@1aboveio/skills 0.20.4 → 0.21.0

package/README.md

@@ -73,6 +73,7 @@ planned same-named skill. The coordinator still owns group and new-member consen
 - `payment` — Payment: fraud-analysis, payment-analysis, tld-detection, similar-domain
 - `document-utilities` — Document Utilities: doc-to-md, reg-parser (required: reg-parser -> doc-to-md)
 - `compliance` — Compliance: compliance-audit, doc-authenticity, report-writing (required: compliance-audit -> doc-to-md; optional: compliance-audit -> reg-parser)
+- `content-creator` — Content Creator: article-review, article-to-image, wechat-article-writer, wechat-publisher, wechat-search
 - `fullstack` — Fullstack: better-auth, monorepo, nextjs-fullstack, prisma-setup, shadcn, zod-v4
 
 Internal dependencies, installed with the groups above and not selectable on their own: `harness-runtime`.
@@ -100,10 +101,11 @@ non-interactively.
 Declined skills are never uninstalled by the coordinator and stay out of the remembered acceptance,
 so the next update asks again.
 
-When a release archives a previously accepted skill, `update` asks before removing that installed
-copy. The default is No; `--yes` accepts archival non-interactively. Removal is performed by the
-same pinned native Skills CLI, only after the declared replacement is installed or already current.
-Declined or failed archival remains in selection memory so a later update offers it again.
+When a selected group still has a leftover retired skill (from prior acceptance or still
+installed), `install` and `update` archive it by default through the same pinned native Skills CLI,
+only after the declared replacement is installed or already current. An interactive prompt defaults
+to Yes; non-interactive runs archive without `--yes`. Declined or failed archival remains in
+selection memory so a later run offers it again.
 
 Install and update are best-effort per member over bulk source commands: they continue after a
 member failure except for skills whose required dependency is unavailable, summarize
@@ -139,6 +141,8 @@ npx skills@1.5.22 remove --global --agent universal claude-code --skill fraud-an
 npx skills@1.5.22 remove --global --agent universal claude-code --skill doc-to-md reg-parser
 # Compliance
 npx skills@1.5.22 remove --global --agent universal claude-code --skill compliance-audit doc-authenticity report-writing
+# Content Creator
+npx skills@1.5.22 remove --global --agent universal claude-code --skill article-review article-to-image wechat-article-writer wechat-publisher wechat-search
 # Fullstack
 npx skills@1.5.22 remove --global --agent universal claude-code --skill better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4
 ```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@1aboveio/skills",
-  "version": "0.20.4",
+  "version": "0.21.0",
   "description": "Install the 1AboveIO first-party skill groups through the native Skills CLI.",
   "license": "MIT",
   "type": "module",

package/runtime/skills/distribution/generated/recipes.json

@@ -24,7 +24,7 @@
       "id": "first-party",
       "type": "first-party",
       "package": "@1aboveio/skills",
-      "version": "0.20.4",
+      "version": "0.21.0",
       "updatePolicy": "pinned-npm-version",
       "members": [
         {
@@ -204,6 +204,36 @@
           "role": "skill",
           "licenseResponsibility": "1aboveio-mit"
         },
+        {
+          "installName": "article-review",
+          "path": "skills/content-creator/article-review",
+          "role": "skill",
+          "licenseResponsibility": "1aboveio-mit"
+        },
+        {
+          "installName": "article-to-image",
+          "path": "skills/content-creator/article-to-image",
+          "role": "skill",
+          "licenseResponsibility": "1aboveio-mit"
+        },
+        {
+          "installName": "wechat-article-writer",
+          "path": "skills/content-creator/wechat-article-writer",
+          "role": "skill",
+          "licenseResponsibility": "1aboveio-mit"
+        },
+        {
+          "installName": "wechat-publisher",
+          "path": "skills/content-creator/wechat-publisher",
+          "role": "skill",
+          "licenseResponsibility": "1aboveio-mit"
+        },
+        {
+          "installName": "wechat-search",
+          "path": "skills/content-creator/wechat-search",
+          "role": "skill",
+          "licenseResponsibility": "1aboveio-mit"
+        },
         {
           "installName": "better-auth",
           "path": "skills/fullstack/better-auth",
@@ -263,7 +293,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -297,7 +327,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -341,7 +371,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -374,7 +404,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -417,7 +447,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -445,7 +475,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -483,7 +513,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -511,7 +541,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -549,7 +579,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -579,7 +609,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -619,7 +649,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -647,7 +677,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -685,7 +715,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -715,7 +745,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -742,6 +772,78 @@
         "removesDependencies": false
       }
     },
+    {
+      "id": "content-creator",
+      "displayName": "Content Creator",
+      "dependsOn": [
+        "harness-runtime"
+      ],
+      "install": {
+        "stopOnFailure": true,
+        "steps": [
+          {
+            "sourceId": "first-party",
+            "sourceType": "first-party",
+            "package": "@1aboveio/skills",
+            "version": "0.21.0",
+            "installPath": null,
+            "members": [
+              "harness-runtime",
+              "article-review",
+              "article-to-image",
+              "wechat-article-writer",
+              "wechat-publisher",
+              "wechat-search"
+            ],
+            "commands": [
+              {
+                "transport": "local",
+                "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime article-review article-to-image wechat-article-writer wechat-publisher wechat-search"
+              }
+            ],
+            "onFailure": {
+              "action": "stop",
+              "message": "Source first-party fetch failed; stop this recipe before running another source."
