Production-ready Claude skills for developers, knowledge workers, and marketers.
Skills are model-invoked playbooks: Claude loads them automatically when your request matches. These aren't "act as an expert" prompts — each one encodes a real, repeatable workflow with output templates, edge cases, and judgment rules that beat an improvised prompt.
🚀 New — free & open source: SocialForge — 20 Claude skills that turn any capable model into a full AI social-media growth agency (hooks, viral scripts, competitor analysis, repurposing, trends, engagement + an orchestrator). MIT. ⭐ it if it helps.
⚡ Also open source: Aether OS — a model-agnostic AI agent kernel: persistent memory, a governed multi-agent orchestrator (council/debate), an eval harness, and cost tracking. Apache-2.0. For builders who want the infrastructure under the skills.
Install in Claude Code:
/plugin marketplace add moses607/skillforge
/plugin install skillforge-free@skillforge| Skill | What it does |
|---|---|
pr-review-guardian |
Structured PR review with severity-tagged findings and a merge verdict |
commit-craft |
Conventional Commits + changelogs from a diff |
incident-postmortem |
Blameless postmortems with 5-Whys root cause and verifiable action items |
meeting-to-actions |
Transcript → decisions, owned action items, follow-up draft |
inbox-triage |
Emails → priority tiers + drafted replies |
weekly-review |
Guided GTD-style weekly reset and next-week planning |
landing-copy |
Conversion copy via hook → problem → mechanism → proof → CTA |
social-repurpose |
One long asset → platform-native posts (X, LinkedIn, IG, newsletter) |
newsletter-forge |
One-idea newsletter issues with strong opens and a single CTA |
The higher-leverage skills — the ones worth paying for — ship separately under a commercial license:
| Skill | What it does |
|---|---|
test-gap-hunter |
Finds untested behaviors (not just lines) and generates the missing tests |
api-contract-designer |
Versioned REST/GraphQL contracts + error model + OpenAPI stub |
refactor-surgeon |
Behavior-preserving, risk-ranked refactoring plans that stay green each step |
decision-brief |
One-page decision docs with a steelmanned counter-argument |
second-brain-capture |
Atomic, linkable Zettelkasten-style notes from any input |
okr-builder |
OKRs with task-vs-outcome tests and anti-gaming guardrails |
cold-email-sequence |
Multi-touch B2B outbound with deliverability discipline |
seo-brief |
Intent-driven content briefs a writer can execute to rank |
positioning-canvas |
Component-based product positioning (Dunford method) |
👉 Buy SkillForge Pro on Gumroad — $39
Instant download after checkout: a versioned .zip with all 9 Pro skills, under a commercial license (single user / single org). Includes 12 months of updates.
See CATALOG.md for the full skill list, PRICING.md for tiers, and GO-TO-MARKET.md for the distribution playbook.
The craft beneath the prompts — how to drive the model itself. Works with Opus 4.8, Sonnet 5, Haiku 4.5, Fable 5, and Claude Code.
| Skill | What it does |
|---|---|
token-economist |
Cut tokens, cost, and latency without losing quality |
loop-architect |
The Looping method: agentic loops that actually converge |
context-engineer |
Treat the context window as your scarcest resource; stop the model "forgetting" |
model-router |
The right model for each job + multi-model orchestration |
verification-harness |
Every AI output is a hypothesis until exercised — catch confidently-wrong results |
👉 Get SkillForge Master — $49 (5 skills)
Instant download: 5 skills + reference files, commercial license (single user / single org), 12 months of updates.
Five composable meta-skills that turn any capable model (Claude, GPT-class, or your own agent) into an autonomous system — not task templates, but the loops a senior AI engineer runs by hand:
| Skill | What it does |
|---|---|
meta-evolver |
Self-improvement loop: optimize any prompt/skill/agent against a real fitness function |
reality-simulator |
A flight simulator for decisions — model markets, businesses, and games before you commit |
cross-domain-synthesis |
Breakthrough ideas by transferring deep mechanisms across distant fields |
workflow-orchestrator |
Multi-agent teams + a persistent relational/vector/graph memory fabric, with self-healing |
value-amplifier |
Idea → validated MVP → growth simulation → monetization |
They compose: Meta-Evolver tunes the others; the Orchestrator runs them as a team. You buy the loop, not the parts.
👉 Get SkillForge Labs — $149 (bundle of 5)
Instant download: all 5 skills + reference files, commercial license (single user / single org), 12 months of updates.
A skill is a directory with a SKILL.md (YAML frontmatter + instructions). Claude keeps only the one-line description in context and loads the full body when your request matches. Works across Claude Code, the Claude apps, and the API.
Free pack: MIT. Pro pack: commercial (see the Pro download).
Open an issue for bugs or skill requests. PRs welcome on the free pack.