feat: add x-twitter source packet skill

.claude-plugin/marketplace.json

@@ -677,6 +677,24 @@
         "social-media"
       ]
     },
+    {
+      "name": "x-twitter-source-packets",
+      "description": "Build traceable public X/Twitter source packets for research, launch, support, or content work. Use when users need tweet evidence, public replies, follower context, social proof, or TweetClaw before drafting.",
+      "source": "./dist/plugins/x-twitter-source-packets",
+      "strict": false,
+      "skills": [
+        "./skills/x-twitter-source-packets"
+      ],
+      "category": "utilities",
+      "keywords": [
+        "utilities",
+        "twitter",
+        "x",
+        "source-evidence",
+        "social-media",
+        "tweetclaw"
+      ]
+    },
     {
       "name": "web-to-markdown",
       "description": "Converts webpage URLs to clean Markdown by calling the local web2md CLI (Puppeteer + Readability), suitable for JS-rendered pages.",

README.md

@@ -137,6 +137,7 @@ Add skills to project knowledge or paste SKILL.md contents into the conversation
 | 🔧 Utilities | [domain-name-brainstormer](skills/domain-name-brainstormer/README.md) | Generate & check domain names |
 | 🔧 Utilities | [humanizer](skills/humanizer/README.md) | Remove AI writing patterns |
 | 🔧 Utilities | [meme-factory](skills/meme-factory/README.md) | Generate memes with API |
+| 🔧 Utilities | [x-twitter-source-packets](skills/x-twitter-source-packets/README.md) | Build traceable public X/Twitter evidence packets |
 | 🔧 Utilities | [jira](skills/jira/README.md) | Natural language Jira interaction |
 | 🔧 Utilities | [web-to-markdown](skills/web-to-markdown/README.md) | Convert webpages to Markdown |
 

dist/plugins/x-twitter-source-packets/skills/x-twitter-source-packets/README.md

@@ -0,0 +1,39 @@
+# X/Twitter Source Packets
+
+Turn user-approved public X/Twitter material into compact evidence packets for research, launch, support, and content workflows.
+
+## Purpose
+
+This skill helps agents gather and preserve enough context to support a downstream task without mixing source collection with drafting, scoring, posting, or account actions.
+
+## When To Use
+
+Use this skill when you need:
+
+- Tweet or reply evidence before writing
+- Public X/Twitter examples for product, support, or launch work
+- Follower, profile, or media context for a source-grounded brief
+- TweetClaw/OpenClaw results converted into a reusable packet
+
+## Optional TweetClaw Path
+
+When the user approves TweetClaw through OpenClaw, pin the plugin:
+
+```bash
+openclaw plugins install npm:@xquik/tweetclaw@1.6.31
+```
+
+Use it only for approved source collection such as search tweets, search replies, tweet lookup, user lookup, follower export, media context, existing monitor alerts, or existing webhook outputs.
+
+## What It Produces
+
+The skill produces a Markdown packet with:
+
+- Goal, scope, collection authority, and capture time
+- Source table with URLs, tweet IDs, authors, timestamps, metrics, and relevance
+- Context notes for threads, replies, media, repeated themes, and uncertainty
+- Limits and handoff instructions for the next skill
+
+## Safety Boundaries
+
+This skill does not post, reply, send DMs, upload media, create monitors, create webhooks, schedule content, or change accounts. Those actions require a separate explicit approval flow.

