LLM Mart Basic

Use when the user says "loop me" or asks to design a recurring workflow. Don't use for remote, credential, publish, deploy, or irreversible changes.

Use when the user needs a compact read-only current-work view from Git state, recorded test evidence, and optional graph.yaml. Not for tasks that need source or remote-system changes.

Use when a project is between phases, the author asks what to do next, too many threads are open, or work needs re-entry. Not for gating whether one named task may proceed.

Use when a user commits to a direction and asks to plan, brief, or research it; modes score, breakdown, shape, visual. Not for codebase audit: use plan-review. Not for four-phase review: use autoplan.

Use when a design dispute has at least two live interpretations and the caller wants worlds made explicit or a plain-language recommendation. Not for selecting a design or source/remote changes.

Use when asked to judge whether to adopt, switch, reject, or revisit technology, library, pattern, or architecture, or a second opinion. Not for scoping: use brainstorm. Not for forks: use decide.

Use when a user asks for a gut-check on a decision or action, or asks whether enough is known to proceed. Not for numeric confidence scoring.

Use when the user asks for a flattened view of roadmaps and next actions. Not for multi-session route planning: use wayfinder.

Use when the user wants to harden a chosen but tentative artifact into one durable result. Not for remote, credential, publish, deploy, or irreversible changes.

Use when starting a project or feature, requirements are unclear, or a change crosses modules. Not for implementing from an existing spec: use spec-driven-implementation.

Use when work has distinct modes and the user wants states, events, guards, outcomes, illegal transitions, not a prose todo list. Not for remote, credential, publish, deploy, or irreversible changes.

Use when a user wants to design an abstraction boundary that collapses a complex implementation into a simpler interface without leaking internal state. Not for implementation.

Use when user wants an async questionnaire, a discovery questionnaire, or a knowledge gap needs answers outside the repo. Not for direct conversation: use askme. Not for agent research: use research.

Use when settled conversation decisions need synthesis into an agent-ready implementation spec, stopping before publication. Not for turning plans into tickets: use to-tickets.

Use when a settled plan needs implementation tickets published as blocker-linked slices, tracer bullets, or expand-contract sequencing. Not for implementation: use work.

Use when a message contains `TODO ADD: <requirement>`. Not for deepening coarse lists: use todos-enhance. Not for resyncing lists: use todos-update. Not for retitling/reordering/completing todos.

Use when tasks are too vague, read as headings, or the user asks to sophisticate the todos. Not for stale reconciliation: use todos-update. Not for adding requirements: use todo-add.

Use when user asks to update todos, resync the task list, say what to do next, or plan and tree have drifted apart. Not for coarse lists: use todos-enhance. Not for adding requirements: use todo-add.

Use when a greenfield project or large feature build will not fit in a single agent session. Don't use for implementation, remote credential changes, or work that fits in one session.

Use when the user wants adversarial stress-testing of a proposed architecture, structure, or shape. Not for tasks that require source or remote-system changes.

Fourteen posts of being wrong in production, compressed to checkboxes

Healthy nodes, a quiet network, 300 restarts in three days, and a latency budget measured in milliseconds

Discovery worked. Ping worked. Every TCP connection timed out, and later the tunnel only worked when someone had a terminal open.

Every VM came back. The cluster did not. Declarative systems converge on config, and the datapath isn't config.

A surprising share of AI-in-the-terminal failures aren't the AI. They're zsh, and a version of bash from 2006.

A Claude Code plugin turns standalone project configuration into a namespaced, installable extension that teams and communities can update as one unit.

None of the safety came from the model. It came from six boring habits.

Skills package instructions and references. Subagents run work in a separate context and return results. They solve different problems and can be composed deliberately.

Six hours in, one step left, everything green, and the incident that didn't happen

CLAUDE.md carries persistent project context. Skills load reusable procedures when relevant. Separating stable facts from task-specific workflows keeps both easier to maintain.

Twenty minutes recovering secrets that never existed, and the one sentence from a human that ended it

An API request routing a model's tool call through an approval gate to a remote MCP server

31 config keys, two audits, and why the first one was wrong in both directions

The official MCP Registry stores standardized server metadata rather than package code. Publishers verify a namespace, describe installation or remote access, and submit immutable versions.

Everyone looks at the Dockerfile. The file that actually leaked the key was the project file.

Remote MCP authorization uses established OAuth standards, but secure integration still requires issuer validation, least-privilege scopes, protected token handling, and server-side enforcement.

"Copy it over and switch the reference" is two steps, and the outage lives in the one nobody checks

stdio fits local processes and prototypes. Streamable HTTP fits hosted services and shared integrations. The right choice follows where the capability runs and who must reach it.

