AI & ML

forexgod21 Bundle

Activate whenever a multi-agent system or single complex agent fails and the question is WHERE and WHY — post-incident analysis of agent failures, debugging multi-step agent pipelines, "the agent did something wrong but I don't know which part broke," designing error taxonomies or failure telemetry for agent systems, or distributing responsibility for a failure across a delegation chain. Fire because the default failure analysis in agent systems is "the model hallucinated" — a non-diagnosis that blames the last visible step and fixes nothing, guaranteeing the failure returns wearing a different output.

forexgod21 Bundle

Activate whenever authorization travels through a chain of agents — agent A authorizes agent B which tasks agent C — and the question is what each link in the chain is permitted to do. Trigger on designing permission systems for multi-agent chains, delegation token or grant design, auditing who authorized an agent's action, provenance questions ("where did this agent get the right to do that"), and any system where permissions could silently widen as they travel. Fire because the natural drift of chained authorization is expansion — each hop interprets its grant generously — and only a structural narrowing rule holds the line.

forexgod21 Bundle

Activate whenever an AI agent or assistant operates with multiple instruction sources — system prompts, project instructions, knowledge files, memory, retrieved documents, and live user messages — and those sources can conflict. Trigger on designing agent instruction hierarchies, debugging "the AI ignored my instructions," resolving contradictions between stored context and current requests, or any multi-source configuration work. Fire because every multi-source system WILL generate conflicts, and a system without a written resolution order resolves them randomly — which users experience as unreliability.

forexgod21 Bundle

Activate whenever an AI agent persists information across interactions — long-term memory stores, RAG knowledge bases, user preference records, learned procedures, shared agent memory, or any system where today's stored content shapes tomorrow's behavior. Trigger on designing memory write paths, auditing what an agent has memorized, diagnosing behavior that degraded after specific interactions, or securing knowledge bases that agents retrieve from. Fire because persistent memory converts a one-shot manipulation into a standing compromise: a poisoned context window expires at session end, but a poisoned memory attacks every future session that retrieves it.

forexgod21 Bundle

Use this skill whenever the user needs to map, analyze, or design the trust and authority structure across a multi-agent AI system. Triggers when the user asks about agent hierarchies, delegation chains, trust relationships between agents, or says things like "map my agent network", "who has authority over what in my AI system", "design a delegation structure for", "I need to know which agent controls which", or "how should authority flow in my multi-agent setup." Always activate this skill when the user needs to visualize or audit how tasks, trust, and decision authority are distributed across multiple AI agents or human-AI teams.

forexgod21 Bundle

Claude continuously checks that its interpretation of a task remains aligned with the original human intent throughout execution — not just at the start. When drift between interpreted task and actual intent is detected, Claude surfaces it immediately rather than executing against a misaligned understanding.

forexgod21 Bundle

Activate whenever AI agents delegate tasks to other agents or humans DURING execution — runtime task handoffs, dynamic sub-agent spawning, mid-mission authority transfer, or any multi-agent system where delegation decisions happen live rather than at design time. Trigger on designing delegation safety checks, reviewing multi-agent systems for runtime failures, or any incident where a delegated task went wrong and nobody can say which safeguard should have caught it. Fire because design-time safety review cannot see runtime conditions — the delegation that was safe on the whiteboard executes in a world the whiteboard never met.

forexgod21 Bundle

Walden University MS/PhD academic execution skill for the YVYC Tempo Mission. Activate for all Walden coursework — discussion posts, response posts, papers, literature reviews, and research assignments. Inherits the full scholar-practitioner doctrine and extends it with Walden-specific APA 7 precision, rubric-dominant execution, and 90-day sprint discipline. This skill does not throttle, soften, or drift. It operates at PhD-level standard from first prompt to final submission.

forexgod21 Bundle

Activate whenever an agent task needs a precise, checkable specification before execution — defining what "done" means for an autonomous task, setting resource and time bounds, writing acceptance criteria an agent's output can be verified against, specifying sub-agent tasks in a delegation chain, or "the agent did something, but not what I meant." Trigger on task definition for autonomous agents, agent evaluation harness design, and any handoff where the receiver needs to know exactly what success looks like. Fire because the gap between what a user MEANT and what an agent OPTIMIZED is where autonomous systems fail most expensively — and an unspecified task is an invitation to that gap.

