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

grounded-engineering

v0.5.0

Published

Source-backed engineering practices for professional software development and AI-assisted work.

Readme

Grounded Engineering

Grounded Engineering is a source-backed catalog of practical engineering habits for professional software development, including AI-assisted development, traditional development, and repository onboarding.

The project turns observations from mature engineering and agent repositories into small, reviewable practices. Each recommendation keeps its evidence, scope, confidence, and validation status visible so contributors can improve it instead of treating it as unexplained doctrine.

Repository layers

research/       Source observations, pinned references, and category audits
practices/      Short, reusable engineering-practice cards
integrations/   Consumer-specific translation guidance for agent instruction files
scripts/        Deterministic local validation

The layers are intentionally separate. Research preserves provenance and decision history; practice cards are the concise consumption surface; integrations explain how to adapt cards to a local repository without creating another source of truth.

Current status

v0.5.0 is the current release. It ships the historical eight-card baseline pack, the thirteen-card ai-assisted pack, provider-neutral Markdown output, the Codex AGENTS.md adapter, the Claude root CLAUDE.md adapter, the read-only check command for managed-block drift detection, the per-source validation-provenance floor, and the opt-in check:sources online drift check.

The historical baseline pack metadata remains at v0.2.0 with pack_version: 1.0.0 so existing adopters can stay green while update propose is still reserved. The ai-assisted profile adds the full thirteen-card set, but it does not rewrite existing policy; it creates a reviewable proposal that the consuming repository must explicitly accept.

The recommendations are point-in-time observations against pinned sources. Source changes do not silently rewrite cards; re-auditing is a deliberate maintenance step. A card cannot report a validation state stronger than not_validated without recording, for each source, the full revision set it was checked against. A source re-pin is caught offline until the card is re-validated.

Grounded Engineering is built and maintained with AI agents in real engineering workflows.

Use it

  1. Read the relevant card in practices/.
  2. Follow its evidence links and review its applicability and boundaries.
  3. Translate only the needed guidance into your repository's canonical policy and agent adapters.
  4. Put deterministic requirements in hooks, CI, permissions, or approval workflows rather than relying on prose alone.
  5. Run the local validator before proposing a change.
npm install
npm test

The validator is local and deterministic. It does not fetch sources or make repository changes.

Source drift checking is an opt-in, report-only operation:

npm run check:sources

The command uses the GitHub API to compare commit-pinned source blobs with the current repository heads. It refuses to run unless the catalog validates, and does not run as part of the default validator or test suite. Drift affecting a validated card is reported with a reminder to move it to needs_review and re-audit it; the command makes no repository changes.

Adopt a profile

Every adoption flow is reviewable and local-first. preview and create are read-only with respect to canonical policy; create stores a proposal under .grounded-engineering/proposals/<proposal-id>/.

Run it straight from npm, no clone required:

# one-off, nothing installed
npx grounded-engineering adopt preview --profile ai-assisted --adapter claude

# or install the command
npm install -g grounded-engineering
grounded-engineering adopt preview --profile baseline

Profiles:

  • baseline: the historical eight-card Context & Instructions starter pack. Its pack metadata stays pinned at v0.2.0 for compatibility with existing adopters.
  • ai-assisted: all thirteen current cards, adding the five Agent & Skill Design practices on top of the baseline set.

Adapter choices:

  • neutral (default): writes provider-neutral Markdown to GROUNDED_ENGINEERING.md.
  • codex: writes into a card-keyed managed block in AGENTS.md.
  • claude: writes into a card-keyed managed block in the repository-root CLAUDE.md.

v0.4.0 supports one adapter target per repository manifest. Applying a second adapter remains deferred until update propose exists.

The full preview → create → review → apply → check flow, using the installed command:

grounded-engineering adopt preview --profile ai-assisted --adapter claude
grounded-engineering adopt create --profile ai-assisted --adapter claude
# Review proposal.yaml, plan.md, and diff.patch; complete local_decisions.
grounded-engineering adopt apply <proposal-id> --confirm
grounded-engineering check

Provider adapters

By default adopt emits provider-neutral Markdown to GROUNDED_ENGINEERING.md. Pass --adapter codex to target AGENTS.md, or --adapter claude to target the repository-root CLAUDE.md:

grounded-engineering adopt preview --profile baseline --adapter codex
grounded-engineering adopt preview --profile ai-assisted --adapter claude

Both provider adapters write only inside card-keyed managed blocks and preserve bytes outside those boundaries. The Codex adapter creates or updates AGENTS.md. The Claude adapter creates or updates only the root CLAUDE.md; if .claude/CLAUDE.md, nested CLAUDE.md, or CLAUDE.local.md are present, the tool reports them during preflight but does not edit them in v0.4.0. If a Codex override file governs instruction resolution, the tool reports it and generates nothing rather than writing to a file Codex will not read.

Apply writes the selected target and .grounded-engineering/manifest.yaml only after checking the saved preconditions and explicit local decisions. Existing policy and instruction files are not silently overwritten; managed updates are bounded by card-keyed markers.

grounded-engineering check is read-only. It reads the manifest and target files from the current repository, but loads the selected pack and practice cards from the pack bundled with the installed CLI. Exit code 0 means clean, 1 means drift or repository-state mismatch, and 2 means invocation error. update propose is still reserved in v0.4.0, and custom --cards selection remains preview-only.

The current release's compatibility contract is exercised by the repository's tests and validator and is summarized in the release notes.

Evidence posture

The default public evidence form is a paraphrase with an immutable source link and a locator such as a section heading, file path, or pinned commit. Verbatim excerpts are exceptional and require an explicit redistribution decision. Large vendor prompt files and instruction documents are not copied into this repository.

Every practice identifies whether it is observed, recommended, or locally validated. Validation notes are generalized and do not disclose private repository names, paths, tracker IDs, or other internal details.

Contributing

Start with CONTRIBUTING.md and research/README.md. Contributions should improve the evidence, the practice, or the translation boundary without collapsing those layers together.

License

Original repository content is available under the MIT License. Third-party sources remain subject to their own licenses and terms; links, attribution, and redistribution decisions are recorded with the evidence.