nyx-skills

Aevatar scheduling & workflow automation: scheduled_agent_creator (cron/one-shot), the long-running task automation playbook, workflow creation semantics (Scope Workflow vs Ornn publish), agent_builder lifecycle, and scheduled-run credential lifetimes (5-minute broker token at fire) with a token_expired triage protocol.

Aevatar channel & delivery how-to: capability tools (code_execute, nyxid_proxy, GitHub PAT fallback, channel bots), token_expired/401 credential triage for capability tools, LLM route selection slash commands, channel_registrations (staged Lark provisioning), and agent_delivery_targets binding.

Configure and prove Aevatar codex_exec for one NyxID account, choosing either operator-managed OpenSandbox or a private NyxID node-backed SSH service. Use for managed allowlist/binding readiness, node registration, forced-command hardening, service binding, Codex authentication/workspace configuration, mandatory public-sample verification, or diagnosing managed sandbox and private SSH failures.

Mount and run harmless Aevatar workflows that prove codex_exec works through either the operator-managed OpenSandbox target or a private NyxID node-backed SSH target. Use after configuring managed access or a personal node, before real tasks, and when diagnosing identity, allowlist, binding, sandbox, service, principal, or Codex runner failures.

Decide — honestly — whether a thing the user wants to build on Aevatar is possible, what its prerequisites are, or why it cannot be done, BEFORE anyone starts building. Use this first whenever a user describes a goal rather than a concrete artifact — "can aevatar do X", "I want a bot that…", "build me something that posts to Twitter / reads my GitHub / replies on Telegram", "is it possible to…", "automate … every day", "let Lark Base trigger a workflow". It teaches the one hard premise (every third-party capability is brokered by NyxID), the two distinct surfaces (outbound connector vs inbound channel), external HTTP trigger options such as Lark Base automation, how to check what is actually connectable, the prerequisite for each capability class, what is host-gated (and so not self-serve), and what is genuinely impossible without new NyxID/Aevatar platform work — so you can negotiate scope and give the user a straight answer plus next steps instead of over-promising. It scopes; it does not build (hand off to

Entry point, panorama, and router for the entire Aevatar skill family — load this FIRST whenever someone wants to build, run, publish, schedule, externally trigger, or operate anything on Aevatar ("create an agent team", "make a workflow / member", "publish or bind a service", "register it with NyxID", "set up a recurring / cron run", "invoke my service", "let Lark Base trigger my workflow"), wants to know whether something is even possible ("can Aevatar do X?", "能不能用 aevatar 实现"), or just wants to know what Aevatar can do. It teaches the object model (scope → team → member[workflow|script|gagent] → service → schedule/external trigger), how to authenticate as a NyxID-bearer REST client, how to resolve your scope, and the two caller modes (client REST vs in-session server-side tools). It does not do the work itself — it routes you to the right companion skill (feasibility-advisor, workflow-authoring, team-builder, service-publisher, scheduler, plus diagnostics probes and the safety-net fallback), held together

Create and manage cron schedules that fire an Aevatar service on a recurring basis, authenticated as the scope owner via NyxID — over the REST API. Use when a user wants to "schedule", "run on a cron", "set up a recurring run", "run every day/hour/Monday", "automate this service on a timer", "preview a cron", "pause/resume/disable a schedule", or "run it now" — or hits token_expired on a scheduled run's late steps. It builds the schedule against a published service (identity + endpoint + payload + serving revision), uses scope-owner NyxID auth (which requires the owner's NyxID broker binding), documents the fire-time credential's fixed 5-minute lifetime and how to design runs around it, and covers preview, enable/disable, run-now, update, and delete. Publish the service first with the service-publisher skill.

Publish an Aevatar member, team, or workflow as an invocable service and (host permitting) register it with NyxID, then verify, invoke, or wire external HTTP triggers such as Lark Base automation — all over the REST API. Use when a user wants to "publish/bind a service", "expose my workflow/team as a service", "register it with NyxID", "make it callable", "get the service slug/URL", "invoke my service", "let Lark Base call my workflow", "trigger this workflow from an external webhook", or "version/deploy/roll out a service". It covers the simple scope binding, reading back a member's published service, the full account-level service lifecycle (revision → publish → deploy → rollout), how to confirm the NyxID registration (slug + status), how to invoke an endpoint, and how to distinguish direct NyxID proxy triggering from host-gated externalExposure. Build the team/member first with the team-builder skill.

