[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M71FRH7T0N6NANFPM85NPVAQ":3,"skill-same:01M71FRH7T0N6NANFPM85NPVAQ":71},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":23,"publishedAt":44,"linkOnly":45,"markdownContent":46,"markdownBody":47,"frontMatter":48,"refName":49,"commitSha":50,"skillDirectory":51,"forksCount":14,"repositoryHtmlUrl":52,"repositoryHomepageUrl":53,"ownerHtmlUrl":54,"ownerBlogUrl":55,"ownerDescription":6,"updatedAt":56,"install":57,"resources":70},"01M71FRH7T0N6NANFPM85NPVAQ","agent-conf-sync",null,"Runs the cyncia batch script (`sync-all`) to regenerate Cursor, Claude Code, GitHub Copilot, VS Code, JetBrains Junie, and Codex configuration from a generic source tree. Use when the user asks to \"sync agent config\", \"regenerate rules\u002Fskills\u002Fagents\u002Fmcp\", \"update .cursor \u002F .claude \u002F .github \u002F .junie \u002F .vscode \u002F .codex \u002F .agents from source\", \"run sync-all\", or similar — so they don't have to assemble CLI flags or pick Windows vs Unix scripts themselves. The skill detects the OS, finds the script, and infers `-i` \u002F `-o` \u002F `--tools` \u002F `--items` \u002F `--clean` from the natural-language request.","crestreach\u002Fcyncia","crestreach","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F169358583?v=4","01M71FR759G1CF6T5BAPQ1XKJ4","CrestReach","01M71FR7663FW7N09D3DSZRWJG",1,"MIT","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Fskills\u002Fagent-conf-sync\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"sync","conf","agent conf sync","unix","scripts","update","request","run","assemble","cli","jetbrains","batch","natural-language","code","when","MCP","clean","copilot","pick","github copilot","2026-11-10 10:12:30",false,"---\nname: agent-conf-sync\ndescription: Runs the cyncia batch script (`sync-all`) to regenerate Cursor, Claude Code, GitHub Copilot, VS Code, JetBrains Junie, and Codex configuration from a generic source tree. Use when the user asks to \"sync agent config\", \"regenerate rules\u002Fskills\u002Fagents\u002Fmcp\", \"update .cursor \u002F .claude \u002F .github \u002F .junie \u002F .vscode \u002F .codex \u002F .agents from source\", \"run sync-all\", or similar — so they don't have to assemble CLI flags or pick Windows vs Unix scripts themselves. The skill detects the OS, finds the script, and infers `-i` \u002F `-o` \u002F `--tools` \u002F `--items` \u002F `--clean` from the natural-language request.\n---\n\n# agent-conf-sync\n\nThis skill invokes **`scripts\u002Fsync-all.sh`** (macOS\u002FLinux\u002FGit Bash) or **`scripts\u002Fsync-all.ps1`** (Windows PowerShell) from the `cyncia` repo.\nThe user describes what they want in plain language; you pick the right script, resolve paths, infer flags, run it, and summarize the result.\n\n## Source tree format (`\u003Csource_root>`)\n\nA source root must contain `AGENTS.md`.\nIt may optionally contain any of these folders; each is independent and missing folders are skipped with a console note:\n\n| Folder | One per | Purpose |\n|---|---|---|\n| `agents\u002F\u003Cname>.md` | file | One agent \u002F subagent definition. |\n| `skills\u002F\u003Cname>\u002FSKILL.md` | folder | One Agent-Skills-format skill (plus optional extra files in the same folder). |\n| `rules\u002F\u003Cname>.md` | file | One rule with YAML frontmatter (`description`, `applies-to`, `always-apply`). |\n| `mcp-servers\u002F\u003Cname>.json` | file | One MCP server definition (body of the per-server config object, no `mcpServers` wrapper). |\n\n## What `sync-all` writes\n\nFor each selected tool, `sync-all` writes:\n\n| Tool | Generated paths under `\u003Coutput_root>` |\n|---|---|\n| `cursor` | `AGENTS.md` (copy), `.cursor\u002F{agents,skills,rules}\u002F`, `.cursor\u002Fmcp.json` |\n| `claude` | `CLAUDE.md` (AGENTS + rules merged), `.claude\u002F{agents,skills}\u002F`, `.mcp.json` |\n| `copilot` | `.github\u002Fcopilot-instructions.md`, `.github\u002F{agents,skills,instructions}\u002F` |\n| `vscode` | `.vscode\u002Fmcp.json` (also read by Copilot Chat in VS Code) |\n| `junie` | `.junie\u002FAGENTS.md` (AGENTS + rules merged), `.junie\u002F{agents,skills}\u002F` |\n| `codex` | `AGENTS.md` (copy), `AGENTS.override.md` when `codex-rules-mode=agents-override`, `.codex\u002Fagents\u002F*.toml`, `.agents\u002Fskills\u002F`, `.codex\u002Fconfig.toml` `mcp_servers` tables when `codex-sync-mcp=true` |\n\n`sync-agent-guidelines` always emits the **full** `AGENTS.md` \u002F `CLAUDE.md` \u002F `.junie\u002FAGENTS.md` — `--items` does not trim it.\nClaude and Junie have no per-rule files by default (rules are appended into the guidelines file).\nCodex does not generate `.codex\u002Frules` from Cyncia Markdown rules because Codex `.rules` files are Starlark command policy; by default, Codex Markdown rule bodies are merged into root `AGENTS.override.md` instead.\n\nThis skill is only about **invocation and reporting**.\nFor source-format details (rule frontmatter fields, secret-token translation, agent ↔ MCP linkage) see the upstream `README.md`.\n\n## When to apply\n\nApply when the user wants to (re)generate tool-specific AI-assistant config from a single generic source tree.\nTrigger phrases include:\n\n- \"sync all\", \"run sync-all\", \"sync agent config\", \"sync the config\"\n- \"regenerate \u002F update the rules \u002F skills \u002F agents \u002F guidelines \u002F mcp \u002F `AGENTS.override.md`\"\n- \"rebuild `.cursor` \u002F `.claude` \u002F `.github` \u002F `.junie` \u002F `.codex` \u002F `.agents`\"\n- \"sync MCP servers\", \"regenerate `.vscode\u002Fmcp.json`\", \"update Claude `.mcp.json`\"\n- \"sync for Cursor and Claude\", \"only Copilot\", \"only Codex\", \"skip Junie\"\n- \"clean sync\", \"prune stale files\", \"mirror deletions\"\n\nDo **not** apply when the user is asking about the *format* of rules\u002Fskills\u002Fagents, or editing source files.\nFor that, point them to `README.md`.\n\n## Step 1: Detect the OS and shell\n\nPick the entrypoint based on the shell you'll actually execute in:\n\n| Environment | Entrypoint | Flag style |\n|-------------|-----------|-----------|\n| Windows native PowerShell (`pwsh`, `powershell.exe`) | `scripts\\sync-all.ps1` | `-InputRoot`, `-OutputRoot`, `-Tools`, `-Items`, `-Clean` |\n| macOS, Linux, BSD | `scripts\u002Fsync-all.sh` | `-i`, `-o`, `--tools`, `--items`, `--clean` |\n| Git Bash \u002F WSL on Windows | `scripts\u002Fsync-all.sh` (prefer Bash) | same as Unix |\n\nDetect with: `$env:OS -eq \"Windows_NT\"` in PowerShell; `uname` in Bash.\nDo **not** assume Unix just because the repo is cross-platform — use what the current session runs.\n\n## Step 2: Locate the script root (`CONFIG_SYNC_ROOT`)\n\nResolve the directory that contains `scripts\u002Fsync-all.sh` and `scripts\u002Fsync-all.ps1`:\n\n1. If the workspace root has `scripts\u002Fsync-all.sh`, use the workspace root.\n2. Else if the user named a path (e.g. `vendor\u002Fcyncia`), use that.\n3. Else search the workspace for `**\u002Fscripts\u002Fsync-all.sh` and use its grandparent.\n4. If still ambiguous or missing, **ask once** for the path.\n\nUse **absolute paths** when invoking the script.\n\n## Step 3: Infer the required parameters\n\n### Input root — `-i` \u002F `-InputRoot` (required)\n\nThe directory that contains `AGENTS.md` and optionally any of `agents\u002F`, `skills\u002F`, `rules\u002F`, `mcp-servers\u002F` (see [Source tree format](#source-tree-format-source_root) above).