unified static analysis of a repository, combining git history intelligence, AST symbol mapping, and project metadata. A developer uses it to generate a repo map and query hotspots, code ownership, or bus factor. It initializes and updates incrementally, stores data in the platform state directory, and validates output with a map-validator agent.1installs