+            }
+          }
+        ]
+      },
+      "update": {
+        "mode": "rerun-install-recipe",
+        "stopOnFailure": true,
+        "steps": [
+          {
+            "sourceId": "first-party",
+            "sourceType": "first-party",
+            "package": "@1aboveio/skills",
+            "version": "0.21.0",
+            "installPath": null,
+            "members": [
+              "harness-runtime",
+              "article-review",
+              "article-to-image",
+              "wechat-article-writer",
+              "wechat-publisher",
+              "wechat-search"
+            ],
+            "commands": [
+              {
+                "transport": "local",
+                "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime article-review article-to-image wechat-article-writer wechat-publisher wechat-search"
+              }
+            ],
+            "onFailure": {
+              "action": "stop",
+              "message": "Source first-party fetch failed; stop this recipe before running another source."
+            }
+          }
+        ]
+      },
+      "uninstall": {
+        "command": "npx skills@1.5.22 remove --global --agent universal claude-code --skill article-review article-to-image wechat-article-writer wechat-publisher wechat-search",
+        "removesDependencies": false
+      }
+    },
     {
       "id": "fullstack",
       "displayName": "Fullstack",
@@ -755,7 +857,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -787,7 +889,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime",
@@ -827,7 +929,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime"
@@ -853,7 +955,7 @@
             "sourceId": "first-party",
             "sourceType": "first-party",
             "package": "@1aboveio/skills",
-            "version": "0.20.4",
+            "version": "0.21.0",
             "installPath": null,
             "members": [
               "harness-runtime"
@@ -885,7 +987,7 @@
           "sourceId": "first-party",
           "sourceType": "first-party",
           "package": "@1aboveio/skills",
-          "version": "0.20.4",
+          "version": "0.21.0",
           "installPath": null,
           "members": [
             "harness-runtime",
@@ -917,6 +1019,11 @@
             "compliance-audit",
             "doc-authenticity",
             "report-writing",
+            "article-review",
+            "article-to-image",
+            "wechat-article-writer",
+            "wechat-publisher",
+            "wechat-search",
             "better-auth",
             "monorepo",
             "nextjs-fullstack",
@@ -927,7 +1034,7 @@
           "commands": [
             {
               "transport": "local",
-              "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 && npm --prefix \"$(realpath \"$HOME/.agents/skills/engineering-runtime/scripts\")\" ci --ignore-scripts"
+              "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing article-review article-to-image wechat-article-writer wechat-publisher wechat-search better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 && npm --prefix \"$(realpath \"$HOME/.agents/skills/engineering-runtime/scripts\")\" ci --ignore-scripts"
             }
           ],
           "onFailure": {
@@ -945,7 +1052,7 @@
           "sourceId": "first-party",
           "sourceType": "first-party",
           "package": "@1aboveio/skills",
-          "version": "0.20.4",
+          "version": "0.21.0",
           "installPath": null,
           "members": [
             "harness-runtime",
@@ -977,6 +1084,11 @@
             "compliance-audit",
             "doc-authenticity",
             "report-writing",
+            "article-review",
+            "article-to-image",
+            "wechat-article-writer",
+            "wechat-publisher",
+            "wechat-search",
             "better-auth",
             "monorepo",
             "nextjs-fullstack",
@@ -987,7 +1099,7 @@
           "commands": [
             {
               "transport": "local",
-              "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 && npm --prefix \"$(realpath \"$HOME/.agents/skills/engineering-runtime/scripts\")\" ci --ignore-scripts"
