AI & ML

synthesisengineering Bundle

Fail-closed pre-send enforcement for agent-drafted correspondence. A PreToolUse hook blocks every message-sending or draft-creating tool call unless the outgoing text passes a deterministic register scan AND a fresh, single-use grounding ledger — sha256-bound to the exact message — attests that the composing agent read the full thread, searched prior correspondence, and mapped every factual claim to a source. Use when setting up, debugging, or composing under the guard; when a send is blocked; or when asked about message grounding, voice enforcement, or pre-send gates.

nahid-sparktales Bundle

See what an agent actually did — one trace per run with nested model, tool and retrieval spans, token and latency accounted per step, and failures clustered by mechanism instead of read one at a time. Use when an agent misbehaves in ways you cannot reproduce, when cost or latency is unexplained, when "it sometimes fails" is the whole bug report, or before writing evals when you do not yet know which failures exist. Not for judging whether an output is correct (agent-evals), and not a replacement for the application's own monitoring.

nahid-sparktales Bundle

Get parseable, trustworthy structured results out of a model — schema design, the enforcement mechanism the provider actually offers, validation at the boundary, bounded retry that feeds the error back, and recognising where a schema stops buying correctness. Use when a model's output feeds code rather than a human, when parsing keeps failing or fields come back plausible-but-wrong, or when designing an extraction, classification or grading step. Not for designing the tools an agent calls (tool-design), and not for free-form prose quality.

nahid-sparktales Bundle

Instrument a service so the questions asked during an incident are answerable from data already being collected — rate, errors, latency distribution and saturation per route, structured events carrying a correlation id that survives process and queue boundaries, and alerts on symptoms users feel. Use before a service or a new critical path goes to production, after an incident that ended in "we had no data for that", or when adding a dependency, queue or job whose failure would be silent. Not for fighting a live outage (incident-response), not for profiling a known-slow path (performance-profiling), and not for tracing agent or LLM runs (llm-observability).

nahid-sparktales Bundle

Write or revise a prompt so it holds up — output contract, instruction placement, examples that earn their place, an escape hatch for bad input — and measure the change against a saved set of cases instead of one good-looking run. Use when a prompt is being authored or patched, when output is inconsistent or the wrong shape, when a model or version changes, or when someone reports a prompt as fixed. Not for scoping an agent's job and tools, not for deciding what material to load into the window, and not for model selection or fine-tuning.

nahid-sparktales Bundle

Decide what an authenticated caller may do — pick the permission model, put the check at one enforcement point close to the data, and default to deny. Use when adding roles or permissions, scoping data per tenant or per owner, building an admin-only path, reviewing an endpoint that trusts a client-supplied id, or when someone reports seeing data that is not theirs. Not for establishing identity (that is authentication), and hiding a control in the UI is never the enforcement.

rjxshr1 Bundle

Prepare a model file, link or modeling brief for FDM printing through sizing, slicing, authorized printer dispatch and result review. Use printable-modeling when geometry must be created or edited.

synthesisengineering Bundle

Discipline for catching the lazy-shortcut antipattern in AI-assistant output. Makes the costume vocabulary explicit so agents recognize when their drafts have slid into deferral, dismissal, or false consultation. Includes the constraint-first protocol, sub-agent dispatch and acceptance hygiene, and a pre-response self-check. Use when asked to: avoid shortcuts, audit for laziness, check for deferral, enforce best solution, no shortcuts, anti-shortcut, constraint-first, sub-agent hygiene, costume vocabulary.

synthesisengineering Bundle

Start or resume any synthesis project in any harness on any machine with full verified context: classify the session (continuing, fresh, or wrong-project paste), load index plus CONTEXT plus REFERENCE plus recent sessions, respect live foreign claims, surface cross-machine changes, and upgrade stale project formats. Use when pasting a console resume prompt or (re)opening project work.

synthesisengineering Bundle

Agent-native skill installer and manager for the synthesis skills ecosystem. Handles installation, drift detection, synthesis merge for conflicts, provenance tracking, and cross-repo coordination. Use when asked to: install skills, update skills, check skill drift, manage skills, skill status, skill inventory, sync skills.

