Posts tagged: projects Archive

clocks

2026-02-25T05:53:01Z

One of the best things I’ve vibe-coded

Nexus

2026-02-05T05:31:53Z

Nexus is a task tracker and second-brain for collaborating with AI agents on software development. It keeps project boards, task details, subtasks, daily notes, and Git worktree context together in one system.

Why Nexus #

I built Nexus because I work with AI agents across multiple parallel work streams and need a way to:

Track what agents are working on and have them check-in regularly
Capture lightweight memories without forcing everything to become executable work immediately
Keep task-to-branch context when starting work
Use daily notes inside the same system

Architecture #

Human
  |
  +-- Agent Harness (GitHub Copilot CLI)
  |     +-- Agents (orchestrator + specialists)
  |     +-- Platform hooks / commands
  |
  +-- nexus CLI (PEP 723 single-file script)
  |     |
  |     +-- Nexus API (FastAPI, port 8090)
  |           +-- Nexus DB (Postgres 16 + pgvector)
  |
  +-- Go TUI (primary interface)
  +-- Git (worktrees, branches)

Components #

Core Stack #

Nexus API — FastAPI REST backend. Projects, tasks, tags, dependencies, comments, and global discoveries. Tasks flow through backlog → wip → done with first-class readiness (is_ready) and blocking (is_blocked) markers.
Nexus DB — Postgres 16 with pgvector for vector storage (future AI memory features).
nexus CLI — PEP 723 single-file Python script. All agent-to-Nexus communication goes through this. JSON output for agents, human-readable for humans.
Go TUI — Terminal UI, the primary interface for dayFalse-to-day task management.

Agent Integration #

Copilot CLI plugin (copilot/) — GitHub Copilot integration with custom skills:
- nexus — Core task and lifecycle commands
- nexus-specialist — Task-specific expertise
- nexus-delegation — Delegation patterns
- daily-summary — Daily progress summaries
Hooks — Session lifecycle hooks that sync state between Nexus and agent sessions:
- session-start.sh — Initialize worktree metadata
- session-end.sh — Archive session state
- session-sync.sh — Periodic checkpoint
- Liveness probes for agent monitoring

Task Lifecycle #

start → implement → review → merge → done → finish

nexus start <id> — Creates nx-<id> worktree from base branch
nexus review <id> --verdict PASS — Authoritative merge gate
nexus merge <id> — Rebases and merges --no-ff
nexus done <id> — Marks task complete
nexus finish <id> — Removes worktree and deletes branch

Discoveries (Second Brain) #

Discoveries are durable memory captures that may be global, project-linked, or task-linked. They’re not executable work — they don’t get branches, worktrees, or the full task lifecycle. When ready, they can be promoted into tasks.

nexus discovery inbox — Triage view for new/promotable/stale discoveries
nexus discovery list — Browse all discoveries
nexus remember "..." — Lightweight capture
nexus discovery promote <id> — Convert to executable task

Key Features #

Task tracking — Projects, tasks, subtasks, tags, dependencies, comments
Readiness tracking — Explicit is_ready marker for backlog refinement
First-class blocking — is_blocked + blocked_reason with DB-enforced invariants
Discoveries — Lightweight memory that doesn’t force work creation
Typed memory links — Graph-style relationships between discoveries, tasks, projects, and Copilot sessions
Worktree management — Automatic branch/worktree creation and cleanup
Session persistence — Agents can persist state to DB for cross-session continuity
Archive + restore — Hygiene mechanism for discoveries (archived items stay durable but hidden from active views)
Multi-timezone — Daily notes default to America/Chicago, configurable via NEXUS_TIMEZONE

Scope #

Nexus is scoped to one machine per deployment. No remote centralization planned. The client runs per-machine against a local Postgres instance.

Rationale: If you need cross-team collaboration and centralized tracking, existing solutions (GitHub Issues, Jira, Azure DevOps + MCP servers) already solve this well. Nexus focuses on the individual developer working with AI agents on a single workstation.

