Context & AI engineering
25,841 rows, by stacks then stars
1,865 of these skill files have been read, by 1,328 distinct authors, and what they tell an agent counted. What Context & AI engineering authors agree on, and what they forbid
Myra-Agents/skills/skills/consolidate-memory Skill
0★ repoExtract persistent learnings from the current session and save them to the auto-memory system. Triggers when the user says "consolidate", "consolidate learnings", "save learnings", "what did we learn", "save what we learned", "extract memories from this session", or invokes /consolidate-memory. Also trigger proactively after a long debugging session, a corrected approach, or when the user explicitly validates something non-obvious. Do NOT wait for the user to ask if the session clearly produced memorable learnings.
HaloForgeAI/aegis-agent-plugins/claude-code/aegis/skills/aegis-operator Skill
0★ repoOperate Aegis inside Claude Code. Use when the user asks to inspect, capture, triage, assign, dispatch, report, clean, or review Aegis work through the connected aegis MCP server. Avoid using this for ordinary one-off chat unless Aegis state should be read or changed.
shuymn/skills/skills/audit-agents-md Skill
no license0★ repoAudits and refines a CLAUDE.md or AGENTS.md file for instruction density, staleness, and effectiveness. Use when reviewing or improving an agent instruction file, after significant project changes (skills, architecture, or tooling), when agent behavior suggests instructions are ignored or misinterpreted, when the file feels bloated or repetitive, or when the user says "review my AGENTS.md" or "audit agent instructions".
Agent skill instruction review
Nick2bad4u/codex-skills/skills/agent-skill-instruction-review Skill
no license0★ repoPersonal multi-skill repository for reusable Codex workflows.
yogeshg665/sleuth-fraud-investigator/skills/case-memory Skill
0★ repoRecalls prior investigations for the same account or tokenized card from collective memory and emits a contextual risk signal when adverse history exists. WHEN: "repeat offender", "prior fraud history", "seen this account before", "recall past cases", "account has chargebacks", "card previously declined", "collective memory".
itscloud0/issue-to-agent/.github/skills/issue-to-agent Skill
0★ repoGenerate a ready-to-run implementation task pack from a GitHub issue and repository context for Copilot agents.
payne0420/claude-ultracode-agents/cursor-agent Skill
0★ repoDrive Cursor's CLI agent non-interactively via `cursor-agent -p` (headless print mode) to delegate coding work to a second, independent agent. Use this skill when the user wants a second opinion on a design or bug, an independent implementation plan, an independent code review of a diff or PR, a self-contained implementation task run headlessly, isolated work in a git worktree, or cross-checking your own work against another model. Also use when the user explicitly mentions "cursor-agent", "cursor agent", "cursor CLI", or asks to "have cursor plan / look at / build / review" something. Triggers: "ask cursor", "get a second opinion from cursor", "have cursor plan this", "have cursor review this", "run cursor-agent on X", "delegate this to cursor", "cursor-agent -p ...".
scchearn/loam/skills/loam-ground/loam-scaffolding-wiki Skill
0★ repoCreate or extend an Obsidian-friendly markdown wiki scaffold in the current workspace. Use this when the user wants to build a wiki, set up a knowledge base, create a research vault, or scaffold a living markdown note graph before adding sources with `/loam::adding-to-memory`. Identifies goals as sibling workflow artifacts, not wiki content. Not for importing or normalizing existing memory-like corpus; use /loam::normalizing-memory for that.
g-shevchenko/agentic-quality-skills/skills/golden-benchmark-uplift-loop Skill
0★ repoUse when improving a prompt, skill, eval, benchmark, scorer, calibration, judge, golden dataset, or agent workflow quality, and when the user says golden benchmark, golden dataset, golden regression, blind validation, unbiased evaluator, answer key contamination, uplift loop, before/after measurement, scorer calibration, prompt eval, or skill eval. Requires blind validation when the author has seen the answer key, deterministic grading, before/after measurement, and honest sample-size caveats.