\nEach folder is independent — `sync-all` skips the corresponding subscript (with a console note) when its source dir is absent.\nOnly `AGENTS.md` is mandatory; if it is missing, `sync-all` exits with an error.\n\nMapping from user phrasing:\n\n- \"from `.agent-config`\", \"internal tree\", \"authoring root\", \"this repo's own source\" → `\u003CCONFIG_SYNC_ROOT>\u002F.agent-config`\n- \"from examples\", \"the examples tree\" → `\u003CCONFIG_SYNC_ROOT>\u002Fexamples`\n- \"from my source\", \"from `\u003Cpath>`\" → that path\n- \"in place\", \"same as output\" → same path as output root\n\nIf nothing is stated, look at the workspace: presence of `AGENTS.md` + `agents\u002F` + `skills\u002F` + `rules\u002F` at some root indicates it's the source.\nAsk if unclear.\n\n### Output root — `-o` \u002F `-OutputRoot` (required)\n\nThe project root where `.cursor\u002F`, `.claude\u002F`, `.github\u002F`, `.junie\u002F`, `.codex\u002F`, `.agents\u002F`, the root `AGENTS.md` copy, and root `AGENTS.override.md` are written.\nUsually the consumer project root, or this repo's root when regenerating its own outputs.\n\n### `--tools` \u002F `-Tools` (optional; default from config)\n\nComma-separated subset of `cursor,claude,copilot,vscode,junie,codex` (case-insensitive, no spaces).\nIf omitted, `sync-all` uses `default-tools` from `.cyncia\u002Fcyncia.conf`; if that property is missing, the built-in default is all six supported tools.\n\n| Phrase | Value |\n|--------|-------|\n| \"all tools\", \"everything\", no tool mentioned | omit flag (uses `default-tools`; built-in default all six) |\n| \"only Cursor\" | `cursor` |\n| \"Cursor and Claude\", \"not Copilot, not Junie\" | `cursor,claude` |\n| \"skip Junie\" | `cursor,claude,copilot,vscode,codex` |\n| \"only VS Code\", \"just MCP for VS Code\" | `vscode` |\n| \"Copilot with its MCP\" | `copilot,vscode` (Copilot Chat in VS Code reads `.vscode\u002Fmcp.json`) |\n| \"only Codex\" | `codex` |\n\nSynonyms: **cursor** ← \"Cursor\"; **claude** ← \"Claude\", \"Claude Code\"; **copilot** ← \"Copilot\", \"GitHub Copilot\"; **vscode** ← \"VS Code\", \"VSCode\", \"Visual Studio Code\"; **junie** ← \"Junie\", \"JetBrains\"; **codex** ← \"Codex\", \"OpenAI Codex\".\n\n### `--items` \u002F `-Items` (optional)\n\nSingle comma-separated list of **basenames without extension** (e.g. `delegate-to-aside`, `commit-style`).\nApplied to agents, skills, and rules together.\n\n| Phrase | Value |\n|--------|-------|\n| \"only the aside skill\", \"just `delegate-to-aside`\" | `delegate-to-aside` |\n| \"just `foo` and `bar`\" | `foo,bar` |\n| \"full sync\", \"all of them\", no item mentioned | omit flag |\n\n**Caveat:** `--items` does not trim the guidelines merge.\n`sync-agent-guidelines` always emits the full `AGENTS.md` + merged rules for Claude\u002FJunie\u002FCodex.\nMention this in the report if the user expected partial guidelines.\n\n### `--clean` \u002F `-Clean` (optional; default: off)\n\nSet when the user asks to **remove stale outputs**, **prune**, **mirror source deletions**, **empty generated dirs first**, or says \"clean sync\".\n\n**Warning — data loss risk:** `--clean` empties the entire output directory for **every** sync step on **every** selected tool before writing: `.cursor\u002F{agents,rules,skills}`, `.claude\u002F{agents,skills}`, `.github\u002F{agents,instructions,skills}`, `.junie\u002F{agents,skills}`, `.codex\u002Fagents`, and `.agents\u002Fskills`.\nAny hand-authored files in those dirs are deleted.