Future Wants #

Daemon agent — Watch task tracker for new issues, spawn agent sessions in appropriate worktrees
Daily reporting — Ask an agent what all the other agents did today using Nexus as source of truth
EOY goal tracking — Agent summarization by Quarter or smart goal category
EKS deployment — Deploy API + Postgres to EKS to run more parallel agent stacks (work use case)
A canvas-style front-end that allows me to better visualize and plan out task-dependencies

Cannalyze

2025-12-27T06:44:31Z

A self-hostable habit tracker with special attention given to gardening habits

QR Harbor

2025-12-26T05:13:22Z

The QR side of Digital Harbor

Features #

self-service QR code generator
- frames
- colors
- brandingicons
dynamic link resolution
analytics
simple API
- see hovercode
- QR Code Monkey

Infra #

dynamic qr codes
1. tunnel
2. UI for 302s
3. UI for QRs

Mindful

2025-12-26T05:10:27Z

Reminder and alarm app

Features #

quick templates
works everywhere (mobile, desktop)

TODOs #

premium sign-up
test password reset

Notifiq

2025-12-26T05:08:54Z

Note: this began in Event Details Updates via Notifications as a Service and is the first iteration of this idea

Projects #

Docs #

Consider the Pushover Docs for inspiration on notifiq docs as well

Notes #

Use Cases #

Event Details and updates
IT ticketing system
Volunteer reminder system

Requirements #

apprise server/library
CMS with users and roles + a database

updates need to POST to apprise

notification backbone

SMS
email (requires domain to offer a real service)
- maybe managers could bring their own email server?
- no, just make it configurable - people setup the whole thing or they pay me

infra

Temporal instead of webserver patterns may lead to easier scaling

Drafting #

Been working in Windsurf with gpt-5 mostly on outlining the requirements for soonish
My immediate goal is to have very detailed specs for the project to best utilize AI while building it without allowing the AI to drive any meaningful aspect of the development

Ideas #

WorkOS for auth for B2B for soonish
remind.notifiq.net/
as a usecase for soonish
- configured reading times, so just get reminded at the next one
- phase 2 of this idea might be grabbing the content of the url and sending the notification with some of that content

Message Priority #

Events should have a priority of some kind so that business hours can be considered easily
Should channels have similar types of preferences? I suppose I need to think about the different user flows
- automated event makes critical ticket outside business hours - force update subscribers based on…. whatever they’re subscribed with/desired channels
- event update is critical - all subscribers need to know on their desired channels
- those above 2 are the same
- event update is not critical - do the signal to everyone but channels take into consideration that it’s info only, so respect a business hours/quiet hours and notify during appropriate time.

Frontends #

Use NiceGUI for a default events front end as a toy example to demonstrate notifiq use cases
Use react for something like pager duty maybe - just to see what kind of things windsurf can do with it

Competitors #

zenduty

Progress #

November 22 2025 #

added auto-sub of creator to event (visibility)
- auto subscription_selector made for autosub:default
- subscription_selector for event.tag == channel.tag as well

Soonish

2025-09-13T06:01:24Z

Personal events tracker dashboard

Features #

share events with other users
grid/calendar view
create/edit events
notified about any updated to events you’re invited to send updates to anyone you’ve invited to an event

TODOs #

premium sign-up
test password reset

Compose Stack Example Repo x Package Manager

2025-09-02T17:10:34Z

The idea is something like the AUR, which is ultimately a pointer to a repo with a script that does some setup…

Perhaps my Homelab As A Service idea could have its own package manager or repository of one-click type of applications. This could end up being basically any amount of compose files or something with a simple front-end, which is kind of what portainer does though I’m not fluent in its capabilities.