Build an Aevatar agent team and its members over the REST API. Use when a user wants to "create a team", "add a member", "make a workflow member / script member / gagent member", "set the team's entry point", or "assemble agents into a team". It creates the team, creates members whose implementation is a workflow (most common), a script, or a hosted gagent, binds each member's concrete implementation (the workflow YAML is attached here), waits for the async binding to succeed, and sets the team entry member. Author the workflow YAML first with the workflow-authoring skill; publish the result as a service with the service-publisher skill.

Use AFTER something goes wrong while using Aevatar — a user hits an error, failure, or confusing behavior and you must find whether it lives in Aevatar, NyxID, or Ornn, then act. Triggers - "aevatar is erroring", "why did my workflow fail", "my scheduled run did not fire", "my bot does not reply", "connector 401/403", "skill won't pull/upload", "is this an aevatar, nyxid, or ornn bug", "file an issue", "am I using this right". It attributes the failure by tracing the request path, pulls that layer's real public source for a code-grounded root cause citing file and line, then branches - draft and, only on explicit user confirmation, file a precise GitHub issue when behavior violates the layer's published contract, or explain the correct usage from the code when it is a usage mistake. The after-it-breaks counterpart to aevatar-feasibility-advisor; never auto-files, de-dups first, never claims a root cause without a code citation. Works locally (git + gh) and server-side (nyxid_proxy + api-github).

Author, validate, and persist an executable aevatar workflow from a natural-language request — use it when the user wants to create, build, set up, or automate a multi-step task as a runnable aevatar workflow (make a workflow that…, automate…, build a pipeline…, set up a recurring…). It generates workflow YAML, dispatch-validates it, then saves it as a reusable workflow that can be re-run and watched in the observatory. Not for running an existing workflow — search for that and start it instead.

Builds deterministic Lark budget variance payloads and, when asked, runs the full budget monitoring flow through a NyxID service instead of direct Lark credentials.

Method for producing short vertical videos from ChatGPT image generation through a fixed browser tab, local TTS narration, timed subtitle burn-in, cover exports, and a manifest. Use when an operator needs a reproducible image-to-short workflow with pinned session consistency, per-beat visual prompts, 1080x1920 safe-area rules, local ffmpeg/PIL assembly, and private access kept outside the skill.

Chrono AI 非技术团队 onboarding — 安装 NyxID CLI, 配置 AI 工具 MCP, 克隆战略仓库, 运行每日更新

Run a real GitHub-based team health / performance audit for the Chrono AI engineering team. Triggers on "team health", "团队健康", "全员审计", "全员绩效", "绩效审计", "check the team". Runs a fixed self-contained script via code_execute and returns a red/yellow/green report; posts it to Lark.

Universal try-catch fallback for the aevatar model. Use whenever, after a genuine attempt, you cannot complete the user's request with available server-side capabilities — no matching skill/workflow/connector/tool, a terminal failure, or a task that inherently needs the caller's local environment (files, shell, local context). Instead of failing opaquely or fabricating, return the original request verbatim to the calling agent so it can finish with its own local tools. Generic by design — addresses "the calling agent" with no hardcoded client or skill names.

Teach an aevatar agent to run Firecrawl web-research/agent jobs through NyxID (submit, poll, then read the result).

Operational manual for an AI agent driving the fkst CONTROL PLANE (fkst-control-plane) over its HTTP API, exclusively through the NyxID CLI credential proxy (`nyxid proxy request …`). Covers the five core user flows — create a GitHub repo under my account, bootstrap a repo's `.fkst/` directory, trigger a fkst-substrate session with my input arguments, check the status of all my substrate sessions, and stop/terminate a specific session — with the EXACT request body and response shape for every call so the agent never has to guess. Invoke when the user says "create a repo via the control plane", "bootstrap .fkst", "trigger a substrate session", "start/stop a fkst session", "check my sessions", or "drive the fkst control plane through nyxid".