\nThis is especially risky for `.github\u002F` since users often keep unrelated content there (workflows live in `.github\u002Fworkflows\u002F` — not wiped — but a user might have put manual instruction files in `.github\u002Finstructions\u002F`).\nCodex MCP clean is narrower: it removes\u002Freplaces only `mcp_servers` sections in `.codex\u002Fconfig.toml`, preserving unrelated Codex config.\nFlag clean risk in your pre-run note whenever `-Clean` is inferred, and name the specific dirs or Codex config sections affected by the `--tools` selection.\n\n## Step 4: Validate and run\n\nBefore running:\n\n1. Check that `\u003Cinput_root>\u002FAGENTS.md` exists. If not, stop and tell the user.\n2. Use absolute paths in the command.\n3. Capture full stdout and stderr.\n\nExample — macOS \u002F Linux:\n\n```bash\nbash \"\u002Fabs\u002Fpath\u002Fto\u002Fcyncia\u002Fscripts\u002Fsync-all.sh\" \\\n  -i \"\u002Fabs\u002Fpath\u002Fto\u002Fsource-root\" \\\n  -o \"\u002Fabs\u002Fpath\u002Fto\u002Foutput-root\" \\\n  --tools cursor,claude \\\n  --items delegate-to-aside\n```\n\nWith clean:\n\n```bash\nbash \"\u002Fabs\u002Fpath\u002Fto\u002Fcyncia\u002Fscripts\u002Fsync-all.sh\" \\\n  -i \"\u002Fabs\u002Fpath\u002Fto\u002Fsource-root\" \\\n  -o \"\u002Fabs\u002Fpath\u002Fto\u002Foutput-root\" \\\n  --clean\n```\n\nExample — Windows PowerShell:\n\n```powershell\n& 'C:\\abs\\path\\to\\cyncia\\scripts\\sync-all.ps1' `\n  -InputRoot 'C:\\abs\\path\\to\\source-root' `\n  -OutputRoot 'C:\\abs\\path\\to\\output-root' `\n  -Tools 'cursor,claude' `\n  -Items 'delegate-to-aside'\n```\n\nOn non-zero exit, surface the error and suggest fixes (missing dirs, unknown tool name, missing `AGENTS.md`).\n\n## Step 5: Post-run report (required)\n\nReply with a **compact** summary, not the full log:\n\n1. **Command** — one line: OS\u002Fshell, script path, effective `-i`\u002F`-o`\u002F`--tools`\u002F`--items`\u002F`--clean`.\n   If `--tools` was omitted and `default-tools` is visible, mention the configured default that was used.\n2. **Deletions \u002F clean** — list paths from log lines matching `cleaned` \u002F `removed` \u002F `Clean`.\n   If `--clean` was not used, state that no clean step ran (overwrite in place only).\n3. **Generated \u002F updated** — from lines with `->` (e.g. `cursor agent ->`, `copilot skill ->`) and `== tool ==` headers.\n   Group by tool.\n   Call out `AGENTS.md`, `AGENTS.override.md`, `CLAUDE.md`, `.github\u002Fcopilot-instructions.md`, `.junie\u002FAGENTS.md`, `.codex\u002Fconfig.toml`, and `.codex\u002Fagents\u002F*.toml` when they appear.\n4. **Basis** — one short sentence explaining what you inferred from the user's request.\n\n## Edge cases\n\n- **MCP-only sync:** likewise, call `scripts\u002F\u003Ctool>\u002Fsync-mcp.{sh,ps1} -i \u003Csrc>\u002Fmcp-servers -o \u003Cout>` directly.\n  The Bash variants require `jq`.\n- **Rules-only sync:** `sync-all` has no such flag.\n  Call the per-tool script directly (e.g. `scripts\u002Fcursor\u002Fsync-rules.sh -i \u003Csrc>\u002Frules -o \u003Cout>`).\n- **Installed cyncia files \u002F monorepo:** always resolve real filesystem paths before running.\n- **Input equals output:** allowed; scripts skip the redundant `AGENTS.md` copy.\n- **Git Bash on Windows:** prefer the `.sh` script with `bash`; only use `.ps1` when the session is clearly native PowerShell.\n","# agent-conf-sync\n\nThis skill invokes **`scripts\u002Fsync-all.sh`** (macOS\u002FLinux\u002FGit Bash) or **`scripts\u002Fsync-all.ps1`** (Windows PowerShell) from the `cyncia` repo.\nThe user describes what they want in plain language; you pick the right script, resolve paths, infer flags, run it, and summarize the result.