SkillAgentSearch skills...

content

Content operations: editorial calendar, marketing, publishing, social media management.

Install / Use

npx skills add notque/vexjoy-agent --skill content

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

83/100

Category

Automation

Supported Platforms

Universal

Our assessment of content

content scores 83/100 on our quality scale, 2265th of 2,931 Automation skills we index.

Its SKILL.md is 11 KB long, well organised into 25 sections with 1 code example: a thorough specification that gives an agent plenty to work with.

It has 425 GitHub stars, a meaningful sign that others use it.

Substance
29/30
Structure
17/20
Description
12/15
Adoption
11/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated 12 days ago, so content is actively maintained.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-10-15. Automated pattern scan on 2026-10-15. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

content compared with similar skills

All 4 of these similar skills score higher than content; compare them before choosing.

SkillScoreStarsUpdatedFormat
content (this skill)by notque8342512d agoSKILL.md
Agent-Reachby Panniantong10097.6ktodayCLAUDE.md
Scraplingby D4Vinci10087.7ktodayMCP Server
LocalAIby mudler10049.5ktodayMCP Server
rufloby ruvnet10074.7ktodayMCP Server

Frequently asked questions

How do I install content?
Run npx skills add notque/vexjoy-agent --skill content. The install tabs above show the steps for each supported agent.
Which AI agents does content work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is content safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is content still maintained?
The repository was last updated 12 days ago, so content is actively maintained.

name: content description: "Content operations: editorial calendar, marketing, publishing, social media management." user-invocable: true allowed-tools:

  • Read
  • Write
  • Edit
  • Bash
  • Grep
  • Glob
  • WebFetch routing: force_route: false not_for: "prose voice and style (use writing), code documentation (use docs-sync-checker)" triggers:
    • "editorial calendar"
    • "content pipeline"
    • "content schedule"
    • "marketing"
    • "SEO"
    • "campaign"
    • "email sequence"
    • "outline post"
    • "pre-publish check"
    • "upload to wordpress"
    • "wordpress draft"
    • "post to X"
    • "tweet this"
    • "post tweet"
    • "read Bluesky"
    • "Bluesky feed"
    • "moderate Reddit"
    • "modqueue"
    • "headline"
    • "content strategy"
    • "audit links"
    • "audit images"
    • "batch edit posts"
    • "repurpose this"
    • "news collection" category: content pairs_with:
    • writing
    • workflow

Content Skill

Six modes: Calendar (editorial pipeline), Marketing (SEO, campaigns, email, competitive), Publish (outline, pre-publish, SEO, audits, WordPress), Social-X (X/Twitter posts and threads), Social-Bluesky (public feed reading), Social-Reddit (moderation). Classify the request and follow the matching section.

Mode Selection

| Mode | Signals | |------|---------| | Calendar | editorial calendar, content pipeline, ideas, scheduling, topics, headlines, news, series, repurpose | | Marketing | SEO audit, campaign, email sequence, content strategy, competitive analysis, brand review, performance | | Publish | outline post, pre-publish check, SEO optimize, batch edit, audit links, audit images, taxonomy, WordPress upload | | Social-X | post to X, tweet, thread, X API | | Social-Bluesky | read Bluesky, Bluesky feed, search Bluesky | | Social-Reddit | moderate Reddit, modqueue, reports |


Calendar Mode

Manage editorial content through 6 pipeline stages: Ideas, Outlined, Drafted, Editing, Ready, Published. All state lives in a single content-calendar.md file.

Workflow

  1. READ -- Load and parse content-calendar.md. Validate all sections exist. Do not trust memory; always re-read the actual file.
  2. EXECUTE -- Perform the requested operation:
    • View: count per stage, upcoming scheduled (14 days), in-progress, recent publications (30 days). Flag stuck content (14+ days in one stage).
    • Add idea: check for duplicate titles (case-insensitive), append to Ideas.
    • Move: forward only through defined sequence (Ideas -> Outlined -> Drafted -> Editing -> Ready -> Published). Record YYYY-MM-DD timestamp at each transition.
    • Schedule: set date on Ready items. Date must be today or future.
    • Archive: move Published entries older than current month to Historical sections.
  3. WRITE -- Persist changes, re-read file to verify.

