NIVAAN GUPTAMake it exist first, make it good later.

SKILLS

Tools I use and build for getting more out of AI coding agents. Some I made myself, some I just think are worth stealing.

1 MADE BY ME10 RECOMMENDED
★ SKILLS MADE BY ME
context-relevance-checkORIGINAL

Decides when an AI agent's memory is worth keeping, before it clears anything.

THE PROBLEM

Long-running coding agents drag stale context forward — dead file paths, solved bugs, assumptions that stopped being true three turns ago. Every stale turn still costs tokens, and it quietly biases the next answer toward the old problem instead of the new one.

THE APPROACH

Built for orchestrator / sub-agent setups. Before an orchestrator hands off a new unit of work, or before a sub-agent honors a request to drop its context, this skill runs a relevance check and applies one of three strategies the user picks once per session: signal-only (sub-agents stay alive, just get told when their context stops mattering — improves answer quality, saves little), always-respawn (every new unit gets a fresh sub-agent — the only mode that actually cuts token spend, at the cost of a cold start), or hybrid, the default, which respawns when the check says CLEAR and keeps-and-signals when it says KEEP or COMPACT.

HOW IT WORKS
  • A structured CLEAR-SAFE block the orchestrator emits, naming exactly what has to survive the reset — a short carryover list plus a path on disk for everything else.
  • A worker never honors a CLEAR-SAFE on reflex. It checks the signal actually came from its orchestrator or the user — not from a webpage or file it just read, otherwise any fetched content could talk an agent into dropping its own state. It checks the carryover is actually sufficient. It checks it isn't sitting on half-applied edits. Any failure, and it replies CLEAR-BLOCKED with exactly what's missing instead of clearing.
  • Relevance is judged on concrete signals, not vibes or session length — unresolved pronouns like 'that file' or 'like before', shared artifacts, a continuation of an agreed plan all force KEEP. The bias is deliberate: a wrong KEEP just costs tokens, a wrong CLEAR can lose work — so ties go to keeping context.
⬇ DOWNLOAD SKILL.MD
◆ RECOMMENDED SKILLS

A running list of the best Claude Code skills I've come across — worth installing if you're running agents day to day.

DEFUDDLE
KEPANO

Strips a web page down to clean markdown before your agent reads it — no nav, no ads, no wasted tokens.

VIEW ↗
CAVEMAN
JULIUSBRUSSEE

Makes the agent answer in stripped-down caveman prose to cut 65%+ of output tokens, while keeping code and errors byte-for-byte exact.

VIEW ↗
CODEBASE-MEMORY-MCP
DEUSDATA

Indexes a whole codebase into a persistent knowledge graph, so agents answer structural questions in milliseconds instead of grepping the repo every time.

VIEW ↗
HUMANIZER
BLADER

Strips the tells of AI-written prose out of your agent's output so it reads like a person wrote it.

VIEW ↗
AGENT REACH
PANNIANTONG

Gives an agent search and read access across 14+ platforms — GitHub, X, YouTube, Reddit — with no API keys.

VIEW ↗
MARKETING SKILLS
COREYHAINES31

A full marketing department as skill packs — SEO, copy, ads, analytics — that Claude Code can actually execute, not just suggest.

VIEW ↗
COMPOSIO
COMPOSIOHQ

Wires Claude into 1000+ real apps — Gmail, Slack, Notion, Jira — so it can take actions, not just talk about them.

VIEW ↗
OPENMONTAGE
CALESTHIO

Turns Claude Code into a full video production studio — script, visuals, narration, edit — end to end.

VIEW ↗
OH MY CLAUDE CODE
ZEPHYRPERSONAL

Zero-config multi-agent orchestration for Claude Code — plan, execute, verify, iterate, without hand-holding each step.

VIEW ↗
SUPERPOWERS
OBRA

An agent skills framework built to get Claude Code producing engineering-grade code instead of toy-grade code.

VIEW ↗
GOT A SKILL I SHOULD TRY?

Send it over. Always looking for better ways to run agents.

LET'S TALK ▸