Stop vibe coding.
Ship production-grade software.

npm install -g 100xprism && 100xprism install

curl -fsSL https://raw.githubusercontent.com/\
rajitsaha/100xprism/main/get.sh | bash

Either way, 100xprism install clones the toolkit to ~/100xprism and provisions your AI tools. The npm package is a thin launcher — modules, hooks and plugins live in that clone, which 100xprism update keeps current.

Set up a project

cd your-project && 100xprism init

The full CLI

These are every subcommand bin/100xprism.js dispatches.

Keeping up to date

100xprism update                    # pull latest, then reconcile skills + plugins
100xprism update --plugins-only     # refresh plugins only
100xprism update --no-slim          # skip the one-time always-on index slim
100xprism slim --dry-run            # preview what slimming would change
npm install -g 100xprism@latest     # upgrade the launcher itself

install and update are fully reconciling, not append-only. Every run adds newly shipped skills, slash commands and curated plugins; updates changed ones in place; and removes those deleted or merged upstream. Your own hand-authored skills and commands, and any plugins you enabled yourself, are never touched.

Multi-project setup

# every repo under ~/projects
for dir in ~/projects/*/; do
  [ -d "$dir/.git" ] && (cd "$dir" && 100xprism init)
done

To apply on every future clone, install a git template hook:

mkdir -p ~/.git-templates/hooks
cat > ~/.git-templates/hooks/post-checkout << 'HOOK'
#!/usr/bin/env bash
[ "$3" = "1" ] || exit 0                          # branch checkout only
PROJECT_ROOT="$(git rev-parse --show-toplevel)"
[ -f "$PROJECT_ROOT/.cursorrules" ] && exit 0     # already set up
command -v 100xprism >/dev/null && (cd "$PROJECT_ROOT" && 100xprism init)
HOOK
chmod +x ~/.git-templates/hooks/post-checkout
git config --global init.templateDir ~/.git-templates

GitHub Actions templates

mkdir -p .github/workflows
cp ~/100xprism/github-actions/ci.yml      .github/workflows/ci.yml
cp ~/100xprism/github-actions/release.yml .github/workflows/release.yml

ci.yml runs on every push and PR — lint, real-database tests and E2E. release.yml runs on v*.*.* tags and drives semantic release.

AI tools — configuration emit

What each tool receives when you run 100xprism init.

Token observability — a separate axis

Emit support and measurement support are not the same list. Measurement capability is declared per collector in scripts/adapters/registry.py and is deliberately kept separate from adapter implementation, so activity can never be silently treated as billable usage.

Operating systems

Database engines

/db executes SQL and migrations against a named connection. /query answers plain-English analytics questions and reaches a wider set of engines through their own CLIs.

Known documentation drift

Found by auditing every documented feature against its implementation. Listed here so nobody has to rediscover them.

Everything else

Lifecycle

Quality

Engineering

Data & infrastructure

Documentation & architecture

Retention classes

Every module derives one. A module can override it with retention: in frontmatter.

Measured footprint by surface

100xprism optimize                   # must-only for user scope + this repo
100xprism optimize --all-projects
100xprism optimize --skills=profile  # widen to detected project profiles
100xprism optimize --skills=all      # restore every module
100xprism audit --json               # standing-context estimate + inventory

Run it

100xprism tokens                  # offline dashboard, auto-refreshes every 30s
100xprism value                   # delivery economics
100xprism tokens --json           # versioned provider-counter report
100xprism tokens --json --tool codex
100xprism audit --json            # standing-context estimate

The four token purposes

Every observed token is classified into one of four purposes, then priced per model. Cache share matters because a high one means much of your context was served through prompt caching — it does not mean the work was free.

What the dashboard actually renders

Every view below exists today in 100xprism tokens.

How delivery is counted

/pack                    # every pack, install state, and anything detected here
/pack detect             # only what matches the current project
/pack add databricks     # install a pack
/pack remove databricks  # reverse only what 100xprism installed

Packs are ownership-tracked per install — /pack remove only removes what it added, never a plugin or command you enabled yourself. Where the vendor ships a CLI, that path is preferred; otherwise 100xPrism installs directly per platform.

Databricks is currently the only pack.

Frequently asked questions