+              "command": "npx skills@1.5.22 add . --global --agent universal claude-code --skill harness-runtime e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing article-review article-to-image wechat-article-writer wechat-publisher wechat-search better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 && npm --prefix \"$(realpath \"$HOME/.agents/skills/engineering-runtime/scripts\")\" ci --ignore-scripts"
             }
           ],
           "onFailure": {
@@ -998,7 +1110,7 @@
       ]
     },
     "uninstall": {
-      "command": "npx skills@1.5.22 remove --global --agent universal claude-code --skill e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 harness-runtime"
+      "command": "npx skills@1.5.22 remove --global --agent universal claude-code --skill e2e-test ensure-coverage implement-and-pr rush-issues rush-release review-pr smoke engineering-runtime cloud-build cloud-debug cloud-deploy docker google-cloud mergify podman app-debug python-backend airflow pyspark fraud-analysis payment-analysis tld-detection similar-domain doc-to-md reg-parser compliance-audit doc-authenticity report-writing article-review article-to-image wechat-article-writer wechat-publisher wechat-search better-auth monorepo nextjs-fullstack prisma-setup shadcn zod-v4 harness-runtime"
     }
   }
 }

package/runtime/skills/distribution/scripts/bundles.mjs

@@ -204,18 +204,22 @@ export function newMemberCandidates(action, groupIds, priorState, transport = {}
   return candidates;
 }
 
-// Archived members are offered only to users who previously accepted them in the same group.
-// This keeps retirement explicit and prevents a release from treating filesystem presence as
-// consent. A successful removal drops the old acceptance; a decline or failure preserves it so a
-// later update offers the archive again.
+// Archived members of a selected group are offered when they were previously accepted or are
+// still installed. Presence of a named leftover is not group inference: the group is already
+// selected. A successful removal drops the old acceptance; a decline or failure preserves it so a
+// later run offers the archive again.
 export function archivedMemberCandidates(action, groupIds, priorState, transport = {}) {
-  if (action !== 'update' || !priorState) return [];
+  if (!ACTIONS.has(action)) return [];
   const recipes = readRecipes(resolveBundleTransport(transport).recipesPath);
   const selected = new Set(groupIds);
-  const acceptedByGroup = new Map(priorState.groups
+  const acceptedByGroup = new Map((priorState?.groups ?? [])
     .map((group) => [group.id, new Set(group.acceptedMembers)]));
+  const leftover = (name) => Boolean(
+    transport.skillsRoot
+    && statSync(join(transport.skillsRoot, name), { throwIfNoEntry: false }),
+  );
   return (recipes.archives ?? []).filter((archive) => selected.has(archive.formerGroupId)
-    && acceptedByGroup.get(archive.formerGroupId)?.has(archive.installName));
+    && (acceptedByGroup.get(archive.formerGroupId)?.has(archive.installName) || leftover(archive.installName)));
 }
 
 // Prior acceptance for the new-member consent delta on explicit selections (picker, --group,
@@ -686,8 +690,8 @@ install without --group/--all opens the interactive picker; update without them
 remembered selection (\${XDG_STATE_HOME:-~/.local/state}/1aboveio-skills/selection.json).
 When a remembered group's current recipe declares members you have not accepted yet, each new skill
 is offered individually and defaults to No; --yes accepts them all, --dry-run lists them without
-prompting. Retired accepted skills are likewise kept by default; --yes archives them through the
-native Skills CLI after their replacement is current. TDD implicit model invocation is enabled by default; --disable-tdd makes TDD explicit-only,
+prompting. Retired leftovers in a selected group are archived by default through the native Skills
+CLI after their replacement is current; decline interactively to keep them. TDD implicit model invocation is enabled by default; --disable-tdd makes TDD explicit-only,
 and --enable-tdd restores implicit invocation. --manual-tdd remains a compatibility alias for
 --disable-tdd.