\n\n## Source tree format (`\u003Csource_root>`)\n\nA source root must contain `AGENTS.md`.\nIt may optionally contain any of these folders; each is independent and missing folders are skipped with a console note:\n\n| Folder | One per | Purpose |\n|---|---|---|\n| `agents\u002F\u003Cname>.md` | file | One agent \u002F subagent definition. |\n| `skills\u002F\u003Cname>\u002FSKILL.md` | folder | One Agent-Skills-format skill (plus optional extra files in the same folder). |\n| `rules\u002F\u003Cname>.md` | file | One rule with YAML frontmatter (`description`, `applies-to`, `always-apply`). |\n| `mcp-servers\u002F\u003Cname>.json` | file | One MCP server definition (body of the per-server config object, no `mcpServers` wrapper). |\n\n## What `sync-all` writes\n\nFor each selected tool, `sync-all` writes:\n\n| Tool | Generated paths under `\u003Coutput_root>` |\n|---|---|\n| `cursor` | `AGENTS.md` (copy), `.cursor\u002F{agents,skills,rules}\u002F`, `.cursor\u002Fmcp.json` |\n| `claude` | `CLAUDE.md` (AGENTS + rules merged), `.claude\u002F{agents,skills}\u002F`, `.mcp.json` |\n| `copilot` | `.github\u002Fcopilot-instructions.md`, `.github\u002F{agents,skills,instructions}\u002F` |\n| `vscode` | `.vscode\u002Fmcp.json` (also read by Copilot Chat in VS Code) |\n| `junie` | `.junie\u002FAGENTS.md` (AGENTS + rules merged), `.junie\u002F{agents,skills}\u002F` |\n| `codex` | `AGENTS.md` (copy), `AGENTS.override.md` when `codex-rules-mode=agents-override`, `.codex\u002Fagents\u002F*.toml`, `.agents\u002Fskills\u002F`, `.codex\u002Fconfig.toml` `mcp_servers` tables when `codex-sync-mcp=true` |\n\n`sync-agent-guidelines` always emits the **full** `AGENTS.md` \u002F `CLAUDE.md` \u002F `.junie\u002FAGENTS.md` — `--items` does not trim it.\nClaude and Junie have no per-rule files by default (rules are appended into the guidelines file).\nCodex does not generate `.codex\u002Frules` from Cyncia Markdown rules because Codex `.rules` files are Starlark command policy; by default, Codex Markdown rule bodies are merged into root `AGENTS.override.md` instead.\n\nThis skill is only about **invocation and reporting**.\nFor source-format details (rule frontmatter fields, secret-token translation, agent ↔ MCP linkage) see the upstream `README.md`.\n\n## When to apply\n\nApply when the user wants to (re)generate tool-specific AI-assistant config from a single generic source tree.\nTrigger phrases include:\n\n- \"sync all\", \"run sync-all\", \"sync agent config\", \"sync the config\"\n- \"regenerate \u002F update the rules \u002F skills \u002F agents \u002F guidelines \u002F mcp \u002F `AGENTS.override.md`\"\n- \"rebuild `.cursor` \u002F `.claude` \u002F `.github` \u002F `.junie` \u002F `.codex` \u002F `.agents`\"\n- \"sync MCP servers\", \"regenerate `.vscode\u002Fmcp.json`\", \"update Claude `.mcp.json`\"\n- \"sync for Cursor and Claude\", \"only Copilot\", \"only Codex\", \"skip Junie\"\n- \"clean sync\", \"prune stale files\", \"mirror deletions\"\n\nDo **not** apply when the user is asking about the *format* of rules\u002Fskills\u002Fagents, or editing source files.\nFor that, point them to `README.md`.\n\n## Step 1: Detect the OS and shell\n\nPick the entrypoint based on the shell you'll actually execute in:\n\n| Environment | Entrypoint | Flag style |\n|-------------|-----------|-----------|\n| Windows native PowerShell (`pwsh`, `powershell.exe`) | `scripts\\sync-all.ps1` | `-InputRoot`, `-OutputRoot`, `-Tools`, `-Items`, `-Clean` |\n| macOS, Linux, BSD | `scripts\u002Fsync-all.sh` | `-i`, `-o`, `--tools`, `--items`, `--clean` |\n| Git Bash \u002F WSL on Windows | `scripts\u002Fsync-all.sh` (prefer Bash) | same as Unix |\n\nDetect with: `$env:OS -eq \"Windows_NT\"` in PowerShell; `uname` in Bash.