npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

ompstack

v0.2.1

Published

Risk-routed engineering workflow for OhMyPi

Readme

ompstack for OhMyPi

A small pstack-inspired workflow built around OhMyPi's native skill, task/subagent, custom-agent, and slash-command primitives.

It intentionally does not recreate OhMyPi's bundled task, scout, reviewer, or security-reviewer agents. The plugin adds:

  • skills/ompstack/ — risk routing, preflight contract, workflow evaluation, and real-surface playbooks
  • skills/ompstack-create-verification/ — creates project-native verify-<surface> skills and feature maps
  • skills/ompstack-maintain-verification/ — maintains a verification capability without changing product code
  • skills/ompstack-decision-trail/ — proportional append-only evidence trails for long-running work
  • agents/ompstack-architect.md — read-only design advisor
  • agents/ompstack-verifier.md — trusted runtime/behavior verifier instructed not to edit; execution tools are not a write sandbox
  • commands/ompstack.md and commands/ompstack-maintain-verification.md — convenience entry points

Install

Install the published plugin on another machine:

omp plugin install ompstack

Update an existing installation after a new npm release:

omp plugin upgrade ompstack

Install unreleased source from GitHub

Use a GitHub source install only when testing an unreleased commit:

omp plugin install github:ChiThang-50Cent/ompstack

Develop locally

Link the checkout without copying files into an OMP configuration directory:

omp plugin link .

Restart OMP, or run /reload-plugins in the active interactive session, before invoking the installed skill or command.

OMP discovery is first-wins when multiple plugins or configuration roots define the same skill, command, or agent name. If an update is not visible, confirm which plugin path won discovery, reload plugins, and spawn a fresh custom agent before diagnosing the new contract.

The same plugin layout is used for local links and direct GitHub installs:

.
├── package.json
├── skills/ompstack/
├── agents/
└── commands/

Invoke

Use the slash command:

/ompstack Fix the intermittent session-expiration regression. Reproduce it first.

Or invoke the skill directly:

/skill:ompstack Add idempotent webhook handling and verify duplicate delivery behavior.

The skill may also be selected automatically from its description when the task matches.

Design goal

Keep the pstack ideas that transfer cleanly to OMP:

risk route
   ↓
resolve project verification capability
   ↓
minimal discovery/design
   ↓
conditional native Todo progress state (parent-owned)
   ↓
one owner per write lane
   ↓
fan-in
   ↓
doctor + deterministic gates + real-surface drive
   ↓
independent review / behavior verification
   ↓
fresh affected proof + maintained feature map

Native Todo is conditional parent progress state for genuinely multi-phase, fan-in, blocked, explicit-progress, or handoff work. It is neither the task scheduler nor an audit log: Task/Hub owns worker lifecycle, while session artifacts and the decision trail retain durable evidence. The parent checks the native list before mutation and never overwrites a non-empty unrelated Todo list.

Avoid copying Cursor-specific cloud-agent, overnight-loop, PR-auto-merge, and Graphite machinery into a local OMP skill without a native equivalent. Reuse OMP-native project skills, task artifacts, transcripts, and Agent Hub instead.

Project-specific verification belongs at .omp/skills/verify-<surface>/SKILL.md. A matching skill names how to launch, Doctor-check, drive, observe, and clean up the real surface; its features/ directory records user-POV coverage. Create one explicitly with /skill:ompstack-create-verification; audit an existing one with /ompstack-maintain-verification.

When no matching project capability exists, the verification phase selects the closest proof driver: Browser through Eval for web interaction, live command interaction for CLI/TUI, an existing consumer drive for API/service behavior, DAP debugger observations for live-state mechanisms, and LSP plus an existing behavior pin for symbol refactors. Static checks support these surfaces; they do not replace them.

For autonomous, multi-phase, high-risk, or handoff work, ompstack-decision-trail keeps material decisions in .omp/audit/<task-slug>.tsv while linking to native history://, agent://, and artifact evidence.

Prewalk, Advisor, session handoff/export, and Memory are opt-in OMP operator facilities. Ompstack never enables or configures them: Prewalk cannot become a route requirement; Advisor remains inspection-only concern coverage rather than a completion gate; persisted session artifacts remain primary handoff evidence; and any memory:// context must be cited and revalidated against the current repository. Never share an export or capture a lesson without explicit authorization.

See docs/DESIGN.md, docs/BENCHMARK_EVIDENCE.md, and examples/usage.md.

Validate

Run the deterministic plugin contract checks before changing routing, custom agents, or playbooks:

bun run check

The check validates primary-route versus overlay/phase/progress-tracking wiring, structured custom-agent capabilities (tools, model, and blocking), preflight rules, self-contained task examples, and complete golden routing coverage. It remains a static contract check; the Eval playbook defines the separate blinded paired behavioral evaluation required for a workflow-policy change.

Optional unattended evidence adapter

scripts/run-verification-contract.mjs executes a versioned verification contract and emits one immutable evidence file per attempt. A v2 contract binds command identity, candidate and protected snapshots, timeout, output cap, oracle result schema, minimum executed tests, and a declared trust level.

scripts/run-orchestrated-task.mjs is an optional external controller. It runs an initial worker, evaluates the candidate, permits at most one configured repair worker after a valid NOT_VERIFIED result, and appends lifecycle events to a JSONL journal. It accepts completion only from a fresh VERIFIED candidate snapshot:

bun scripts/run-orchestrated-task.mjs --spec /trusted/orchestration.json
bun scripts/run-orchestrated-task.mjs --check /trusted/state/<run-id>/run.json

The controller is not an OMP extension and does not turn local execution into a security boundary. It records a caller-declared trust level; only a separately owned executor, oracle, and artifact store can make isolated or ci-attested meaningful.

Run the behavioral tests for the runner and controller with:

bun test tests

Attribution

ompstack adapts selected pstack workflow concepts to OMP-native primitives. See NOTICE.md and third_party/PSTACK_LICENSE.