Operate a user's GitHub account through NyxID's credential-brokering proxy (service slug api-github) — read and write repositories, files, issues, pull requests, commits, branches, Actions, gists and anything else the GitHub REST API exposes, all on the user's behalf and without ever handling a raw token. NyxID injects the user's GitHub credential server-side. Use when an agent needs to read from or act on GitHub for a user who has connected their GitHub account in NyxID.

Translate Japanese text into Chinese with support for tone, context, and output style preferences.

Check one Lark approval instance or list approval tasks with the typed lark_approvals_get / lark_approvals_list tools and report the normalized status (approved / rejected / running, is_terminal, pending task and approver). Doubles as the end-to-end test for approval status tracking and as a daily "查审批" helper.

Create and operate Lark Base (多维表格 / Bitable) tables through the bot's NyxID-brokered bitable/v1 + drive/v1 APIs — create a new Base and grant the requester access, list a Base's tables and a table's fields, read and filter records, create or update single rows, and batch-write many rows. Handles the app_token + table_id extraction from a Base share URL, writes by field NAME with the right typed value shapes (date = ms timestamp, person = open_id, single/multi-select = existing option names), grants the requester full_access on any Base it creates, and doubles as the bitable-scope probe — on a scope 403 it reports the exact missing permission. Lark Base here is the Bitable database product, distinct from Lark Sheets — use the Sheets skill for plain grid spreadsheets.

Calendar assistant over the bot's NyxID-brokered Lark calendar/v4 APIs — list calendars, check free/busy across people, find a common slot, read your agenda for a window, and create or update events with attendees. Knows the event-time shape (a timestamp + timezone object, not RFC3339) and that attendees are added in a separate call, and doubles as the calendar-scope probe — on a scope 403 it reports the exact missing permission instead of failing.

Recall and summarize Lark conversations — pull a chat's history over a time window through the real im/v1 history API, reconstruct discussions, and answer "who said what" with quoted evidence. Use for "总结这个群今天聊了什么 / 上周谁提过X / 找一下关于Y的讨论". Requires the bot tenant scope im:message.group_msg; reports the exact missing scope verbatim when absent.

Create and edit Lark cloud documents (Docx) through the bot's NyxID-brokered docx/v1 + drive/v1 APIs — create a document and grant the requester access, append headings and text paragraphs, read a document's plain-text content, and share it with other people or make it tenant-readable. Handles the Docx block model (block_type + text_run elements) and the document-root block_id, grants the requester full_access on any doc it creates, and doubles as the docx-scope probe — on a scope 403 it reports the exact missing permission.

Send, reply to, react to, read, and find Lark messages and chats through the bot's NyxID-brokered im/v1 APIs — proactively post to a chat or DM, thread-reply, add or remove emoji reactions, read a single message or pull a chat's history over a time window, search or list the chats the bot is in, and download an image/file from a message. Handles the "content must be a JSON string" footgun and receive_id_type selection, and doubles as the im-scope probe — on a scope 403 it reports the exact missing permission instead of failing silently.

Builds the connector-ready Lark approval instance payload for the onboarding email approval flow. Use after Aevatar has received a Lark/Base onboarding event and before calling the NyxID Lark approval connector.

Read and write Lark Sheets (电子表格) through the bot's NyxID-brokered sheets APIs — find a spreadsheet and its sheet ids, read a cell range, append rows to the bottom, and overwrite a range. Handles the range format (sheetId!A1:D10) and the valueRange wrapper, extracts the spreadsheet_token from a share URL, and doubles as the sheets-scope probe — on a scope 403 it reports the exact missing permission. Lark Sheets here is the spreadsheet product, distinct from Lark Base / Bitable — use the Base skill for Bitable tables.

Index, router, and dependency map for the aevatar Lark/Feishu skill family on ornn — load this FIRST when someone wants the bot to do anything in Lark (send a message, recall a chat, write a doc / sheet / Base record, manage a calendar event, run an approval). It names the six core operation primitives (im / docx / sheets / bitable / calendar / approval), states the one shared premise (every call goes through nyxid_proxy slug api-lark-bot, scope-probe-first), routes intent to exactly one primitive, lists which business workflows consume them, and flags the current cleanup items. It does not call Lark itself — it routes to the right ops skill.

Builds Lark team-health command parsing, progress-message payloads, and report-message payloads from collected GitHub and AI analysis results.