Calendar Sub-modes

| Request | Load | |---------|------| | Topic brainstorming, idea generation | references/topic-brainstormer.md | | Headlines, title generation | references/headlines.md | | Series planning, multi-part content | references/series-planner.md | | Content repurposing, platform variants | references/content-engine.md | | News collection, item qualification | references/news-collection.md |


Marketing Mode

Seven sub-modes. Always load references/llm-marketing-failure-modes.md alongside mode-specific references for detection heuristics against generic copy, keyword stuffing, fabricated metrics.

| Sub-mode | Signals | Framework | Load | |----------|---------|-----------|------| | SEO_AUDIT | SEO audit, keywords, content gaps, technical SEO | RESEARCH -> AUDIT -> PRIORITIZE | references/seo-audit.md | | CAMPAIGN | Campaign plan, product launch, lead gen, budget | BRIEF -> PLAN -> CALENDAR | references/campaign-planning.md | | CONTENT | Content strategy, editorial calendar, funnel mapping | AUDIT -> PLAN -> DELIVER | references/content-strategy.md | | EMAIL | Email sequence, drip campaign, nurture flow | SEQUENCE -> DRAFT -> AUTOMATE | references/email-sequences.md | | COMPETITIVE | Competitor research, battlecard, positioning | RESEARCH -> COMPARE -> POSITION | references/seo-audit.md + references/campaign-planning.md | | BRAND | Brand review, voice check, messaging consistency | AUDIT -> EVALUATE -> RECOMMEND | references/content-strategy.md | | PERFORMANCE | Marketing report, campaign results, ROI | GATHER -> ANALYZE -> REPORT | references/campaign-planning.md |

SEO Audit Quick Steps

  1. Gather: URL/domain, audit type, target keywords, competitors (identify 2-3 via web search if not provided).
  2. Keyword research: classify by intent (informational, navigational, commercial, transactional). Assess difficulty and opportunity.
  3. On-page audit: title tags, meta descriptions, H1/H2, keyword usage, internal linking, image alt text, URL structure.
  4. Technical: page speed, mobile, structured data, crawlability, HTTPS, Core Web Vitals.
  5. Content gaps: competitor topic coverage, freshness, thin content, funnel gaps.
  6. Output: executive summary, keyword table (15-25), issues table, content gap recommendations, technical checklist, competitor matrix, prioritized action plan.

Publish Mode

Eight sub-modes covering the blog publishing pipeline.

| Sub-mode | Signals | Load | |----------|---------|------| | Outline | outline post, blog structure, article outline | references/outline.md | | Pre-publish | pre-publish check, Hugo validation, frontmatter | references/pre-publish.md | | SEO | check SEO, optimize SEO, meta description | references/seo.md | | Batch-edit | batch edit posts, bulk frontmatter, find/replace | references/batch-edit.md | | Link-audit | audit links, broken links, link health | references/link-audit.md | | Image-audit | audit images, alt text, image accessibility | references/image-audit.md | | Taxonomy | audit taxonomy, fix tags, merge categories | references/taxonomy.md | | WordPress | upload to WordPress, create draft, edit post | references/wordpress-upload.md |

Outline Quick Steps

  1. Gather: topic, target audience, desired format, word count range, SEO target keyword.
  2. Research: check existing coverage, identify gaps, gather supporting data.
  3. Structure: working title, hook, thesis, 3-7 sections with sub-points, conclusion CTA.
  4. Validate: logical flow, sufficient depth per section, SEO keyword placement plan.

Pre-publish Quick Steps

  1. Check: title, date, slug, description, tags, categories, featured image, author, draft=false.
  2. Content: spell check, broken internal links, image paths, code blocks render, meta descriptions.
  3. SEO: title tag (<60 chars), meta description (120-160 chars), keyword in H1 and first paragraph.

Social-X Mode

OAuth-authenticated X/Twitter posting via x-api-poster.py. 4-phase pipeline with confirmation gate.

Workflow

  1. VALIDATE -- Check credentials: python3 $HOME/.claude/scripts/x-api-poster.py post --dry-run --text "ping". Requires X_API_KEY, X_API_SECRET, X_ACCESS_TOKEN, X_ACCESS_SECRET, X_BEARER_TOKEN. Read-only ops need only X_BEARER_TOKEN.
  2. CONFIRM -- Show user the exact text. For threads: python3 $HOME/.claude/scripts/x-api-poster.py thread --dry-run --texts "part 1" "part 2". Enforce 280-char limit per tweet. Wait for explicit user confirmation.
  3. POST -- Execute with --confirmed flag: python3 $HOME/.claude/scripts/x-api-poster.py post --confirmed --text "text". For media: --media /path/to/image.png.
  4. VERIFY -- Report tweet URL and rate limit status.