 `;
@@ -862,29 +866,24 @@ async function consentNewMembers(candidates, { yes, stdin, stdout, stderr }) {
   return declined;
 }
 
-async function consentArchivedMembers(candidates, { yes, stdin, stdout, stderr }) {
+async function consentArchivedMembers(candidates, { yes, stdin, stdout }) {
   const declined = new Set();
-  if (yes) {
-    for (const candidate of candidates) stdout.write(`Accepting archive of retired skill "${candidate.installName}" (--yes)\n`);
-    return declined;
-  }
-  if (!stdin.isTTY) {
+  if (yes || !stdin.isTTY) {
+    const how = yes ? '--yes' : 'default Yes';
     for (const candidate of candidates) {
-      stderr.write(`Keeping retired skill "${candidate.installName}": no interactive terminal (default No; re-run with --yes to archive)\n`);
-      declined.add(candidate.installName);
+      stdout.write(`Archiving retired skill "${candidate.installName}" (${how}; replaced by ${candidate.replacement})\n`);
     }
     return declined;
   }
   const prompter = createLinePrompter({ stdin, stdout });
   for (const candidate of candidates) {
     const answer = await prompter.ask(
-      `Archive retired skill "${candidate.installName}" (replaced by ${candidate.replacement})? [y/N] `,
+      `Archive retired skill "${candidate.installName}" (replaced by ${candidate.replacement})? [Y/n] `,
     );
-    if (/^y(es)?$/i.test(answer)) stdout.write(`Archiving ${candidate.installName}\n`);
-    else {
+    if (/^n(o)?$/i.test(answer)) {
       stdout.write(`Keeping ${candidate.installName}\n`);
       declined.add(candidate.installName);
-    }
+    } else stdout.write(`Archiving ${candidate.installName}\n`);
   }
   return declined;
 }
@@ -993,18 +992,21 @@ export async function runCli(args, {
     } else if (candidates.length > 0) {
       declined = await consentNewMembers(candidates, { yes: parsed.yes, stdin, stdout, stderr });
     }
-    const archiveCandidates = archivedMemberCandidates(parsed.action, groupIds, priorState, transport);
+    const skillsRoot = join(env.HOME || homedir(), '.agents', 'skills');
+    const archiveCandidates = archivedMemberCandidates(parsed.action, groupIds, priorState, {
+      ...transport,
+      skillsRoot,
+    });
     let declinedArchives = new Set();
     if (archiveCandidates.length > 0 && parsed.dryRun) {
       for (const candidate of archiveCandidates) {
-        stdout.write(`Retired skill "${candidate.installName}" (replaced by ${candidate.replacement}) — would prompt to archive (default: No); --yes accepts archival\n`);
+        stdout.write(`Retired skill "${candidate.installName}" (replaced by ${candidate.replacement}) — would archive by default; decline interactively to keep\n`);
       }
     } else if (archiveCandidates.length > 0) {
       declinedArchives = await consentArchivedMembers(archiveCandidates, {
         yes: parsed.yes,
         stdin,
         stdout,
-        stderr,
       });
     }
     // Digest skip (#1127): `update` classifies the accepted members against the installed tree and
@@ -1013,7 +1015,6 @@ export async function runCli(args, {
     // from classification: they never run, so they are not buckets.
     let buckets;
     let skipMembers;
-    const skillsRoot = join(env.HOME || homedir(), '.agents', 'skills');
     if (parsed.action === 'update') {
       buckets = classifySelectedMembers(parsed.action, groupIds, { ...transport, enableTdd, skillsRoot });
       skipMembers = new Set(buckets.unchanged.map((entry) => entry.member));

package/runtime/skills/engineering/engineering-runtime/scripts/workflow-policy.mjs

@@ -86,7 +86,7 @@ export const WORKFLOW_TRUSTED_SOURCES = deepFreeze({
     id: 'first-party',
     type: 'first-party',
     package: '@1aboveio/skills',
-    version: '0.20.4',
+    version: '0.21.0',
   },
   matt: {
     id: 'matt-pocock',

package/skills/compliance/doc-authenticity/SKILL.md

@@ -1,7 +1,8 @@
 ---
 name: doc-authenticity
-version: 0.4.0
+version: 0.4.1
 description: "Screen PDF documents and raster images for fabrication, tampering, Photoshop or other image-editor use, and content validity. Detects file type, runs format-specific authenticity forensics, classifies bank statements, conversation logs, invoices, agreements, and other documents, then applies type-specific validity checks. Produces separate authenticity (NO_EVIDENCE_OF_MODIFICATION / SUSPICIOUS / LIKELY_MODIFIED / INCONCLUSIVE) and validity (VALID / INVALID / INCONCLUSIVE) verdicts. Use whenever a user asks whether a document, screenshot, scan, statement, invoice, contract, chat log, receipt, or payment evidence is genuine, original, edited, Photoshopped, manipulated, fake, or valid."