forexgod21 Bundle

Activate whenever an agent executes multi-step work, uses tools, spends tokens, or produces a deliverable the operator will need to audit, debug, or trust. Governs action logging, cost tracking, decision-quality review, and replay-for-debugging. Converts agent work from an honor system into an inspectable record.

forexgod21 Bundle

Activate whenever an AI agent personalizes its behavior using stored knowledge about a user — memory-equipped assistants, agents that learn preferences, systems that infer user intent from history, or any deployment where "what the agent knows about you" shapes "what the agent does for you." Trigger on designing personalization features, auditing how user memory influences agent decisions, privacy reviews of personalized agents, or diagnosing an agent that treated a request differently based on who was asking. Fire because personalization is a safety surface, not only a UX feature: benign stored details can bias an agent into legitimizing requests it should refuse, or refusing requests it should grant.

forexgod21 Bundle

Use when deciding which context-sensitive human interactions an AI agent may handle, must flag, or must escalate to a human. Triggers on questions about AI handling emotionally or relationally complex situations — "how should my agent handle upset customers", "when should AI hand off to a human", "what interactions should AI never handle alone" — or any request to define the boundaries of AI social interaction.

forexgod21 Bundle

Activate whenever an agent's context gets compressed, summarized, truncated, or grows long enough to degrade performance — designing compaction and summarization strategies, multi-hour agent sessions approaching context limits, "the agent forgot what we decided," degrading quality as conversations grow, or handoffs where one context must become a smaller one. Trigger on any long-context or context-management design. Fire because compression and length are the twin quiet killers of long-horizon work: compaction loses load-bearing facts by summarizing them away, and raw length loses them by burying them — and both failures are silent until the agent acts on what it no longer effectively holds.

forexgod21 Bundle

Activate whenever a user corrects an AI agent's behavior, output pattern, formatting, tone, or approach — and the correction should hold for the rest of the session and beyond. Trigger on any user correction ("stop doing X," "I told you to Y," "why are you doing this again"), on designing feedback-integration behavior for agents, and on repeated-mistake diagnosis. Fire because the default failure of AI systems is treating corrections as one-time patches instead of standing rules — forcing users to re-issue the same correction until they stop trusting the system.

tomjiu Skill

Use when building batch automation, agent pipelines, scheduled jobs, or multi-step unattended scripts. Prefer durable plans, checkpoints, and authorized scopes only.

tomjiu Skill

This skill should be used when the user asks about browser automation, testing web pages, scraping content, filling forms, taking screenshots, or monitoring console/network activity. Activates for E2E testing, web scraping, form automation, or debugging web applications.

doronkatz Skill

Use this skill when you need to store or search MEMANTO persistent memories. It defines mandatory guidelines for best practices, memory types, confidence levels, tagging, and patterns for effective agent memory usage.

doronkatz Bundle

Use when diagnosing or repairing Codex MCP server startup failures, connector auth errors, missing tools, plugin cache drift, skill-loader warnings, broken skill symlinks, or config.toml syntax and path issues.

doronkatz Bundle

Use the Appllama MCP (mcp.appllama.io) well — research real top-grossing mobile apps, their screens, flows, and UI elements, then build from what you learn. Load when the Appllama MCP is connected and the task involves building a mobile app or screen, researching app design patterns, studying onboarding/paywall/feature flows, improving an existing screen, or whenever an appllama_* / search_apps / list_app_screens tool is available. Covers the tool map, pagination, expiring media, and the full build-from-research playbooks.

doronkatz Bundle

Official skill for integrating Firebase AI Logic (Gemini API) into web applications. Covers setup, multimodal inference, structured output, and security.

doronkatz Bundle

Produces a topic-segmented post-meeting summary for attendees with decisions highlighted and actions captured inline per topic (plus a consolidated action view at the end). Auto-populates topic skeleton from a sibling meeting-agenda when available and reconciles planned vs. actual topics. Accepts transcripts from Zoom, Meet, Otter, Fireflies, Krisp MCP, or manual notes; runs on variable-quality input without blocking. For synthesizing user research interviews across participants, use discover-interview-synthesis.

doronkatz Bundle

Run a repo-wide cross-cutting governance audit via the pm-skill-auditor sub-agent. Aggregates the enforcing validator suite, re-derives aggregate counters, and surfaces cross-cutting issues no single validator catches, graded P0/P1/P2/P3 with a machine-readable status. Use for pre-release readiness checks or a periodic repo health audit.