\nDo **not** assume Unix just because the repo is cross-platform — use what the current session runs.\n\n## Step 2: Locate the script root (`CONFIG_SYNC_ROOT`)\n\nResolve the directory that contains `scripts\u002Fsync-all.sh` and `scripts\u002Fsync-all.ps1`:\n\n1. If the workspace root has `scripts\u002Fsync-all.sh`, use the workspace root.\n2. Else if the user named a path (e.g. `vendor\u002Fcyncia`), use that.\n3. Else search the workspace for `**\u002Fscripts\u002Fsync-all.sh` and use its grandparent.\n4. If still ambiguous or missing, **ask once** for the path.\n\nUse **absolute paths** when invoking the script.\n\n## Step 3: Infer the required parameters\n\n### Input root — `-i` \u002F `-InputRoot` (required)\n\nThe directory that contains `AGENTS.md` and optionally any of `agents\u002F`, `skills\u002F`, `rules\u002F`, `mcp-servers\u002F` (see [Source tree format](#source-tree-format-source_root) above).\nEach folder is independent — `sync-all` skips the corresponding subscript (with a console note) when its source dir is absent.\nOnly `AGENTS.md` is mandatory; if it is missing, `sync-all` exits with an error.\n\nMapping from user phrasing:\n\n- \"from `.agent-config`\", \"internal tree\", \"authoring root\", \"this repo's own source\" → `\u003CCONFIG_SYNC_ROOT>\u002F.agent-config`\n- \"from examples\", \"the examples tree\" → `\u003CCONFIG_SYNC_ROOT>\u002Fexamples`\n- \"from my source\", \"from `\u003Cpath>`\" → that path\n- \"in place\", \"same as output\" → same path as output root\n\nIf nothing is stated, look at the workspace: presence of `AGENTS.md` + `agents\u002F` + `skills\u002F` + `rules\u002F` at some root indicates it's the source.\nAsk if unclear.\n\n### Output root — `-o` \u002F `-OutputRoot` (required)\n\nThe project root where `.cursor\u002F`, `.claude\u002F`, `.github\u002F`, `.junie\u002F`, `.codex\u002F`, `.agents\u002F`, the root `AGENTS.md` copy, and root `AGENTS.override.md` are written.\nUsually the consumer project root, or this repo's root when regenerating its own outputs.\n\n### `--tools` \u002F `-Tools` (optional; default from config)\n\nComma-separated subset of `cursor,claude,copilot,vscode,junie,codex` (case-insensitive, no spaces).\nIf omitted, `sync-all` uses `default-tools` from `.cyncia\u002Fcyncia.conf`; if that property is missing, the built-in default is all six supported tools.\n\n| Phrase | Value |\n|--------|-------|\n| \"all tools\", \"everything\", no tool mentioned | omit flag (uses `default-tools`; built-in default all six) |\n| \"only Cursor\" | `cursor` |\n| \"Cursor and Claude\", \"not Copilot, not Junie\" | `cursor,claude` |\n| \"skip Junie\" | `cursor,claude,copilot,vscode,codex` |\n| \"only VS Code\", \"just MCP for VS Code\" | `vscode` |\n| \"Copilot with its MCP\" | `copilot,vscode` (Copilot Chat in VS Code reads `.vscode\u002Fmcp.json`) |\n| \"only Codex\" | `codex` |\n\nSynonyms: **cursor** ← \"Cursor\"; **claude** ← \"Claude\", \"Claude Code\"; **copilot** ← \"Copilot\", \"GitHub Copilot\"; **vscode** ← \"VS Code\", \"VSCode\", \"Visual Studio Code\"; **junie** ← \"Junie\", \"JetBrains\"; **codex** ← \"Codex\", \"OpenAI Codex\".\n\n### `--items` \u002F `-Items` (optional)\n\nSingle comma-separated list of **basenames without extension** (e.g. `delegate-to-aside`, `commit-style`).\nApplied to agents, skills, and rules together.\n\n| Phrase | Value |\n|--------|-------|\n| \"only the aside skill\", \"just `delegate-to-aside`\" | `delegate-to-aside` |\n| \"just `foo` and `bar`\" | `foo,bar` |\n| \"full sync\", \"all of them\", no item mentioned | omit flag |\n\n**Caveat:** `--items` does not trim the guidelines merge.