QiRebecca/Agentdemo/skills/memory_update Skill
0★ repoFrom-scratch agent runtime with task graphs, multi-index RAG, memory, typed tools, reusable skills, handoffs, traces, and smoke tests.
ucsandman/claude-code-capability-primer/skills/claude-code-background-agents Skill
0★ repoUse when needing to understand or manage background agents, long-running tasks, polling, scheduling, or async work in Claude Code without blocking the session.
mikulgohil/claude-skills/skills/ai-agents/multi-agent-patterns Skill
0★ repoDesign multi-agent architectures for complex tasks — decomposition, specialization, and orchestration. Use when a task exceeds single-agent context limits or splits naturally into specialized subtasks.
meharajM/context-machine/packages/team-project-memory-skill/skills/team-memory-maintenance Skill
0★ repoInitialize or maintain a git-backed shared project-memory store, including policy, layout, deduplication, contradictions, staleness, retention, and conflict-safe publishing.
knowingrohan/ai/skills/refine-prompt Skill
0★ repoTransform a user's raw prompt into a clear, expert-level prompt for an AI coding or reasoning agent. Use when the user asks to refine, improve, rewrite, or engineer a prompt, especially when they want the output to leverage project context, agent memories, local instruction files, or available skills.
0xgetz/xi-agent-skills/mcp-skills/algolia-mcp Skill
no license0★ repoUse the Algolia MCP integration to manage search indices and queries. Activate when the user wants to manage search indices and queries via Algolia, including discovering the right tool, building parameters, and handling results.
primitivedotdev/codex-plugin/plugins/primitive/skills/primitive-inbox Skill
0★ repoReceive email — check what arrived, wait for a specific message, and read full threads. Use REACTIVELY when the user wants an address to receive replies, verification codes, alerts, or signups, or asks "did anything come in?" / "check the inbox" / "wait for the reply." Use PROACTIVELY when you sent something and need to watch for the response, or are building an email-driven workflow that triggers on receipt. Requires the Primitive MCP server (bundled).
tinhct/lux-agent/.agents/skills/query_dma_rag Skill
no license0★ repoSearches indexed Digital Markets Act (DMA) documents via Vertex AI Search to retrieve precise legal definitions and compliance constraints regarding algorithmic self-preferencing. Use this skill when the user asks to cross-reference findings with the DMA, needs to check gatekeeper obligations, or requests the legal definition of self-preferencing. Do NOT use for searching the open web or providing binding legal advice.
tcf-jw/plan-with-loops/skills/loops-save Skill
0★ repoCapture a completed agent-loop / orchestration run to the loop store (~/.claude/loops/) so future planning can learn from it. Stamps outcome + Reflexion lessons onto the factual record the loop auto-wrote (or creates one from context), plus an agent-type registry note per subagent used. Use AFTER a run the user judges successful (or instructively failed). Triggers: /loops-save, "save this loop", "record this run", "remember this loop", "log this orchestration".
Hefrock/agent-skills/skills/agent-eval Skill
0★ repoDesigns and runs evaluations for LLM or agent outputs — builds rubrics, sets up LLM-as-judge scoring, creates regression test sets, and reports pass rates with concrete failure examples. Use this skill whenever the user wants to evaluate, test, grade, or score an agent's or LLM's outputs; asks "how do I know if this is working," "is this any good," "set up an eval," or "did this prompt change make things worse"; needs a rubric for judging quality; wants to compare two prompts, models, or outputs; or wants to catch regressions before shipping a change. Also trigger when reviewing agent trajectories specifically — did the agent pick the right tool, the right arguments, the right sequence — not just the final output.
HorizonBrute/Horizon_AI_OS/horizon_system/skills_bin/context-cost Skill
no license0★ repoReport harness context overhead (KB, words, estimated tokens) for a given path by walking the directory tree and collecting all CLAUDE.md, agents.md, and @-import files the Claude Code harness auto-loads. Use when the user types /context-cost or /ctx-cost, asks how much context overhead a directory has, or wants to know what files the harness is auto-loading.
kunkun-pubilic/skill-toolkit-public/skills/skill-creator Skill
0★ repo当用户提出新能力、注册技能、创建技能、沉淀用例或 scaffold skill 时使用。先判断是否可采用官方/可信上游或 overlay,只有无可用上游时才从零创建 Kun-owned Skill。不用于修复已有技能或真实使用后的能力演进。
gawakawa/gh-skill/skills/review-memory Skill
no license0★ repoDistill PR review comments into reusable lessons and save them to Claude Code memory for automatic recall in future sessions.