Event Details Updates via Notifications as a Service

2025-09-02T17:09:34Z

I’ve moved these notes to Soonish

Homelab As A Service

2025-09-01T07:09:49Z

$HOME #

I don’t know what to call my opinionated homelab distribution yet but I will try to have some goals for today or tomorrow for simple steps

drawing up all the pieces will take a while but is necessary
ansible playbook to make zfs datasets for vms or containers is easy first utility to bank somewhere

Homelab #

started chatting with jipity about some automation… I have so many ideas that circle my head and I have a hard time implementing anything for fear of committing halfway to something…
I think I can take baby steps to mitigate that issue and one way is to automate the first thing that annoys me - zfs dataset creation for new apps. I assume that I want a new zfs dataset if I’m going to provision something into my homelab so that means
- dataset creation on the appropriate node
- permissions
- updates to zfs-ops (eventually dataops in homelab-mono) for the backup to be added explicitly (or removed if I’m removing one - for example some old VM datasets I no longer need)
I also mind-dumped after starting the convo with jipity, using speakr to get a short summary of a brain dump is pretty awesome especially considering the purely local self-hosted nature. Here’s the summary of a pretty concentrated 6 minute monologue on things I’m thinking about:
- Summary is imperfect - but again, for self-hosted hardware and getting purely private LLM-assisted brainstorming is really cool. However, I do plan on sinking some thought into the big boys (claude or jippity) as I very slowly begin construction

speakr self-hosted summary

Key Issues Discussed

Platform and service idea for consulting
Product as an opinionated home lab targeting small teams or on-prem setups with Docker Swarm + ZFS
Focus on privacy, data ownership, compute ownership while providing cloud options (AWS)
Hardware recommendations part of services
Use AWS resources when possible but self-hosting is also viable option
Tail scale for business networking and network ACLs provided by the vendor

Key Decisions Made

Deploy servers using Docker Swarm + ZFS with an opinionated approach to data management, backups, application deployment monitoring.
Cloud options include Terraform set up in AWS (and possibly Azure later)
Use U-Corps as a base for containers on top of Ubuntu distro box; consider KeyCloak or Aphalia for authentication
DNS and public services will be handled with NameCheap and Cloudflare using Terraform, not picking each other’s features.
CERT management is Let’s Encrypt
Observability through Harbor telemetry data gathering

I brainstormed a little more witih jipity this morning and got this set of pillars

pillars

Infrastructure (hardware, OS, baseline networking)

DataOps (storage, backup, placement)

Orchestration (containers, jobs, IaC)

Networking & Connectivity (service-to-service, ingress/egress)

Security & Identity (auth, secrets, policies, hardening)

Observability (metrics, logging, tracing, health)

Automation & CI/CD (playbooks, pipelines, GitOps)

Resilience & Recovery (HA, redundancy, DR, self-healing)

Experience (DX, onboarding, docs, templates)

Features #

I’m taking notes in affine here to capture requirements grouped by feature or relevant level of the stack (ie. OS package installs vs repository setup vs pipeline recommendations etc.)

Tools #

komodo for server deployment etc… it actaully has a ton of things in it for managing compose stacks on multiple nodes right now, with swarm support coming
- in the mean time this comment gives a short example on using swarm’s overlay network and managing the stack from within komodo (kind of)

Shotput

2025-07-16T06:18:00Z

This is from I built a simple app for adding images to my blog

Overview #

Shotput is a simple webapp backed by a git repo that allows you to paste a screenshot from your clipboard, it gets put into the repo and pushed, and you get a link back to using that image via statically’s CDN in your website.

Example #

pasting an image let’s you copy the markdown link for easy pasting

![20250607130609_dadd33eb.png](https://cdn.statically.io/gh/pypeaday/images.pype.dev/main/blog-media/20250607130609_dadd33eb.png)

Concerns #

Currently this is deployed with an ssh key mounted into the container that I put into a zfs dataset manually. Per 2025-07-16 Notes I’m struggling with the key sharing model…

Pilgrims Creek

2025-07-09T08:00:25Z

What

Podcast with Cameron for weary theologians. We spend time discussing interesting topics, approaches to theology, etc. all from the perspective of travelers who are tired and need a place to just have good conversation. We’re not here to fight theological points or debate, but rather to explore the patterns in the Bible, equip ourselves and others with theological tools, and overall enjoy a good time with believers interacting with fun and challenging ideas about our faith

Thoughts To Nostr

2025-07-08T08:30:35Z

I am trying to use temporal at home for running automated pipelines. It’s definitely overkill for my use case but this is America…

So my first project is a system that takes my Thoughts and posts them to Nostr as I briefly discussed in Testing a Postiz Change Locally (IT WORKS!).