+disable-model-invocation: true
 ---
 
 # Document and Image Authenticity

package/skills/compliance/doc-authenticity/agents/openai.yaml

@@ -0,0 +1,8 @@
+interface:
+  display_name: "Document Authenticity"
+  short_description: "Explicit-only document and image authenticity screening"
+
+policy:
+  # Codex counterpart to SKILL.md disable-model-invocation: true.
+  # Keeps explicit invocation while blocking description-based selection.
+  allow_implicit_invocation: false

package/skills/content-creator/article-review/LICENSE

@@ -0,0 +1,3 @@
+MIT License
+
+Copyright (c) 2026 1AboveIO

package/skills/content-creator/article-review/SKILL.md

@@ -0,0 +1,65 @@
+---
+name: article-review
+description: >
+  Evaluate and score articles across multiple dimensions including originality,
+  readability, shareability, timeliness, and factual accuracy.
+  Use when a user shares an article, pastes article text, sends a URL, or asks to
+  review, rate, critique, evaluate, 评价, 点评, 分析, or 看看 any article, post, or 文章.
+  Default template is 公众号-支付行业.
+---
+
+# 文章评审引擎
+
+模板化的文章评审系统。评审标准由模板文件定义，引擎负责执行评审流程并输出标准化报告。
+
+**开始任何评审前，先读取 `references/report-format.md`**，熟悉输出结构。
+
+## 可用模板
+
+- **公众号-支付行业** (`references/template-wechat-payments.md`) — 面向支付/金融科技从业者的公众号内容（默认）
+
+如需其他模板，或用户请求的模板不存在，读取 `references/template-guide.md` 创建新模板，再继续评审。
+
+## 输入处理
+
+**粘贴文字：** 直接进入评审流程。
+
+**URL：** 用 `web_fetch` 抓取文章正文。若被拦截（如微信需要登录），提示用户粘贴文章内容后继续。
+
+## 评审流程
+
+### 第一步：确认模板
+在每次评审开始时，读取对应模板文件（默认 `references/template-wechat-payments.md`），获取维度列表和 Level 2 细则。用户未指定模板时使用默认。
+
+### 第二步：通读文章
+通读全文，理解核心主张、论据、结构、受众定位。
+
+### 第三步：事实核查（Fact-Check Pass）
+在逐维度评分之前，先做事实核查：
+
+1. 提取文章中所有可验证的主张：
+   - 数字、统计、比例、排名
+   - 引用他人言论或研究结论
+   - 行业事实陈述（市场规模、政策、技术参数等）
+
+2. 对每条主张用 `web_search` 核查，标注结果：
+   - ✅ **已核实** — 找到一致的可信来源
+   - ❌ **有误** — 找到与之矛盾的可信来源，记录正确值
+   - ⚠️ **无法核实** — 无法找到来源或来源相互矛盾，标注存疑
+
+3. 将核查结果用于事实准确性维度的评分依据
+
+### 第四步：逐维度评分
+按模板定义的每个 Level 1 维度：
+- 给出 1-10 分
+- 引用文章原文 2-3 处作为依据
+- 给出 1-2 条具体可执行的改进建议（如有）
+
+事实准确性维度的评分必须基于第三步的核查结果，而非主观判断。
+
+### 第五步：输出报告
+按 `references/report-format.md` 中的模板输出完整报告。
+
+## 注意事项
+
+严格按流程执行：评分必须有原文引用，不接受笼统判断；事实准确性必须基于 web_search 核查；改进建议要具体可执行；发布建议根据综合分和事实核查结果综合判断。