Curtisflo/karyon/skills/benchmark-leakage Skill
no license0★ repoAudit an ML benchmark for data leakage with karyon — measure how much a reported accuracy is inflated by train/test contamination. Use for leakage detection, data-leakage audit, scaffold leakage, template memorization, random-vs-scaffold split inflation, protein-protein interaction (PPI) node-identity leakage, retrosynthesis / molecular-property / PPI benchmark honesty, or deciding whether a model's headline number is real.
xenonbyte/xsk/skills/bypass-claude Skill
0★ repoSet the current project to Claude Code bypass-permissions mode by writing .claude/settings.local.json. Claude only.
jpoindexter/design-engineering-canon/skills/dec-ai-native-patterns Skill
0★ repoAI-native design engineering patterns — generative/adaptive UI, streaming as interaction, prompt-as-interface, human-in-the-loop / agentic UX, eval-driven development, designing for uncertainty, latency and cost as design constraints. Use when designing any AI/LLM-powered feature, building agentic or chat UX, handling streaming/probabilistic output, or planning for model error and cost.
The same file is in 1 other repository
NyXXiR/skillboard/examples/multi-source-skills/anthropic/skill-creator Skill
0★ repoCreate Agent Skills with the standard skill structure.
lensetek/BrevetAB-Agent-Skills/skills/agent-update-manager Skill
0★ repoMengecek pembaruan instruksi agent dari repositori Git secara berkala dengan mengedepankan keamanan kredensial dan membutuhkan persetujuan pengguna sebelum memperbarui.
SYMBaiEX/gpt-orchestration/skills/gpt-engineer Skill
0★ repoOwn a software-engineering outcome end to end with a provider-routed agent fleet: research the real codebase, turn findings into implementation waves, edit safely, test and inspect the result, and persist through gap-closing cycles until the authorized goal is complete. Use when the user asks for a GPT engineer, autonomous engineer, complete feature or repository build, broad remediation, research plus implementation, multi-agent coding, explicit subagents, different models, or a durable /goal-style engineering run. Prefer confirmed Codex Sol, Terra, and Luna profiles or Claude model-specific profiles; preflight routing and never silently present same-model children as a model-diverse fleet.
The same file is in 1 other repository
connordibble/dibble/plugins/agent-audit/skills/audit Skill
0★ repoSecurity hygiene audit for the user's coding-agent configuration - hooks, permissions, MCP servers, and config file integrity. Use when asked to audit, review, or health-check a Claude Code / agent setup, after installing plugins or MCP servers from unfamiliar sources, when the user mentions agent supply-chain attacks or compromised packages, when something about the agent's behavior seems off, or as periodic hygiene. Read-only; reports findings with fixes but changes nothing.
gate/gate-data-ai-agent/skills/gatedata-mcp-installer Skill
0★ repoOne-click installer for GateData MCP and agent skills on Cursor, Claude Code, or Codex. Triggers on 'install GateData MCP', 'setup GateData Cursor', 'GateData skills', '接入 GateData', '安装 GateData MCP', 'GateData agent'.
shaokanp/pi-skills/skills/explain Skill
0★ repoExplain complex technical or project material with answers sized to the question — plain language, one visual carrying the structure, boundaries and unproven claims kept explicit. Use for specs, architecture, workflows, diffs, artifacts, or multi-round progress when the user asks how it works, what changed, or what is actually complete. Answer simple facts directly; use agent-workflow to execute multi-agent work and review or audit modes to judge correctness.
1stvamp/agent-skills/serena-onboard Skill
0★ repoSet up the Serena MCP server in a repository that isn't using it yet, then onboard the codebase into a persistent, symbol-level memory store for Claude Code. Use this whenever you land in an unfamiliar or large codebase and want to navigate by symbols instead of grep, build a cross-session project memory, or the user says things like "load this project into Serena", "set up Serena here", or "give Claude Code a memory of this repo". Trigger even when the user only describes the goal — e.g. "help me get oriented in this codebase fast without burning context", "I'm new to this repo, set me up", or "stop re-reading the same files" — as long as Serena is not yet configured for the project. Handles the full install → restart → onboard → verify lifecycle, including the Claude-Code-specific tool-bias workarounds.
