AI & ML

spences10 Bundle

Multi-agent orchestration patterns for Claude Code team mode. Use when coordinating teammates, decomposing complex tasks, or managing shared task lists.

spences10 Bundle

Explore a codebase to find opportunities for architectural improvement, focusing on making the codebase more testable by deepening shallow modules. Use when user wants to improve architecture, find refactoring opportunities, consolidate tightly-coupled modules, or make a codebase more AI-navigable.

microsoft-foundry Skill

Synchronize MCP server configuration between VS Code (.vscode/mcp.json) and Copilot CLI (~/.copilot/mcp-config.json). Use when setting up Copilot CLI for the first time, when MCP servers are added or changed in the repo, or when a user reports missing MCP tools in CLI.

microsoft-foundry Skill

Provides documentation audit checklists and quality standards. Use when reviewing README files, SKILL.md files, agent definitions, copilot-instructions.md, or ARCHITECTURE-FLOW.md for quality and consistency.

microsoft-foundry Skill

Validate that Copilot CLI can see all repo skills, MCP servers, and custom instructions. Use when setting up CLI for the first time, after adding skills, or to diagnose CLI issues.

costa-marcello Bundle

Discovers and installs agent skills from the open ecosystem and CCPM registries. Use when users ask 'how do I do X', 'find a skill for X', 'is there a skill that can...', 'ccpm', 'claude code skill', 'installed skills', or want to extend capabilities.

dbosk Bundle

Back factual and empirical claims with a small two-sided literature review — searches for literature that supports AND refutes/qualifies the claim (COUNTER) — verify sources in full text, and record provenance (how found, why picked, quote, counter-search outcome). Format-independent: papers, reports, memos, evidence logs, survey answers — any output resting on a factual claim, not just TeX. Use proactively when: (1) adding OR reusing a citation, page number, or attribution for a factual/empirical claim (inherited citations are NOT pre-verified), (2) attributing a claim to a source ("X found that…"), (3) the user asks to find a reference or check a citation — or wants an answer or decision backed by evidence, research, or literature, (4) writing related-work, background, or claims sections, (5) mentions of scholar, BibTeX, DOI, arXiv, OpenAlex, Crossref, or a literature search. (6) DELEGATING literature research to a subagent or teammate — the subagent prompt must tell it to load this skill, run every search

costa-marcello Bundle

Creates professional software diagrams using Mermaid's text-based syntax. Use when users request diagrams, visualizations, or documentation of software architecture including class diagrams, sequence diagrams, flowcharts, ERDs, C4 architecture diagrams, state machines, git graphs, or gantt charts. Triggers: 'diagram', 'visualize', 'model', 'map out', 'show the flow', database design, code structure.

microsoft Bundle

Run break/fix demo scenarios with browser verification. Use when asked to demo, break the app, or show the SRE Agent working.

joannis Bundle

Expert guidance on contributing to WendyOS: Yocto builds, agent internals, E2E testing, and system architecture. Use when developers mention: (1) building WendyOS images, (2) meta-wendyos layers or bitbake, (3) wendy-agent development or internals, (4) containerd or nerdctl on WendyOS, (5) E2E tests for wendy-agent, (6) Yocto recipes or bbappend files, (7) mDNS/Avahi service configuration, (8) device identity or UUID generation.

microsoft Skill

Use when an operator or a scheduled task requests a periodic Network Security Group (NSG) and connectivity-path audit of the Zava Learning resource group — enumerate every NSG and its rules, trace the full path (NSGs, UDRs / route tables, Private Endpoints, DNS, and any firewall), flag overly-permissive, shadowed, unused, or orphaned rules, and recommend least-exposure corrections. Read-only; produces findings for a report.

microsoft Skill

Use when an operator or a scheduled task requests a periodic RBAC / least-privilege audit of the Zava Learning resource group — enumerate the real role assignments effective on the group, correlate each identity against actual Activity-Log usage to find standing access that is never used, flag over-privileged and directly-assigned (versus group-governed) identities and stale troubleshooting access, and recommend least-privilege corrections. Read-only; produces findings for a report.

microsoft Skill

Use to produce the root-cause analysis narrative for a resolved or mitigated Zava Learning incident — the incident timeline, the trigger vs. latent root cause (5-Whys), contributing factors, and how to present them. Defines the RCA section format; its output is used in the before/after, recommendations, and reporting deliverables.