Social-Bluesky Mode

Read public Bluesky feeds via AT Protocol. No auth required.

Commands

# Fetch recent posts
python3 ~/.claude/scripts/bluesky_reader.py feed --handle HANDLE --limit 20

# Search posts
python3 ~/.claude/scripts/bluesky_reader.py search --handle HANDLE --query "terms"

# JSON output for pipelines
python3 ~/.claude/scripts/bluesky_reader.py feed --handle HANDLE --json

Endpoint: https://public.api.bsky.app/xrpc/app.bsky.feed.getAuthorFeed. Limit: 1-100 posts. Search is local keyword filter (all words must appear, case-insensitive).


Social-Reddit Mode

Reddit moderation via PRAW. Three modes:

| Mode | Command | Behavior | |------|---------|----------| | Interactive | invoke directly | Fetch, classify, present, confirm actions | | Auto | /loop 10m /reddit-moderate --auto | Auto-action high-confidence, flag rest | | Dry-run | --dry-run | Show recommendations without acting |

Workflow

  1. FETCH -- python3 scripts/reddit-moderate/reddit-mod.py queue --json --limit 25 | python3 scripts/reddit-moderate/reddit-mod.py classify. Loads subreddit context from reddit-data/{subreddit}/.
  2. CLASSIFY -- Fill classification, confidence, and reasoning for each item using the assembled prompt.
  3. PRESENT -- Group by confidence tier (high/medium/low). Show: title, author, report reason, classification, action, reasoning.
  4. ACT -- Execute confirmed actions: python3 scripts/reddit-moderate/reddit-mod.py action --item-id ID --action approve|remove|spam.

Deep References

Load when the task needs detailed patterns, specifications, or examples.

Calendar

| Signal | Reference | |--------|-----------| | Pipeline stage definitions, transitions | references/pipeline-stages.md | | Calendar file format, sections | references/calendar-format.md | | Operation execution details | references/operations.md | | Pipeline health, velocity metrics | references/metrics.md | | Calendar error recovery | references/error-handling.md |

Marketing

| Signal | Reference | |--------|-----------| | SEO audit methodology | references/seo-audit.md | | Campaign planning, budget | references/campaign-planning.md | | Content strategy, frameworks | references/content-strategy.md | | Email sequences, drip campaigns | references/email-sequences.md | | LLM marketing failure modes | references/llm-marketing-failure-modes.md |

Publish

| Signal | Reference | |--------|-----------| | Blog outline methodology | references/outline.md | | Pre-publish Hugo checklist | references/pre-publish.md | | Post-level SEO | references/seo.md | | Batch editing methodology | references/batch-edit.md | | Link health scanning | references/link-audit.md | | Image audit, alt text | references/image-audit.md | | Taxonomy management | references/taxonomy.md | | WordPress API upload | references/wordpress-upload.md |

Social

| Signal | Reference | |--------|-----------| | X API details | references/x-api.md | | AT Protocol endpoints | references/at-protocol-api.md | | AT Protocol troubleshooting | references/at-protocol-preferred-patterns.md | | Reddit classification categories | references/classification-prompt.md | | Reddit script commands | references/script-commands.md |

Scripts

| Domain | Script | Location | |--------|--------|----------| | Reddit moderation | reddit-mod.py | scripts/reddit-moderate/reddit-mod.py | | Link scanning | link_scanner.py | scripts/publish/link_scanner.py | | PageSpeed | pagespeed.py | scripts/publish/pagespeed.py | | WordPress upload | wordpress-upload.py | scripts/publish/wordpress-upload.py | | WordPress edit | wordpress-edit-post.py | scripts/publish/wordpress-edit-post.py | | WordPress media | wordpress-media-upload.py | scripts/publish/wordpress-media-upload.py | | X API poster | x-api-poster.py | $HOME/.claude/scripts/x-api-poster.py | | Bluesky reader | bluesky_reader.py | $HOME/.claude/scripts/bluesky_reader.py |

Related Skills

View on GitHub
GitHub Stars425
CategoryAutomation
Updated12d ago
Forks48

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions
content — Universal Skill: Install & Safety Check | SkillAgent