“elf”的搜索结果

Principled action selection for autonomous agents using Expected Free Energy (EFE) under the Active Inference framework. Each candidate action is scored as G(n) = w_prag * (-E[p_n] * damping) + w_epist * (-Var[p_n] * novelty(n)), where agent beliefs are maintained as per-node Beta distributions and updated via Bayesian observation. Selection is a softmax over -gamma * G, making exploration an automatic consequence of uncertainty rather than a hardcoded epsilon. Precision weights w_prag and w_epist self-adapt via surprise history, shifting the agent between exploitation and exploration regimes without any manually tuned schedule.

中文简介正在准备中。当前显示原文说明。

Operationalizing an expungement website — user intake flows, document generation pipelines, court filing integrations, payment processing, state-by-state workflow variations, customer support for sensitive populations, and the handoff between automated systems and human legal review. Activate on 'expungement operations', 'legal tech SaaS', 'court filing integration', 'document automation pipeline', 'expungement business', 'legal service workflow', 'multi-state legal operations'. NOT for expungement law itself (use national-expungement-expert), web design (use legal-tech-web-design), or legal document drafting (use recovery-app-legal-terms).

中文简介正在准备中。当前显示原文说明。

Classify every destructive/gated action an agent body can attempt (git, filesystem, network, shell, github) into block/approve/allow tiers, and audit whether the pre-tool and post-tool governance backing that classification is real: blocked actions proven zero-side-effect by a negative fixture, every denial backed by a receipt and a transcript event, every block-tier denial paired with a safe alternative, and no unmanaged or same-UID body ever marked "contained." Use when building or reviewing Agent Harbor's C5 governance gate (destructive git blocker, approval request/result, denial receipts), gating a new tool surface before it can run destructive commands, or auditing whether an existing "destructive-action blocked" claim actually has evidence behind it. NOT for proving an existing sandbox boundary actually contains an adversary once an action is isolated (use sandboxed-adversarial-test-harness), deciding where in a DAG to place a human review node or designing the approval UX itself (use human-gate-designer), or designing the general work-receipt schema across an entire agent task (use agent-work-receipt-designer).

中文简介正在准备中。当前显示原文说明。

Expert in developer tool documentation as a product. Activate on: docs site, getting started guide, API reference, developer documentation, Docusaurus, Mintlify, Nextra, ReadMe, migration guide, troubleshooting page, docs information architecture, documentation testing, llms.txt. NOT for: marketing copy (use seo-content-blogging), API design itself (use api-architect), general technical writing (use technical-writer).

中文简介正在准备中。当前显示原文说明。

Expert in cluttering -- the lesser-known fluency disorder characterized by rapid/irregular speech rate, excessive disfluencies, and frequent ADHD comorbidity. Covers differential diagnosis from stuttering, assessment protocols, therapy approaches (rate control, self-monitoring, pragmatic awareness), technology-assisted intervention, and AI tools for people who clutter. Activate on 'cluttering', 'rapid speech disorder', 'speech rate too fast', 'speech disfluency not stuttering', 'disorganized speech', 'ADHD speech problems', 'cluttering vs stuttering', 'speech intelligibility', 'mazed speech', 'telescoping words'. NOT for stuttering-only treatment (overlap is common but they are distinct), hearing disorders, or voice disorders (pitch, resonance).

中文简介正在准备中。当前显示原文说明。

Design and build mathematically rigorous, high-density, interactive data graphics and streaming telemetry visualizations following Mike Bostock's D3.js and Observable methodologies. Use when crafting custom SVG/Canvas charts, reactive streaming time-series, multi-dimensional brush-and-link coordinates, small multiples, or topological swarm graphs. NOT for generic dashboard template clones, basic spreadsheet charts, or off-the-shelf low-effort wrapper libraries.

中文简介正在准备中。当前显示原文说明。

Fundamental patterns for effective agentic behavior. Teaches decomposition, tool orchestration, error recovery, context management, quality self-assessment, and knowing when to stop. Model-agnostic principles that make any agent more effective regardless of domain. Activate on: "how should I structure this agent", "agentic workflow", "agent patterns", "multi-step task", "tool orchestration", "/agentic-patterns", "decompose this", "agent best practices", "chain of actions", "when should the agent stop", "agent loop design". NOT for: creating agent infrastructure (use agent-creator), building DAGs (use jury_rig-architect), specific tool implementation.