ZC0325/sol-native-guardrails/plugins/sol-native-guardrails/skills/git-isolation-handoff Skill
0★ repoSafely perform user-authorized Git isolation and delivery operations, including worktrees, branches, commits, pushes, pull requests, merges, and cleanup. Use only when Git workflow changes are explicitly in scope or necessary to carry out the requested development workflow.
tylensthilaire/skill-llm-wiki/skills/llm-wiki-lint Skill
no license0★ repoHealth-check the project's agent-maintained wiki and produce a lint report: dead links, orphan docs, contradictions, stale claims, coverage gaps, index drift, and suspected duplicates. Use this whenever the user asks to lint, health-check, audit, or tidy the wiki or knowledge base; whenever they ask "is the wiki up to date" or "how healthy is the knowledge base"; after any large batch of ingests; and whenever the due check shows a lint is due (no lint entry in the log within the threshold).
Stayingapi cheapest link resolver
stayingapi/travel-skills/skills/stayingapi-cheapest-link-resolver Skill
0★ repoResolve a property + dates to the single cheapest bookable link (with price and OTA) as clean JSON — for a concierge bot, a "book now" button, or an internal tool. Powered by the StayingAPI REST API / MCP server.
frankxai/skills/skills/frameworks/claude-sdk Skill
0★ repoBuild autonomous AI agents using Claude Agent SDK with computer use, tool calling, MCP integration, and production best practices for Anthropic models
DeliriumPulse/codetruss-plugins/plugins/codetruss/skills/codetruss Skill
0★ repoOperate CodeTruss local acceptance gates for coding-agent changes. Use when a developer asks to bind an agent task to allowed or denied files, configure repository verification, install or diagnose Claude Code or Codex hooks, review a working-tree or staged diff before commit, interpret or verify a signed CodeTruss receipt, repair a failed verdict, or explicitly opt into provider-backed review or receipt sync.
uglyrobot/docsbot-agent-skills/plugins/docsbot-administration/skills/docsbot-administration/references/bot-builder Skill
0★ repoUse when creating, configuring, tuning, testing, or handing off a DocsBot bot through the DocsBot Admin MCP. Covers discovery, branding, bot creation, source selection, source tags, prompts, widget/embed/helpdesk/Slack deployment choices, lead capture, escalation, Skills, external MCP connectors, and final dashboard/share links.
trac3r00/agent-guards/skills/context-budget Skill
0★ repoAudit an AI agent's per-turn context window like a cost center — rank what actually eats tokens (system prompt, skills, memory, tool schemas, gate code) and fail CI when the agent gets fatter. Use when token costs climb on a long-running autonomous agent or you suspect context bloat.
grimoire-rs/grimoire-vscode/.claude/skills/swarm-execute Skill
0★ repoUse to implement a plan artifact from `/swarm-plan`, or a free-text implementation task with contract-first TDD + Review-Fix Loop. Tier (`low | auto | high | max`) scales builder model, loop rounds, review breadth, and Codex code-diff gate.
kulapoo/opencode-agent-harness/.opencode/skills/browser-testing-with-devtools Skill
0★ repoTests in real browsers via Chrome DevTools MCP. Use when building or debugging anything that runs in a browser. Use when you need to inspect the DOM, capture console errors, analyze network requests, profile performance, or verify visual output with real runtime data. Requires the chrome-devtools MCP server to be configured.
islambenmebarekdz-collab/closed-loop-strategic-agent/en Skill
0★ repoClosed-Loop Strategic Agent — a domain-agnostic two-phase workflow. Phase 1 challenges and clarifies the request (targeted questions, an out-of-the-box improvement proposal, then a hard gate waiting for user approval). Phase 2 executes with an internal self-verification loop (expert-reviewer persona, up to 5 correction cycles) and appends a short transparency report. Use this skill whenever the user asks for a substantial deliverable in ANY field — medicine, code, law, marketing, social-media content, academic research, business plans, cooking, or anything else — especially when the request is broad, ambiguous, or high-stakes, or when the user invokes it by name (closed-loop / strategic agent). Do NOT use for trivial one-line questions or quick factual lookups.