microsoft Skill

Use to build the before-and-after visual evidence for a Zava Learning incident. First classify the fault, then render the ONE visual that actually explains what changed — a before/after path or topology diagram for connectivity/config/RBAC faults, or time-series comparison charts for performance/availability faults — plus a state/metric delta table. Never plot a metric that does not tell the story. Its output is used in the RCA and the zava-reporting deliverables.

microsoft Skill

Use to keep the triggering PagerDuty incident in sync with the investigation and to close it out — acknowledge on engagement, mark the incident Mitigated the moment the live fix lands, post a concise symptom-only status/summary note (diagnosis, mitigation, links to the PR / Change Request), and resolve the incident only after the RCA and summary report are complete and recovery is verified. This is the PagerDuty communication and closure layer for a Zava Learning incident.

microsoft Skill

Use to produce the forward-looking recommendations and next steps for a Zava Learning incident — prioritized preventive, detective, and process actions with owners, target dates, and the risk of inaction. Its output is used in the zava-reporting deliverables.

microsoft Skill

Use whenever a Zava Learning investigation produces a durable fix that needs change management — after a GitHub PR is opened for an Infrastructure-as-Code or application code root cause, raise a ServiceNow Change Request referencing the PR and attach the RCA report. The single owner of ServiceNow Change Request and attachment operations.

pluginagentmarketplace Bundle

Test AI systems for code injection vulnerabilities including prompt-to-code attacks and agent exploitation

technickai-openclaw-config Bundle

Manage Asana tasks, projects, and workspaces via MCP

jiunbae Bundle

Runs an explicitly invoked pointer-driven review, plan, work, and feedback loop with independent multi-agent falsification, exact source fences, bounded dispatches, and concurrent-run safety. Supports a non-mutating audit/report mode and a user-authorized full implementation mode until convergence or 128 cycles. Use only when the user explicitly invokes `$rpf`; never trigger implicitly for ordinary review, planning, implementation, commit, or deployment requests.

technickai-openclaw-config Bundle

Make outbound phone calls via Vapi voice AI. Use when the agent needs to call someone on the phone for any reason (reminders, notifications, requests to businesses, or any task that benefits from a real voice conversation). Requires VAPI_API_KEY, a provisioned phone number, and an assistant configured in Vapi.

technickai-openclaw-config Skill

Intelligent task delegation — route to think with deep reasoning for hard problems, or Grok for unfiltered takes. Teaches when to escalate, how to pack context into sub-agent spawns, and how to communicate delays transparently. Default: handle directly on chat (thinking off). Escalate only when the quality gain justifies 30-90 seconds of silence.

technickai-openclaw-config Skill

Write effective prompts for LLM agents — system prompts, workflow instructions, skill files, and agent configurations. Use when creating or improving prompts that agents will execute.

bowen31337 Skill

Per-worktree boot script generator and instance launcher. Generates a self-contained bash script that starts an isolated application instance on a dedicated port, optionally isolates the database (PostgreSQL schema, SQLite file, or container), and polls a health endpoint until the instance is ready. Also supports booting the instance directly from Python without writing a script to disk. Use when: (1) starting an app server for an agent worktree so it doesn't collide with other concurrent agents, (2) generating a reproducible boot script to store alongside the worktree, (3) verifying an instance is healthy before running tests, (4) setting up database isolation per task, (5) scripting multi-agent environments where each agent needs its own running service. Triggers on: boot instance, start isolated server, per-worktree boot, generate boot script, health check, isolate database, launch app for agent, worktree isolation.

bowen31337 Bundle

Screenshot capture skill for recording application state as a visual artifact. Captures browser pages, desktop windows, or terminal output and saves them as timestamped PNG files (or base64-encoded inline artifacts). Use when: (1) documenting the visual state of a running web application, (2) capturing a UI before/after a code change, (3) recording terminal or CLI output as an image, (4) producing evidence artifacts for a bug report or test run, (5) snapshotting intermediate UI states during an automated agent workflow, (6) comparing layouts across breakpoints or themes. Triggers on: take a screenshot, capture the screen, screenshot the app, record the UI, capture application state, visual snapshot, screenshot artifact, capture browser, capture window, harness screenshot.

alexzio00 Bundle

Agent Discipline Engine — lock the goal, run PLAN→DO→VERIFY→FINALIZE→OUTPUT loop, detect success masquerading. Triggers: '/goal-lock', '/goal-lock quick', 'goal lock', 'task harness'.