\n`sync-agent-guidelines` always emits the full `AGENTS.md` + merged rules for Claude\u002FJunie\u002FCodex.\nMention this in the report if the user expected partial guidelines.\n\n### `--clean` \u002F `-Clean` (optional; default: off)\n\nSet when the user asks to **remove stale outputs**, **prune**, **mirror source deletions**, **empty generated dirs first**, or says \"clean sync\".\n\n**Warning — data loss risk:** `--clean` empties the entire output directory for **every** sync step on **every** selected tool before writing: `.cursor\u002F{agents,rules,skills}`, `.claude\u002F{agents,skills}`, `.github\u002F{agents,instructions,skills}`, `.junie\u002F{agents,skills}`, `.codex\u002Fagents`, and `.agents\u002Fskills`.\nAny hand-authored files in those dirs are deleted.\nThis is especially risky for `.github\u002F` since users often keep unrelated content there (workflows live in `.github\u002Fworkflows\u002F` — not wiped — but a user might have put manual instruction files in `.github\u002Finstructions\u002F`).\nCodex MCP clean is narrower: it removes\u002Freplaces only `mcp_servers` sections in `.codex\u002Fconfig.toml`, preserving unrelated Codex config.\nFlag clean risk in your pre-run note whenever `-Clean` is inferred, and name the specific dirs or Codex config sections affected by the `--tools` selection.\n\n## Step 4: Validate and run\n\nBefore running:\n\n1. Check that `\u003Cinput_root>\u002FAGENTS.md` exists. If not, stop and tell the user.\n2. Use absolute paths in the command.\n3. Capture full stdout and stderr.\n\nExample — macOS \u002F Linux:\n\n```bash\nbash \"\u002Fabs\u002Fpath\u002Fto\u002Fcyncia\u002Fscripts\u002Fsync-all.sh\" \\\n  -i \"\u002Fabs\u002Fpath\u002Fto\u002Fsource-root\" \\\n  -o \"\u002Fabs\u002Fpath\u002Fto\u002Foutput-root\" \\\n  --tools cursor,claude \\\n  --items delegate-to-aside\n```\n\nWith clean:\n\n```bash\nbash \"\u002Fabs\u002Fpath\u002Fto\u002Fcyncia\u002Fscripts\u002Fsync-all.sh\" \\\n  -i \"\u002Fabs\u002Fpath\u002Fto\u002Fsource-root\" \\\n  -o \"\u002Fabs\u002Fpath\u002Fto\u002Foutput-root\" \\\n  --clean\n```\n\nExample — Windows PowerShell:\n\n```powershell\n& 'C:\\abs\\path\\to\\cyncia\\scripts\\sync-all.ps1' `\n  -InputRoot 'C:\\abs\\path\\to\\source-root' `\n  -OutputRoot 'C:\\abs\\path\\to\\output-root' `\n  -Tools 'cursor,claude' `\n  -Items 'delegate-to-aside'\n```\n\nOn non-zero exit, surface the error and suggest fixes (missing dirs, unknown tool name, missing `AGENTS.md`).\n\n## Step 5: Post-run report (required)\n\nReply with a **compact** summary, not the full log:\n\n1. **Command** — one line: OS\u002Fshell, script path, effective `-i`\u002F`-o`\u002F`--tools`\u002F`--items`\u002F`--clean`.\n   If `--tools` was omitted and `default-tools` is visible, mention the configured default that was used.\n2. **Deletions \u002F clean** — list paths from log lines matching `cleaned` \u002F `removed` \u002F `Clean`.\n   If `--clean` was not used, state that no clean step ran (overwrite in place only).\n3. **Generated \u002F updated** — from lines with `->` (e.g. `cursor agent ->`, `copilot skill ->`) and `== tool ==` headers.\n   Group by tool.\n   Call out `AGENTS.md`, `AGENTS.override.md`, `CLAUDE.md`, `.github\u002Fcopilot-instructions.md`, `.junie\u002FAGENTS.md`, `.codex\u002Fconfig.toml`, and `.codex\u002Fagents\u002F*.toml` when they appear.\n4. **Basis** — one short sentence explaining what you inferred from the user's request.\n\n## Edge cases\n\n- **MCP-only sync:** likewise, call `scripts\u002F\u003Ctool>\u002Fsync-mcp.{sh,ps1} -i \u003Csrc>\u002Fmcp-servers -o \u003Cout>` directly.