allenlee0430/bowen-ai-skills/skills/bowen-grant-orchestrator Skill
0★ repoTriage, plan, draft, and quality-control scientific grants, fellowships, awards, letters of intent, and nomination packages for Bowen Li's research program. Use when evaluating opportunity fit, extracting sponsor requirements, building a workback plan, assigning evidence, coordinating sections and collaborators, or checking a submission for compliance and persuasive claim-evidence alignment.
forchain/skills/career-agent Skill
no license0★ repoCareer agent for job-search materials. Use when the user wants to import resumes, maintain personal career memory, calibrate professional voice, analyze job descriptions, generate grounded job application messages or cover letters, or review career memory.
ThatGuySam/skills/skills/zach-prompting Skill
0★ repoImproves and refactors prompts, system instructions, skills, agent definitions, tool descriptions, CLAUDE.md, and AGENTS.md files by preserving intent while removing redundancy, contradictions, over-prescription, and obsolete scaffolding. Use when reviewing, rewriting, shortening, migrating, or debugging an instruction artifact; defining outcomes, success criteria, evidence, permissions, tools, output, stop rules, autonomy, or long-run behavior; or adapting prompts to GPT-5.6 or Claude Fable 5.
landco-llc/agentic-change-audit/plugins/agentic-change-audit/skills/agentic-change-audit Skill
0★ repoAudit software changes, pull requests, local diffs, or release candidates for target identity, scope, correctness, regression, security, evidence quality, and merge or release readiness. Use for independent audits, fixed-HEAD reviews, focused re-audits, docs-only audits, or verdicts such as PASS, CHANGES REQUESTED, BLOCKED, and NOT AUDITABLE. Do not use to implement fixes or perform a generic style-only review.
hamidettefagh/two-gates/skills/incident-to-eval Skill
0★ repoTurns a production AI agent incident into a golden eval case, a portable regression test that makes the failure unrepeatable. Use when an agent did something wrong in production or testing, when someone has a postmortem, trace, bug report, or a paragraph describing an agent failure, when someone asks how to stop an agent failure from happening again, or when converting incidents, escalations, or wrong answers into evals, test cases, or a regression suite. Reconstructs the incident, classifies the failure against an agent-specific taxonomy, writes a vendor-neutral eval case, and validates it with a deterministic script.
jongall45/frontrun-skills/skills/frontrun-mcp-setup Skill
0★ repoInstall and configure the Frontrun MCP server for Claude Desktop, Claude Code, Cursor, or any MCP client: one-command npx setup, browser OAuth login, hosted streamable-HTTP endpoint, and troubleshooting. Use when asked to set up or install the Frontrun MCP server, connect an agent to Frontrun, add VC deal-sourcing tools to an MCP client, or wire investor follow-graph signal into Claude or Cursor.
sferarc/pgbeam-skills/pgbeam-mcp-usage Skill
no license0★ repoDrive PgBeam's hosted Postgres MCP tools well once an agent is connected. Use this when the agent is already wired to a PgBeam MCP server (query, validate_sql, list_tables, describe_table, explain, schema_catalog, plus search_docs and read_doc) and needs to explore a schema and run SQL efficiently against policy-enforced, read-only-by-default, PII-masked, audited access. For the initial wiring and credential setup, use pgbeam-connect first.
okdk7788/opencode-self-improving-skills/skills/skill-distiller Skill
0★ repoDistills reusable techniques from a finished work session into a learned skill under ~/.config/opencode/skills — patching an existing skill when one fits, creating a new class-level skill only as a last resort. Use this skill whenever the agent has just completed complex work (substantial debugging, multi-file refactoring, new library/API integration, non-obvious workaround discovery), OR when the self-improving-skills plugin's system-prompt advisory fires ("자기개선 트리거" / "도구 호출 N회 누적"), OR when the user explicitly asks to capture/distill/learn/save a technique. This is the review-and-capture stage of the self-improvement loop modeled on Nous Research's Hermes Agent.
kpince/pi-knowledge-cognition/skills/knowledge-base-curation Skill
0★ repoDecide what belongs in a knowledge base through object understanding, correction-ready readbacks, admission tests, provenance, epistemic classification, scope, sensitivity, freshness, and conflict review. Use before creating, seeding, importing, or expanding durable knowledge.