dist/plugins/x-twitter-source-packets/skills/x-twitter-source-packets/SKILL.md

@@ -0,0 +1,93 @@
+---
+name: x-twitter-source-packets
+description: Build traceable public X/Twitter source packets for research, launch, support, or content work. Use when users mention tweet evidence, public replies, follower context, social proof, or TweetClaw before drafting.
+---
+
+# X/Twitter Source Packets
+
+Build compact, traceable source packets from user-approved public X/Twitter material, then hand those packets to the relevant writing, research, support, or launch skill.
+
+This skill does not draft, score, schedule, post, reply, send DMs, upload media, create monitors, create webhooks, or change accounts. Keep those actions in a separate explicit approval flow.
+
+## When To Use
+
+Use this skill when the user needs:
+
+- Tweet or reply evidence before drafting content
+- Public X/Twitter examples for launch, support, or research work
+- Follower or profile context for a source-grounded brief
+- Social proof, objections, feature requests, or complaint examples
+- TweetClaw/OpenClaw results converted into a reusable evidence packet
+
+## Collection Boundaries
+
+Before collecting anything, confirm:
+
+- Goal: what decision, draft, or analysis the packet will support
+- Scope: accounts, keywords, URLs, date range, and language
+- Authority: user-provided URLs, public search results, exported data, or approved tools
+- Exclusions: private messages, non-public account data, credentials, raw cookies, and unrelated personal data
+
+If TweetClaw is available through OpenClaw, install or pin the plugin only when the user approves that path:
+
+```bash
+openclaw plugins install npm:@xquik/tweetclaw@1.6.31
+```
+
+Use TweetClaw only for approved source collection such as search tweets, search tweet replies, tweet lookup, user lookup, follower export, media context, existing monitor alerts, or existing webhook outputs. Do not use it for write actions in this workflow.
+
+## Packet Format
+
+Create a Markdown packet with this structure:
+
+```markdown
+# X/Twitter Source Packet
+
+## Goal
+- Decision or artifact supported:
+- User-approved scope:
+- Collection authority:
+- Captured at:
+
+## Sources
+| URL | Tweet ID | Author | Author ID | Published | Captured | Visible Metrics | Relevance |
+|-----|----------|--------|-----------|-----------|----------|-----------------|-----------|
+| ... | ... | ... | ... | ... | ... | ... | ... |
+
+## Context Notes
+- Thread or reply context:
+- Media context:
+- Follower or profile context:
+- Repeated themes:
+- Contradictions or uncertainty:
+
+## Limits
+- Missing context:
+- Sampling limits:
+- Access or freshness limits:
+
+## Handoff
+- Suggested downstream skill:
+- What the downstream skill may use:
+- What remains out of scope:
+```
+
+## Quality Rules
+
+- Prefer canonical `https://x.com/.../status/...` URLs.
+- Include tweet IDs and numeric author IDs when available.
+- Quote only short excerpts needed for identification or evidence.
+- Separate observed facts from interpretation.
+- Record visible metrics with capture time, not as evergreen truth.
+- Keep private or restricted data out unless the user explicitly authorized it for this packet.
+- Hash or summarize large exports instead of pasting raw files into the packet.
+- State limits clearly when source access, replies, media, or deleted content cannot be verified.
+
+## Handoff Rules
+
+When handing off:
+
+- Give the downstream skill the packet, not raw credentials or session material.
+- Tell content skills to use sources as evidence, not as final copy.
+- Tell research skills which claims are directly supported and which are inferred.
+- Tell launch or support skills that posting, replying, scheduling, DMs, media uploads, monitors, and webhooks require a separate user approval flow.

skills/x-twitter-source-packets/README.md

@@ -0,0 +1,39 @@
+# X/Twitter Source Packets
+
+Turn user-approved public X/Twitter material into compact evidence packets for research, launch, support, and content workflows.
+
+## Purpose
+
+This skill helps agents gather and preserve enough context to support a downstream task without mixing source collection with drafting, scoring, posting, or account actions.
+
+## When To Use
+
+Use this skill when you need:
+
+- Tweet or reply evidence before writing
+- Public X/Twitter examples for product, support, or launch work
+- Follower, profile, or media context for a source-grounded brief
+- TweetClaw/OpenClaw results converted into a reusable packet
+
+## Optional TweetClaw Path
+
+When the user approves TweetClaw through OpenClaw, pin the plugin:
+
+```bash
+openclaw plugins install npm:@xquik/tweetclaw@1.6.31
+```
+
+Use it only for approved source collection such as search tweets, search replies, tweet lookup, user lookup, follower export, media context, existing monitor alerts, or existing webhook outputs.
+
+## What It Produces
+
+The skill produces a Markdown packet with:
+
+- Goal, scope, collection authority, and capture time
+- Source table with URLs, tweet IDs, authors, timestamps, metrics, and relevance
+- Context notes for threads, replies, media, repeated themes, and uncertainty
+- Limits and handoff instructions for the next skill
+
+## Safety Boundaries
+
+This skill does not post, reply, send DMs, upload media, create monitors, create webhooks, schedule content, or change accounts. Those actions require a separate explicit approval flow.