The most important rule wasn't about what I could change. It was about what I was allowed to display.

Tools perform operations, resources expose readable context, and prompts provide reusable templates. Choosing the correct primitive makes an MCP server easier to understand and govern.

Run the reliability scorecard for one service, or every connected service if omitted - error-budget burn rate where an SLO is defined, trend reporting otherwise

Generate the current on-call handoff brief - what's paging, what's escalated without an owner, last-shift history, and known-flaky alerts to watch

Draft a blameless postmortem for a given incident (by ID or name), or the most recent significant incident within a time window

List all devices at a Domotz-monitored site

Find a Domotz device by name, IP address, or MAC address

Overview of a Domotz site's network health

Pull full detail for one Checkpoint Harmony Email detection and the message behind it

Restore quarantined mail to its recipients in Checkpoint Harmony Email, with task polling

Find messages in Checkpoint Harmony Email by sender, subject, attachment hash or quarantine state

Sweep security events in Checkpoint Harmony Email by type, state, severity and date range

Get summary of recent phishing and training campaigns from KnowBe4

Get security awareness metrics for a KnowBe4 group

View phishing campaign results and click rates from KnowBe4

Check training completion status for users or groups in KnowBe4

Get risk score and risk history for a KnowBe4 user

Full health check for one Proofpoint Essentials customer org

Resolve a Proofpoint Essentials customer org by name or domain and show its details

View recent TAP threat events including blocked messages, delivered threats, and click activity

Decode a Proofpoint URL Defense rewritten URL back to the original URL

Deep-dive threat investigation with forensics, campaign context, and remediation options

Your car as a chat-room agent: Raspberry Pi 5 + dashcam + local AI. CodeWatch's sibling for the garage.

The enterprise gateway for autonomous agents. Identity management, per-channel isolation, credential vault, per-session tamper-evident audit log.

AI agent for Home Assistant — talk to your home, create automations in plain language, analyze cameras with face recognition, and get proactive anomaly alerts.…

A desktop app to prototype agent ideas, inspect every harness step, replay failures, and evaluate performance, all in one place. Local-first, cloud-ready for ma…

Local-first AI memory you can see, edit, and override — portable across Claude Code, Codex, Cursor, Windsurf, and other MCP coding tools.

Web & desktop (Electron) client for the OCTO open workplace — one React + TypeScript codebase shipping browser and PC surfaces, with first-class AI agent UX.

同花顺官方 A股金融数据服务，提供股票实时行情、历史行情、财务报表、指数、板块、涨停等数据，适用于 AI Agent、量化研究和应用开发，支持 API、MCP、CLI 和 Python。Official Tonghuashun (HiThink) A-share financial data service provi…

Open-source desktop SQL workspace for PostgreSQL, MySQL/MariaDB, SQLite and 15+ more databases like DuckDB, ClickHouse, Redis and Firestore. Built-in MCP server…

🤖 Local-assist BOSS Zhipin CLI for AI agents — search, welfare filtering, shortlist, JSON-envelope output; low-risk & compliant by default.

The local-first sidebar AI agent for ComfyUI — runs on your own Claude OR ChatGPT subscription (no API keys, no extra LLM costs). Drives your live graph: edits,…

Persistent memory for Claude Code — identity, context, and continuity across sessions

🔬 A curated collection of 23,000+ agent skills for empirical research across 8 social science disciplines. | 精选 23,000+ AI Agent 技能库，覆盖8大社会科学学科的实证研究。CoPaper.AI…

A curated list of plugins, skills, MCP servers, patch/profile layers, orchestrators & UIs for DeepSeek Harness (DSH). Visualization · PPT · Coding · Agents · Lo…

World Memory Protocol.

Run your own organization of agents.

The most RAM efficient harness

AI-native, free, open-source alternative to Jira, Trello, ClickUp & Monday. Built for Scrum teams where humans and AI agents collaborate as equals — on the same…

Free open-source desktop app that scrapes JAV metadata and generates NFO + cover art for Jellyfin, Emby & Kodi. No Docker, no CLI — one-click install on Windows…

A cross-platform desktop app to manage Agent Skills in one place and sync them to multiple AI coding tools’ global skills directories — “Install once, sync ever…

天枢 (Tianshu) 是一个基于harness工程的终端编程智能体运行时（Tui X Gui），针对DeepSeek V4 做了前缀缓存工程优化（长会话实测稳态命中率 97–99%）和深度适配。它跳出了传统 AI 编程助手把大模型仅当成“工具”的局限，基于认知虚拟机 (CVM)、自感知层和信息素（Stigmergy…