李诞做自媒体/内容工作的思维框架与表达方式。基于《李诞工作手册（全新修订版）》 第一部分（自媒体内容工作，8 个「心」章节 + 问答）的精读提炼，含 6 个核心心智模型、 10 条决策启发式和完整表达 DNA。 用途：作为思维顾问，用李诞的视角审内容、做内容决策、面对数据焦虑与网络争议时给反馈。 当用户提到「用李诞的视角」「李诞会怎么看」「李诞模式」「李诞自媒体心法」时使用。 即使用户只是说「帮我用李诞的角度想想这条内容」「如果是李诞会怎么做这个号」「切换到李诞」也应触发。

Builds Lark monthly attendance approval/reminder payloads and, when asked, runs the full attendance flow through a NyxID service instead of direct Lark credentials.

Lark/Feishu integration via NyxID proxy. Replaces lark-cli — call Lark OpenAPI directly through nyxid proxy request with automatic credential injection.

Invoke an operation on a connected and authorized NyxID service through Aevatar's dynamically typed connected-service tools or the credential-injecting nyxid_proxy fallback. Use when a user asks to read, search, create, update, send, run, download, or otherwise call a NyxID service while keeping upstream credentials inside NyxID and respecting provider scopes, approvals, and service routing. 中文触发包括“调用我的 NyxID 服务”“使用服务”“让服务执行”。

Connect, add, reconnect, or repair a NyxID service through NyxID's secure credential and provider OAuth surfaces, then grant the current Aevatar developer app incremental access to that connected service. Use when a user wants a new NyxID service, sees a missing-connection or missing-credential error, needs provider OAuth, or must authorize an already connected service for NyxID Chat without exposing secrets to the model. 中文触发包括“连接新的 NyxID 服务”“新增服务”“授权服务”。

Inventory and inspect NyxID services available to the current authenticated user across connected instances, the service catalog, proxy availability, and LLM routes. Use when a user asks what services they have, what can be added, whether a service is connected, active, healthy, or callable, which personal or organization source owns it, or needs exact service details before connecting, invoking, or maintaining it. 中文触发包括“我有哪些 NyxID 服务”“查看服务列表”“服务状态”。

Maintain an existing NyxID service by inspecting and updating labels, endpoints, active state, direct or node routing, health, external-key metadata, and deletion, while sending all credential rotation or repair through NyxID's secure UI. Use when a user asks to edit, reroute, enable, disable, repair, rotate credentials for, diagnose, or remove a connected NyxID service. 中文触发包括“修改 NyxID 服务路由”“更新服务”“轮换凭据”“删除服务”。

Brokers credentials for downstream services (OpenAI, Anthropic, GitHub, Lark, custom APIs, SSH, MCP) so the agent never sees raw API keys or OAuth tokens. Use whenever the user asks to call, proxy, or authenticate against a third-party API/service, mentions NyxID, asks to "connect", "add a service", "set up an API key", manage credentials/nodes/MCP, send messages through bot platforms, or wire up SSH access. Operate exclusively through the `nyxid` CLI.

The manual an AI agent loads to operate Ornn — the model-agnostic skill-lifecycle API (an npm-style registry + CLI for agent skills) — via the NyxID CLI (`nyxid proxy request ornn-api …`). Load and follow this skill WHENEVER the user asks to do anything with Ornn skills or skillsets. Skills: search Ornn or find a skill, pull or install a skill (or a specific version), run a skill, build and upload a skill, publish a new version, make a skill public / private / shared, run or read a security audit, deprecate or delete a version, diff two versions, check usage analytics, bind a skill to a NyxID service, link a skill to GitHub or sync from source, manage npm-style dist-tags, or transfer skill ownership. Skillsets — curated multi-skill bundles with a required master prompt: bundle skills into a set, create or publish a skillset, resolve its closure in one call, export a skillset as a Claude Code marketplace plugin, transfer skillset ownership, or diagnose why a shared skillset isn't visible (visibility derives fr

Verify-first promotion intelligence for ANY repo or org. Pulls live GitHub state, runs sshx multi-perspective consensus over it, and returns an evidence-backed board of what is genuinely promotable, how items connect, and what is unverifiable. Use when an agent or operator asks "what's worth promoting / what connects / what's real" for a target repo or org.