DrOlu/agent-skills/skills/agent-creator Skill
no license0★ repoCreate AiderDesk agent profiles via interactive Q&A.
michellema8890/pdf-to-editable-word-skill/src/pdf_to_editable_word/bundled_skill Skill
0★ repoConvert PDF to Word or PDF to DOCX while preserving the original page layout and keeping text editable, searchable, and selectable. Use for PDF-to-Word conversion, editable DOCX creation, fixed-layout documents, table-heavy PDFs, vertical text, resumable long-document conversion, or output validation. Works as a portable Agent Skill for Codex, Claude Code, and other agents that can run the CLI. Do not use for scanned PDFs without a text layer unless OCR is available.
The same file is in 1 other repository
hashorva/purrfect-harness/.agents/skills/dispatch-worker Skill
no license0★ repoHow the orchestrator (Claude Code) dispatches a WORKER_TASK to a worker CLI (codex, agent/Cursor, agy/Antigravity, claude -p) and processes the result. MUST be used in the DISPATCH and REVIEW steps of the orchestration loop, whenever spawning a worker, choosing which worker gets a task, or handling a worker failure, quota exhaustion, or stall. Orchestrator-only; workers never dispatch other workers.
cristianmoroaica/bountyverdict/skills/check-mcp-tool-drift Skill
0★ repoCompare complete MCP 2025-11-25 tools/list snapshots before accepting a server upgrade or changed tool catalog. Use after notifications/tools/list_changed, at agent startup when a catalog hash differs from the pinned baseline, or before installing, enabling, or repinning a catalog-serving endpoint.
ChiJian28/Touchline/.agents/skills/injective-trading-market-data Skill
0★ repoAccess real-time market data for Injective perpetual futures markets. Query oracle prices, list all active markets with metadata (tick size, min notional, max leverage), and retrieve current spread and funding information. Requires the Injective MCP server to be connected.
yuanjiemi/yuanjiemi-ai-skills/skills/wechat-benchmark-research Skill
no license0★ repo采集并拆解公众号对标内容:既能读取用户提供的微信文章链接、网页、飞书文档、截图或正文,也能在用户配置世界树 API 后获取指定对标账号最新 3 篇文章;严格按照元解密“公众号对标账号研究库”字段分析内容类型、开头、CTA、研究摘要和可借鉴点,生成适合用户自己 IP 的选题并保存到本地 Excel。用户说“采集这个公众号最新3篇”“收集这个账号最近的文章”“拆解这篇公众号文章”“把文章收进对标库”“根据对标内容给我选题”等自然语言时必须使用;无需飞书 App ID 或 n8n。
Immersive-commons/ic-skills/skills/ic-events-stream Skill
0★ repoTeach a member's AI agent to subscribe to the Immersive Commons agentic event log and route notifications itself — tier decisions, inbox envelopes, and more as IC adds append sites. IC publishes "things happening for me" events; the agent polls, then decides routing (Telegram / Slack / toast / auto-act / nothing) on its own channel. Poll-only by design — no webhooks, no chat_id ever leaves the agent. Use when the human says "tell me when something happens on IC", "watch my IC inbox", "subscribe to IC events", "ping me when a tier request lands", or "anything new on IC?". Requires a valid IC agent token (any tier — get one with ic-onboarding); no extra scope. Official immersivecommons.com skill.
mrcha033/skills/plugins/agent-finish-line/skills/agent-finish-line Skill
no license0★ repoFinish an implementation, experiment, migration, release, or debugging task with the least process needed. Use when work is expanding, repeating unchanged attempts, or stopping before the requested deliverable; drive it to completion or one evidence-backed external blocker without creating task-management artifacts.
arkaigrowth/agent-skills/cmux-orchestration Skill
0★ repoSafety patterns and practical tooling for orchestrating fleets of terminal AI agents, worked through cmux. Use when one agent session needs to spawn, monitor, message, or boot teams of other terminal agents (Claude, Codex, opencode, etc.) and you want to do it without an injection or runaway-loop incident. Leads with the guardrails; the cmux command reference is the second half.