I have temporal infrastructure setup in homelab-compose and then homelab-temporal should probably be homelab-temporal-pipelines or something, but it’s the app code. I also have a repo homelab-social-media-pipelines - which has been consolidated down into homelab-temporal. once that all works I can destroy the homelab-social-media-pipelines repo

NOTE: everything is moving to homelab-mono

Issue 01 #

I am currently hitting a stupid python issue:

nic in homelab-temporal   main    ×2  ×4  ×7 via   v3.12.8(homelab-temporal)   (dev) 󰒄
⬢ [devbox] ❯ python -c "from homelab_temporal.postiz.postiz_client import PostizClient"
(homelab-temporal)
nic in homelab-temporal   main    ×2  ×4  ×7 via   v3.12.8(homelab-temporal)   (dev) 󰒄
⬢ [devbox] ❯ python scripts/postiz/simple_post.py
Traceback (most recent call last):
  File "/home/nic/projects/personal/homelab-temporal/scripts/postiz/simple_post.py", line 3, in <module>
    from homelab_temporal.postiz.postiz_client import PostizClient, process_for_nostr, process_for_x
ModuleNotFoundError: No module named 'homelab_temporal'

very dumb venv problem…

Resolution #

need to use python -m because the script is outside the project root? see 2025-07-08 Notes and the chatgpt notes for more on this but it works now

Thoughts Order #

as of 2025-07-11 Notes I have the thoughts ordering correct - we post the oldest thoughts up to the newest ones. Now I want more intros for the notes so they don’t get too repetitive

TODOs #

[x] - destroy the homelab-social-media-pipelines repo

[] - think about how to structure things more… monorepo in homelab-temporal might be hard to manage with AI tools due to context management… unsure though

[] - create workflow and activities for thoughts to nostr… revisit diagram to see what you need to build out in the temporal repo

[] - do something like this for github stars and then blog posts

I need to update my templates for nostr og images using this as fodder

<!-- HTML Meta Tags -->
<title>Nostr Design</title>
<meta
  name="description"
  content="A comprehensive resource for designers and developers to build successful nostr products"
/>

<!-- Facebook Meta Tags -->
<meta property="og:url" content="https://nostrdesign.org" />
<meta property="og:type" content="website" />
<meta property="og:title" content="Nostr Design" />
<meta
  property="og:description"
  content="A comprehensive resource for designers and developers to build successful nostr products"
/>
<meta
  property="og:image"
  content="https://nostrdesign.org/img/nostr-cover.jpg"
/>

<!-- Twitter Meta Tags -->
<meta name="twitter:card" content="summary_large_image" />
<meta property="twitter:domain" content="nostrdesign.org" />
<meta property="twitter:url" content="https://nostrdesign.org" />
<meta name="twitter:title" content="Nostr Design" />
<meta
  name="twitter:description"
  content="A comprehensive resource for designers and developers to build successful nostr products"
/>
<meta
  name="twitter:image"
  content="https://nostrdesign.org/img/nostr-cover.jpg"
/>

Dockerizing #

I think I wnt to dockerize my workers and workflows… build them out of the homelab-temporal repo but then run them in compose stacks… that can get stitched together better but for now that’ll probably be a good target. I have things to think about though… for example the blog builder needs my ssh keys, or an ssh key for a user… also git is a little mess up anyways

temporal link

Temporal #

I haev temporal workflows and stuff that all seem to work

on 2025-08-08 Notes I had issues with my nostr relay being too public, so I shut it down, reconfigured postiz and stuff and got it workign again, more restricted and in-memory database

gwl-margarita-tracker

2025-06-30T21:33:28Z

Track discount on margarita refills over the lifetime of this mug

github repo

QuadTask

2025-06-13T20:58:00Z

QuadTask is a simple productivity app that I’m sure will bring in $0 towards my retirement but a few friends use it and it’s a fun project to use as fodder for learning how to pair program with AI.