doronkatz Bundle

Establishes a project's quality bar as a written contract and stops agents quietly lowering it. Interviews the user on which dimensions matter, supplies sane default thresholds when they have no number in mind, records everything in CONSTRAINTS.md, and watches the diff for a weakened bar — new @ts-ignore or eslint-disable suppressions, skipped or deleted tests, assertions stripped out, unimplemented stubs, thresholds edited down. Use when no quality bar is written down, when the user says "set up constraints" or "define our standards", when an agent keeps silencing checks or skipping tests to get to green, when you need a coverage or performance threshold and don't know what number to pick, or when an agent writes more code than anyone will read.

hybridaione Bundle

Redact, anonymize, sanitize, or remove PII locally with Distil-PII and llama.cpp; keep personal data and secret values out of model context, logs, and chat.

aydincan Bundle

Türk hukukçusunun ev içi atıf düzeni: içtihat yalnızca mahkeme, daire, esas/karar numarası, tarih ve doğrulanabilir kaynakla; mevzuat madde/fıkra/bent ile; doktrin yazar-eser-sayfa ile. Model hafızasından karar zikretme yok. Bu beceriyi Türk Hukukunda Atıf ve Kaynak Hijyeni alanındaki sorular, olaylar ve belgeler için kullan (atıf, kaynak hijyeni, Yargıtay, Danıştay, AYM, esas karar numarası, pinpoint atıf, içtihat doğrulama). Alan dışı konularda tetiklenme.

aydincan Bundle

Patent ve faydalı model: patentlenebilirlik şartları (yenilik, buluş basamağı, sanayiye uygulanabilirlik), istem yorumu, hükümsüzlük, çalışan buluşları ve patent hakkına tecavüz; SMK uygulaması. Bu beceriyi Patent ve Faydalı Model alanındaki sorular, olaylar ve belgeler için kullan (patent, faydalı model, yenilik, buluş basamağı, istem, hükümsüzlük, çalışan buluşu, zorunlu lisans). Alan dışı konularda tetiklenme.

harsh817 Bundle

Design or audit the data model for one feature, including entities, fields, constraints, indexes, ownership, migrations, and persistence boundaries. Use before implementing database or schema changes.

harsh817 Bundle

Inspect an existing repository before implementing a feature. Use when a coding agent needs to understand project structure, current behavior, owner modules, tests, documentation, git state, and risk before planning code changes.

ryankolean Bundle

Converts a working Claude prompt into a permanent, installable SKILL.md file with proper frontmatter, activation triggers, execution rules, and examples. Then automatically add it to the Summit Custom Skills table in the repo README and push to GitHub. Chains naturally after prompt-architect. Activates when the user says "make this a skill", "prompt to skill", "install this permanently", or when a prompt-architect output is confirmed working and the user wants to keep it.

ryankolean Bundle

Translates a plain-English task description into a production-quality Claude prompt with structure, examples, and output formatting. Activates when the user says "write me a prompt for this", "turn this into a prompt", "prompt-architect", or when they describe what they want Claude to do but can't get the right output. Bridges the gap between knowing what you want and knowing how to ask for it.

ryankolean Skill

Package the relevant content from the current Claude.ai chat into a clipboard-ready, self-contained prompt for a fresh Claude Code session. Activates when the user says "chat-to-code-prompt", "send to Claude Code", "make a CC prompt", "package this for Claude Code", "handoff to CC", or any variant. Produces one of three formats — task, plan, or context — chosen by the user at activation. Output is a single fenced markdown block at the end of the response so it can be tap-and-hold copied on mobile.

yogsoth-ai Bundle

Structured data extraction from deep-read papers — produces comparison tables (method, dataset, metrics, results, limitations). Used by systematic-survey and deep-survey.

yogsoth-ai Skill

Plan the statistical synthesis approach — model selection, heterogeneity strategy, and reporting

yogsoth-ai Bundle

Plan funnel plots, Egger's test, trim-and-fill, p-curve, and selection model analyses for publication bias

jorinyang Bundle

Configure and troubleshoot Hermes Agent on native Windows (post-WSL).

jorinyang Skill

Use when designing or evaluating agent-native CLI tools, creating skill harnesses, or deciding which bridge pattern to use for accessing a service/app from Hermes.