alexzio00 Bundle

Audits whether a verification (eval/metric/experiment/holdout) actually secures independent external ground truth, or whether the designer, the model, and the scorer are just confirming each other in a circle — via a 21-pattern taxonomy. Read-only. Use before trusting any 'how we'll know it worked' — A/B tests, holdouts, scores, validation — especially when a result feels too clean or self-confirming. 한국어: '이 검증 순환논리 아닌지 봐줘', '이 평가 편파적이야?', '이 벤치마크 셀프체크야?'.

jiunbae Bundle

Synchronizes ML model files across servers. Supports rsync-based transfer with bandwidth control and checksum verification. Use for "모델 동기화", "모델 배포", "rsync 모델", "서버로 전송" requests.

jiunbae Bundle

Guides ML model benchmarks and evaluations. Measures inference speed, memory usage, and accuracy metrics. Use for "벤치마크", "모델 평가", "성능 테스트", "inference 속도" requests.

jiunbae Bundle

Open AIR Workbench to discover local Agent Skills, visualize and round-trip a Skill workflow as AIR (Agent Intermediate Representation), review a native Codex or Claude plan, inspect observable execution evidence, or promote a reviewed plan or trace into a Skill draft. Use only when the user explicitly asks for AIR Workbench, the legacy Workflow Studio, Skill-to-graph or graph-to-Skill conversion, visual workflow editing, plan approval, observable tracing, or plan/trace promotion. Do not trigger for ordinary Skill execution, general planning, or routine Codex/Claude requests.

dcjanus Bundle

用三个相互隔离的干净 subagent 并行做代码审查、由主 agent 判断审查意见价值、修复有效问题并提交推送，直到同一批三个 reviewer 都没有有价值审查意见。适用于用户要求 review/fix loop、clean review cycle、创建新 subagent 审查当前修改、反复 review 到没有问题、或“三个独立 reviewer 都没有有效建议”这类任务。

alexzio00 Bundle

Gate for deciding whether to adopt an external pattern (skill/agent/rule/plugin/MCP/prompt) into your system. Triggers: '/integration-intake [name]', 'should I integrate this', 'is this worth adopting', or sharing a GitHub repo link and asking what to do with it.

whitebeardit Skill

Safely update Jira cards by reading card information, identifying changes, and adding structured comments using the official Atlassian MCP. Use when updating Jira cards with progress, changes, or status updates.

bowen31337 Skill

Append-only agent progress log for tracking step completion within a plan. Agents call this skill to record timestamped entries as they start, finish, fail, or skip individual steps in an execution plan. All entries accumulate in docs/exec-plans/progress.md as a Markdown table — safe for concurrent writes from multiple agents. Use when: (1) starting work on a plan step, (2) marking a step done with a completion note, (3) recording a step failure with error detail, (4) skipping a step that is not applicable, (5) reviewing what steps have been completed so far, (6) generating a per-plan done/total progress summary, (7) coordinating progress visibility across parallel agents. Triggers on: log progress, append progress, mark step done, record step started, step failed, step skipped, progress entry, progress log, plan progress, track step, update progress, what steps are done, show progress, progress summary.

bowen31337 Skill

Shared agent state publisher and query tool. Agents publish intermediate results (discovered endpoints, schema changes, test results, or arbitrary structured data) into docs/exec-plans/shared-state.yaml so other concurrently-running agents can read them. All writes are protected by an advisory file lock for safe concurrent access. Use when: (1) an agent has discovered API endpoints that other agents need, (2) an agent has applied schema changes that downstream agents should know about, (3) an agent has run tests and wants to share pass/fail counts or coverage data, (4) any agent wants to leave structured breadcrumbs for agents that run later, (5) querying what intermediate state has already been published before starting a task. Triggers on: publish result, share intermediate result, discovered endpoints, schema changes, test results, shared state, agent breadcrumbs, inter-agent communication, query shared state.

bowen31337 Bundle

Load and present the most recent plan state for agent context handoff. Reads .claude/plan-progress.md (primary) or .plan_progress.jsonl (fallback) and formats the structured plan state — task, status, accomplished work, in-progress items, next steps, and search hints — so a new agent session can orient itself immediately. Search hints are presented as actionable pointers (file paths, grep patterns, key symbols) that the resuming agent verifies with its own Read/Grep/Glob tools. Use when: (1) starting a new session that continues previous agent work, (2) quickly checking what the current plan state is, (3) injecting saved plan state into an agent system prompt via the SDK, (4) bridging sessions after a context window limit or session restart, (5) presenting a handoff summary before a new agent begins. Triggers on: harness resume, load plan state, what was I working on, resume previous session, show plan progress, continue from last session, where did I leave off, read handoff, present context handoff.