~~The app is live here - quadtask.app~~

The app is live here - quadtask.fly.dev

Updates #

I can add SMS notifications with the SMTP trick
retire quadtask.app and just have it live on fly

New Updatess #

Rebuild as a notifiq app
- includes translating quadtask’s schema into event.content or adding tables to notifiq
- db clean up to get rid of my DDOS artifacts

Dad, Can I Wear This?

2025-06-13T20:58:00Z

Repo

This is a simple FastAPI web app you can docker compose up alongside ollama and get some cute AI driven weather suggestions for the day.

Version 2 #

Cronicle

2025-06-13T20:58:00Z

I use hcio to monitor a few of my backup scripts and when LLMs hit the scene I got interested to try to build my own. I have a pretty good understanding of systems, and I enjoy thinking through them, but I’m not a great front-end designer - AI has helped a bit in that regard.

Cronicle isn’t live anywhere yet, and it’s undergone a few revisions already. I’m not sure if I’ll continue with it, but I’m having fun thinking through it.

Someday

2025-02-17T00:00:00Z

Purpose of the Page #

Projects, tools, and technologies I want to explore someday. This is my queue of things to check out, organized by category. Each item has a checkbox for tracking progress - once explored, I’ll add a summary or link to a longer post.

Self-Hosted Services #

Karakeep (2025-07-19) Read-later and bookmarking service to replace texting links to myself. Want to explore the app and extension, set up RSS feeds for lists.
- I am running Karakeep, would like to hook it into my blog like thoughts but just haven’t taken the time to do it
Rybbit (2025-08-23) Self-hosted Google Analytics alternative. Forked the repo, want to try on quadtask.
Domain Locker (2025-12-09) Domain monitoring service. Stood it up and want to see how it fits into the lab over the next year.

Platform & DevOps #

Komodo (2025-08-23) Platform-as-a-product tool that aligns with my vision for “platform as a product”. Discovered it matches much of what I envisioned.
IronCalc (2025-11-07) Self-hosted spreadsheet solution. Need to manage the build, starting to think about a reusable Forgejo action for Docker builds.
- Unfortunately there’s not much to write about - neat project but I don’t need it
LLDAP (2025-11-10) Lightweight LDAP for homelab SSO. Want to get SSO setup at home for repoflow, considering LLDAP and Authelia for authentication.
Repoflow (2025-08-09) Artifactory replacement for artifact storage. Requested a free license and experimenting with it for homelab platform storage.
- Note: Have blog posts but no dedicated project/glossary page yet

Storage Solutions #

Garage (2026-02-15) Distributed object storage. MinIO is archived, so looking at Garage as a replacement for object storage.
RustFS (2026-02-15) MinIO alternative in Rust. Alternative to MinIO being considered alongside Garage.

Mobile & Apps #

Conduit (2025-08-23) Open-WebUI Android app. Found it but needs to build and sideload the APK.

Desktop & Window Management #

AeroSpace (2025-11-26) Tiling window manager for Mac. Want to try it out for window management.
- I am using AeroSpace on my Mac - it’s beautiful
Fluid Tile (2025-11-26) Tiling window manager option. Saw it as a tiling window manager option to try after getting tiling working on Mac with AeroSpace.

Productivity & Utilities #

ChoresRewards (2025-08-10) Chore tracking with rewards for kids. Looking for a way to track and reward chores for both myself and the kids; considering rewriting it in FastAPI if the auth mechanisms don’t work out.
Lazyjournal (2025-11-15) TUI for host logs. Lazydocker/git-inspired tool for viewing host logs that I learned about.
Oh My Opencode (2025-12-31) Compare with OpenAgents. Need to compare OpenAgents with oh-my-opencode.
Handy (2026-02-03) STT tool for voice control. Started using for speech-to-text on work Mac to talk to neovim and opencode; excited to explore vocally conversing with agents.

Already Explored (Have Pages) #

These projects already have glossary entries or blog posts:

Affine - Self-hosted Notion-like platform (see Affine)
Apprise - Centralized notification service (see apprise)
Forgejo - Self-hosted Git forge (see forgejo)