synthesisengineering Bundle

Generate a structured writing voice profile from sample texts and diagnostic questions. Outputs an agent-instruction voice section that other skills automatically consume. Use when asked to: create voice profile, analyze writing style, extract voice, profile my writing, build voice section, writing DNA, style analysis.

nahid-sparktales Bundle

Decide what actually occupies the model's window — progressive disclosure through an index, retrieval versus inlining, what compaction must preserve, and what loading everything costs. Use when a prompt or agent pulls in many files, docs or tool transcripts, when quality decays over a long session, when choosing between fetching at runtime and pasting up front, or when a context has to be trimmed or summarized. Not for the wording of the prompt, not for an agent's job and tool boundaries, and not for picking retrieval infrastructure.

nahid-sparktales Bundle

Attack and harden an existing auth surface — session fixation and rotation, token verification, horizontal and vertical privilege escalation, password reset and account recovery, MFA bypass. Use when reviewing login, session, token, reset, invite, impersonation or role-elevation code, when someone reports seeing another user's data or an account takeover, or when auth changes are about to ship. Not for designing the login mechanism or permission model in the first place (authentication, authorization), not for infrastructure IAM, and never run against a system you have not been told you may test.

catherine1401 Skill

Create or update an agent skill with focused instructions and only the resources it needs. Use for skills; not policies, plugins, or project implementation.

nahid-sparktales Bundle

Secure an agent system as a permission surface — what authority each tool call runs under, where the confused deputy sits, which controls are enforced outside the model and which are only prompt text, and how far one bad call reaches. Use when granting an agent tools or credentials, wiring in MCP servers or subagents, reviewing an agent that acted beyond what the requester could have done, or before letting an agent touch a shared or production system. Not for wording the ingestion trust boundary in detail (prompt-injection-defense), not for scoping what an agent is for (agent-design), and it never grants an agent permission it did not already have.

synthesisengineering Bundle

The most comprehensive open-source slop detection system available. Catches AI-generation patterns by model family (Claude, GPT, Gemini, Llama, Grok, DeepSeek, Mistral, Qwen), substance and depth failures (beautiful word salad), and the full v3.1.0 catalog refreshed. Includes causal-layer attribution, combined-signal fingerprints, two-axis calibration, ESL safe-harbor, and zone-conditional detection (artifact mode vs full-response mode). Spans the entire LLM era through the compounding-archive principle. Use for content quality, slop detection, AI content auditing, editorial review, content improvement, and publishing standards.

synthesisengineering Bundle

Configure an LLM as a creative writing coach and editor. Use for creative writing, writer setup, configure writer, creative assistant, content creation coaching, storytelling assistance, and writing improvement.

nahid-sparktales Bundle

Treat everything an agent reads but did not author as data rather than instructions — an explicit trust boundary, a tool set fixed before content is read, consequential calls gated on the user, and destinations that cannot be chosen by the content. Use when an agent reads web pages, retrieved documents, emails, tool results, file contents or another agent's output and can also take actions, when adding retrieval or new tools to an agent, or when reviewing an agent for injection exposure. Not for authentication and authorization design, not for secret management, and never satisfied by "the model did not fall for it".

btsouth Skill

Working through Toolport

chinfi-codex Bundle

Session-handoff skill. Compacts the current conversation into a self-contained handoff prompt for a fresh session and prints it inline as one copyable block — writes no files. References existing artifacts (./docs/, ./dev/, commits, issues) by path instead of duplicating them, and names the skills the next session should invoke. Invoke explicitly with the next session's purpose as argument.

markus-global Bundle

Chrome DevTools browser automation via MCP - control a live Chrome browser

markus-global Bundle

Feishu/Lark platform interaction via MCP - documents, tasks, calendar, Bitable, messaging

anorth Skill

Reviews local code changes via a subagent using the review-diff skill, then triages findings and fixes clear problems directly. Use when the user asks for a sub-agent review, a review with a specific model, or parallel reviews with multiple models.

acryldev Skill

Use when the user wants a new model provider, a swappable capability, an agent preset or persona, or a Desktop profile feature.