\n  The Bash variants require `jq`.\n- **Rules-only sync:** `sync-all` has no such flag.\n  Call the per-tool script directly (e.g. `scripts\u002Fcursor\u002Fsync-rules.sh -i \u003Csrc>\u002Frules -o \u003Cout>`).\n- **Installed cyncia files \u002F monorepo:** always resolve real filesystem paths before running.\n- **Input equals output:** allowed; scripts skip the redundant `AGENTS.md` copy.\n- **Git Bash on Windows:** prefer the `.sh` script with `bash`; only use `.ps1` when the session is clearly native PowerShell.\n",{"name":5,"description":7},"main","f2834ae3a8d2b768340afe3edecf03e345e49171","skills\u002Fagent-conf-sync","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia","https:\u002F\u002Fwww.cyncia.net","https:\u002F\u002Fgithub.com\u002Fcrestreach","http:\u002F\u002Fwww.crestreach.com","2026-07-11 02:00:30",{"source":8,"skillName":5,"commands":58,"prompt":68,"zipAvailable":69},[59,62,65],{"tool":60,"command":61},"npx","npx skills add crestreach\u002Fcyncia --skill agent-conf-sync",{"tool":63,"command":64},"bunx","bunx skills add crestreach\u002Fcyncia --skill agent-conf-sync",{"tool":66,"command":67},"pnpm","pnpm dlx skills add crestreach\u002Fcyncia --skill agent-conf-sync","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Fskills\u002Fagent-conf-sync\u002FSKILL.md",true,[],[72,82,90,106,128],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":44},"01M71FRNND0JMSB7KJ25NG3BKV","beta","Second test skill.","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Ftwo-skills\u002Fskills\u002Fbeta\u002FSKILL.md",[],[74,79,80,81],"Test Skill","test","second",{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":44},"01M71FRN094CDHGCEJNA2ZQD67","alpha","First test skill.","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Ftest\u002Ffixtures\u002Ftwo-skills\u002Fskills\u002Falpha\u002FSKILL.md",[],[84,80,89,79],"first",{"skillNo":91,"name":92,"nameZh":6,"description":93,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":94,"categories":95,"keywords":96,"publishedAt":44},"01M71FREXPPVZB51AV724KMPRP","pirate-java-comments","Write code comments like a pirate, but only in Java files.","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Fexamples\u002Fskills\u002Fpirate-java-comments\u002FSKILL.md",[],[97,98,99,100,101,102,103,37,104,105],"pirate java comments","pirate","comments","java","only","files","code comments","but","write",{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":44},"01M71FRC45A867X4D111HMD5RB","delegate-to-aside","Delegates a side question to the `aside` subagent with a properly composed prompt so the main chat context stays clean. Use when the user says \"ask aside\", \"side question\", \"delegate this\", \"run in a subagent\", or otherwise signals they want the question answered in isolated context. Also use proactively when a request is clearly off the current work thread (research, lookups, \"how does X work?\") and would bloat the main context.","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002Fexamples\u002Fskills\u002Fdelegate-to-aside\u002FSKILL.md",[],[113,114,115,116,117,118,31,30,119,120,121,122,40,49,123,124,125,126,38,127],"delegate","delegate to aside","signals","research","delegates","how","work","with","chat","subagent","question","use","thread","prompt","ask",{"skillNo":129,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":130,"categories":131,"keywords":132,"publishedAt":44},"01M71FR98QWRQ3CQGZ5BV53S6A","https:\u002F\u002Fgithub.com\u002Fcrestreach\u002Fcyncia\u002Fblob\u002Fmain\u002F.agent-config\u002Fskills\u002Fdelegate-to-aside\u002FSKILL.md",[],[114,113,115,40,49,123,124,125,126,38,127,122,121,120,119,116,117,118,31,30]]