中文简介正在准备中。当前显示原文说明。

Discipline an AI agent needs to work an issue/task tracker (Jira, Linear, GitHub Issues) as the shared source of truth: pull the right next item instead of inventing work, search before creating to avoid duplicates, write actionable items with checkable acceptance criteria, keep status transitions honest, and link work to items via branches/PRs/commits so progress is legible. Use when an agent picks up tracker work, files a new issue, updates status, or closes an item. NOT for authoring the PR itself (use agent-pr-authoring), roadmap-level planning and prioritization (use legible-roadmap-with-sidequests), or the mechanics of multiple agents coordinating on the same codebase (use multi-agent-coordination).

中文简介正在准备中。当前显示原文说明。

Design RL simulation sandboxes, trajectory datasets, QLoRA/LoRA adaptation plans, eval harnesses, and rollback unhooks for agents learning specific coding behaviors beyond the base model. Use when building or reviewing tool-use training loops, offline trajectories, reward functions, self-improvement harnesses, or Port Daddy agent behavior curricula. NOT for generic ML tutorials, full model training operations, or production deployment without eval gates and safety unhooks.

中文简介正在准备中。当前显示原文说明。

Verify every visual-evidence artifact (screenshot/GIF/recording) attached to a PR carries a provenance manifest binding it to real daemon-backed truth — daemon port, run id, transcript head hash, agent node id, commit, and an honest real/fixture/mock source label — and that operator-control-panel PRs cover the required set of proof states. Use when gating a PR's proof artifacts before merge, auditing whether a "LIVE" visual claim is backed by a real daemon run versus a reused or fixture-backed mock, or defining the required state-coverage set for a control-panel change. NOT for the capture technique itself — headless Playwright, `screencapture -x -l`, non-interruptive capture (use port-daddy-agent-skill's visual-evidence doctrine), designing the receipt body schema an artifact attaches to (use agent-work-receipt-designer), or the broader dogfood/Potemkin product-quality bar (use multi-agent-authoring-product-bar).

中文简介正在准备中。当前显示原文说明。

Install Port Daddy's bounded interactive coordination hooks ("tentacles", from the Giant Squid Harness) into the INTERACTIVE sessions of agentic coding CLIs (Claude Code, Codex, Gemini, Antigravity/agy) by auto-detecting which CLIs are installed and writing PER-PROJECT hook config behind a runtime gate, so coordination fires only inside a Port Daddy project while the daemon is running — never machine- wide-always-on. Use when the user says "install port daddy hooks", "agent CLI hooks", "interactive session hooks", "make coordination hooks fire in interactive sessions", "codex/gemini/agy hooks", or "auto-detect agent CLIs". NOT for generic software/package installation; NOT for installing the Port Daddy daemon or `pd` CLI itself (delegate to the `port-daddy` skill / `pd setup`); NOT for MCP server setup (delegate to `pd mcp install`); NOT for authoring new skills (delegate to `skill-architect` / `skill-creator`); NOT for editing the tentacle scripts themselves (owned by the Giant Squid Harness program; this installer layer is pd-adr-090).

中文简介正在准备中。当前显示原文说明。

The staged pipeline that turns any API, provider payload, scrape, or computed result into registered, rendered platform kinds. Use on 'create kinds for X', 'distill X into kinds', 'point yourself at API X', or as a stage (A/B/V/D/C) of a running data-to-kinds run. NOT for consuming existing kinds (use aidream's workflow-io-kinds / matrx-frontend's shape-system).

中文简介正在准备中。当前显示原文说明。

Recipe for an agent building a small custom web app (an Applet) for one person or business on THEIR OWN store tables, live at aimatrx.com/applets/<slug>. Use when asked to build an app, portal, calendar, tracker, dashboard screen or client view on someone's tables/data in AI Matrx. NOT for platform features (use build-sub-feature) or tables the platform keeps for itself (defineTypedTable).

中文简介正在准备中。当前显示原文说明。

The bar every automated test or guard must clear: green only when the real system works. Use when writing, fixing, or reviewing a test, mock, stub, fixture, snapshot, or self-test; adding a regression test for a bug; or about to call work done because tests pass.