demondamon Skill

AgenticX zero-to-hero quickstart guide. Use when the user wants to get started with AgenticX, create their first project, build their first agent, or run their first workflow. Covers installation, project scaffolding, agent creation, task execution, and CLI basics.

demondamon Skill

Guide for creating custom tools in AgenticX including function decorator tools, MCP tool integration, tool registries, and remote tool access. Use when the user wants to create tools for agents, integrate external APIs as tools, build MCP servers, or extend agent capabilities with custom functions.

chinfi-codex Bundle

Independent code review powered by open-code-review (ocr). Modes: delegate (default — ocr filters files and resolves rules, a fresh independent conversation reviews; no LLM endpoint needed), local, ci (GitLab CI auto-review on MR push). Review-only: never modifies code, writes the review report only.

chinfi-codex Bundle

Execution skill for one confirmed issue. Normally auto-dispatched by the main session as a fresh host-agent conversation (economy-tier model): TDD in a dedicated worktree, per-round scope-drift checks with one-way verify-tier escalation, on-disk evidence for the ticket's tier gate, then independent /ai-review, BLOCKER fixes, and the merge request made ready. Main session supervises through merge — no human gate. Success is judged by on-disk evidence, never by self-report.

chinfi-codex Bundle

One-time per-repo setup for the dev-stage skills. Detects ocr / glab / GitLab CI readiness, collects tracker mode, main branch, quality gate, redlines, repo level, and open-code-review mode (default: delegate — review runs as a fresh independent conversation inside the host agent), then writes ./dev/agents-config.md plus CI/rule seeds. Run once before /tech-spec, /issue-split, /implement, /ai-review.

chinfi-codex Bundle

Documentation-first ticket splitting skill. Turns a confirmed tech spec into tracer-bullet tickets with blocking edges, auto-checkable DoD commands, and per-ticket verify-tiers from agents-config.verify-policy (integration tickets forced to full), gets gate-2 human confirmation, publishes — then the main session auto-dispatches each ticket as a fresh host-agent conversation and supervises until merge with no further human gates. Default mode is DOC_MODE.

swlfaye-sketch Bundle

每日清点国内外主流 AI 平台的免费额度——免费 token、Agent/Coding 工具赠送额度、视频与生图模型试用期、学生教师认证福利，汇总成一份「今天能薅什么、什么快到期」的清单，并可选推送到手机。当用户要求「领赛博鸡蛋」「看看今天有什么免费的额度」「免费 token 汇总」「哪些模型免费」「AI 福利活动」「把清单推给我」，或需要维护这条产线时使用。

ly87ing Bundle

Shape a solution before anything is built or edited — an architecture or migration scheme, a service split, merge, or retirement, an upgrade plan, a fix approach with several candidates, or a comparison of approaches. Use when asked to analyse how something should be changed, to think from zero about the right design, to compare options and say which to follow, to judge whether a scheme is complete or has gaps, or to turn a many-constraint discussion into one plan. Collects every constraint first, inventories live versus dead capabilities, names the deployment model each argument assumes, walks a blast-radius checklist, grades claims as observed, inferred, or unverified, and briefs the implementer. Not for editing the code once the scheme is agreed, nor for proving a built change works.

demondamon Skill

Guide for creating persistent Near desktop digital avatars (数字分身) via natural-language interview and the create_avatar tool. Use when the user wants to create an avatar, digital twin, specialist agent persona, or add someone to the avatar sidebar.

demondamon Skill

Guide for designing and running AgenticX workflows including sequential pipelines, parallel execution, graph-based orchestration, conditional routing, and trigger services. Use when the user wants to create workflows, orchestrate multiple agents, design agent pipelines, or set up complex multi-step processes.

triangle-int Skill

Use when working with the gog CLI for Google Workspace automation, especially when an agent needs JSON output, auth preflight, command guards, scoped accounts, or safe Google API reads/writes across Gmail, Calendar, Drive, Docs, Sheets, Slides, Forms, Apps Script, Contacts, Tasks, People, Groups, Keep, or Admin.

atomicstrata Bundle

Use the installed AtomicMemory CLI for memory search, ingestion, packaging, diagnostics, and agent-safe JSON output.