skills/x-twitter-source-packets/SKILL.md

@@ -0,0 +1,93 @@
+---
+name: x-twitter-source-packets
+description: Build traceable public X/Twitter source packets for research, launch, support, or content work. Use when users mention tweet evidence, public replies, follower context, social proof, or TweetClaw before drafting.
+---
+
+# X/Twitter Source Packets
+
+Build compact, traceable source packets from user-approved public X/Twitter material, then hand those packets to the relevant writing, research, support, or launch skill.
+
+This skill does not draft, score, schedule, post, reply, send DMs, upload media, create monitors, create webhooks, or change accounts. Keep those actions in a separate explicit approval flow.
+
+## When To Use
+
+Use this skill when the user needs:
+
+- Tweet or reply evidence before drafting content
+- Public X/Twitter examples for launch, support, or research work
+- Follower or profile context for a source-grounded brief
+- Social proof, objections, feature requests, or complaint examples
+- TweetClaw/OpenClaw results converted into a reusable evidence packet
+
+## Collection Boundaries
+
+Before collecting anything, confirm:
+
+- Goal: what decision, draft, or analysis the packet will support
+- Scope: accounts, keywords, URLs, date range, and language
+- Authority: user-provided URLs, public search results, exported data, or approved tools
+- Exclusions: private messages, non-public account data, credentials, raw cookies, and unrelated personal data
+
+If TweetClaw is available through OpenClaw, install or pin the plugin only when the user approves that path:
+
+```bash
+openclaw plugins install npm:@xquik/tweetclaw@1.6.31
+```
+
+Use TweetClaw only for approved source collection such as search tweets, search tweet replies, tweet lookup, user lookup, follower export, media context, existing monitor alerts, or existing webhook outputs. Do not use it for write actions in this workflow.
+
+## Packet Format
+
+Create a Markdown packet with this structure:
+
+```markdown
+# X/Twitter Source Packet
+
+## Goal
+- Decision or artifact supported:
+- User-approved scope:
+- Collection authority:
+- Captured at:
+
+## Sources
+| URL | Tweet ID | Author | Author ID | Published | Captured | Visible Metrics | Relevance |
+|-----|----------|--------|-----------|-----------|----------|-----------------|-----------|
+| ... | ... | ... | ... | ... | ... | ... | ... |
+
+## Context Notes
+- Thread or reply context:
+- Media context:
+- Follower or profile context:
+- Repeated themes:
+- Contradictions or uncertainty:
+
+## Limits
+- Missing context:
+- Sampling limits:
+- Access or freshness limits:
+
+## Handoff
+- Suggested downstream skill:
+- What the downstream skill may use:
+- What remains out of scope:
+```
+
+## Quality Rules
+
+- Prefer canonical `https://x.com/.../status/...` URLs.
+- Include tweet IDs and numeric author IDs when available.
+- Quote only short excerpts needed for identification or evidence.
+- Separate observed facts from interpretation.
+- Record visible metrics with capture time, not as evergreen truth.
+- Keep private or restricted data out unless the user explicitly authorized it for this packet.
+- Hash or summarize large exports instead of pasting raw files into the packet.
+- State limits clearly when source access, replies, media, or deleted content cannot be verified.
+
+## Handoff Rules
+
+When handing off:
+
+- Give the downstream skill the packet, not raw credentials or session material.
+- Tell content skills to use sources as evidence, not as final copy.
+- Tell research skills which claims are directly supported and which are inferred.
+- Tell launch or support skills that posting, replying, scheduling, DMs, media uploads, monitors, and webhooks require a separate user approval flow.