中文简介正在准备中。当前显示原文说明。

Find and identify the CBasePlayerPawn::CommitSuicide virtual function's vtable slot in the CS2 server binary using IDA Pro MCP. Use this skill when reverse engineering server.dll or libserver.so to locate the self-inflicted-death handler (used by the "kill"/bot_kill console commands and related gameplay code) by RTTI-walking CBasePlayerPawn's primary vtable and reading its known slot index directly, then confirming the body matches CommitSuicide(bool bExplode, bool bForce) semantics. Trigger: CBasePlayerPawn_CommitSuicide

中文简介正在准备中。当前显示原文说明。

Find and identify the CBaseEntity::TakeDamageOld function in the CS2 server binary using IDA Pro MCP. Use this skill when reverse engineering CS2 server.dll or libserver.so to locate the legacy damage-application entry point by finding the code that references its self-identifying assert/warning message "CBaseEntity::TakeDamageOld: damagetype %d with info.GetDamageForce() == Vector::vZero". Trigger: CBaseEntity_TakeDamageOld

中文简介正在准备中。当前显示原文说明。

Find and identify the CAttributeList_SetOrAddAttributeValueByName function in CS2 server binary using IDA Pro MCP. Use this skill when reverse engineering CS2 server.dll or libserver.so to locate the econ attribute-list workhorse that sets an existing float attribute by name or appends a new one. Resolved via a durable caller anchor: the weapon paint-kit/wear/seed/kill-eater applier function, which is itself located by the literal econ keyvalue-system strings "set item texture prefab" / "set item texture wear" / "set item texture seed". This function backs weaponpaints-style plugins that programmatically set an econ item's paint kit / wear / seed / stickers. Trigger: CAttributeList_SetOrAddAttributeValueByName, SetOrAddAttributeValueByName

中文简介正在准备中。当前显示原文说明。

ユーザーからの修正・指摘を受けた際の学習記録と自己改善ループ。ユーザーから「違う」「そうじゃない」「前にも言ったけど」といった修正を受けた時、同じミスを繰り返していると気づいた時、セッション開始時の教訓確認で使う。修正やフィードバックを受けたら必ずこのスキルを参照する。

中文简介正在准备中。当前显示原文说明。

Find the cheapest AI model that is smart enough for a task, and the cheapest provider to run it on, from live Artificial Analysis benchmarks via Firecrawl. Use when the user asks which model to use, how to cut their LLM or API bill, whether a cheaper model would do, which provider is cheapest or fastest for a model, which open-weights model to self-host, or wants something built with the cheapest model for each step.

中文简介正在准备中。当前显示原文说明。

Runs Hyper-V VMs and Windows Sandbox. Trigger to create, start, stop, check or delete a VM (Gen2, secure boot, virtual TPM, settings proved); to take, list or restore a checkpoint; to test a plugin install on a clean machine from a golden checkpoint; to open a suspicious file or zip in a disposable sandbox with networking off; when Hyper-V says access denied; or say hypervrunner (audit, setup, vm, checkpoint, teardown, cleanroom, sandbox, status). Runs as a Hyper-V Administrators member, never self-elevated. Containers and WSL are dockerrunner's; whether code is trustworthy trustwarden's; secrets in returned files shieldwarden's.

中文简介正在准备中。当前显示原文说明。

Hands work to a local model served by LM Studio and verifies what comes back. Trigger to offload, delegate or batch work to a local, offline or on-device model; to queue unattended runs that keep only what passes; to ask which installed model fits a job or whether to switch; to size a task before delegating; when a local model returns nothing, repeats itself or drifts on a long list; when LM Studio shares the GPU with an image tool, or to free its hold; or say lmstudiorunner (audit, size, queue, run, status). Reads models from the live LM Studio API. Cloud model picks and prompt text are promptwright's; a schedule's guardrails agentwright's.

中文简介正在准备中。当前显示原文说明。

Set up, resume or troubleshoot cmail custom-domain email (domain at any registrar → Cloudflare Email Routing → any inbox; optional GoDaddy automation and Gmail send-as): runs cmail setup itself, asking only for missing input, secrets or risky approvals. Don't use for mail migrations or campaigns.

中文简介正在准备中。当前显示原文说明。

Elite mobile app image-generation skill for creating premium, app-native screen concepts and flows. Designed for iOS, Android, and cross-platform mobile products. Prioritizes clean hierarchy, comfortably readable text, strong multi-screen consistency, controlled color palettes, non-generic creative direction, textured surfaces, image-led composition, tasteful custom iconography, and clean phone mockup framing. By default, screens should be shown inside a subtle premium iPhone or similar phone mockup with a visible frame, while the main focus stays on the app content itself. This skill generates images only. It does not write code.

中文简介正在准备中。当前显示原文说明。

Elite website image-to-code skill for Codex. For visually important web tasks, it must first generate the design image(s) itself, deeply analyze them, then implement the website to match them as closely as possible. In Codex, it must prefer large, readable, section-specific images instead of tiny compressed boards, generate fresh standalone images for sections or detail views instead of cropping old ones, avoid lazy under-generation, avoid cards-inside-cards-inside-cards UI, and keep the hero clean, spacious, readable, and visible on a small laptop.

中文简介正在准备中。当前显示原文说明。