Ryan T’s shelf
“I think Obsidian is practical as an agent first semantic layer. It’s always framed as a “second brain” for us humans, but it’s actually really good for your fleet”
Their post on X ↗
Cursor Agent
Local-only builder. Runs cursor-agent CLI on the user's laptop for experiments and a shop-floor CLI. Cloud agents stay the default for GitHub PRs. No deploy, submit, or send without the user's yes.
Can act for youGitHub0 bookmarks on XShared 29 Aug 2026
Dink Bot
Turns pickleball doubles score sheets (photos, PDFs, or typed results) into a CSV for DUPR's club match importer, and keeps stats from the matches you log. You do the upload; it never logs into DUPR.
Not specified0 bookmarks on XShared 24 Oct 2026
Eval Keeper
Keeps every skill in your bot fleet covered by evals, blocks skill edits that break safety, and flags skills that stop helping after a model change. For anyone running several AI agents on shared skills.
Can act for you0 bookmarks on XShared 14 Oct 2026
Groklexa
Alexa voice bridge for Grok Bot. Connects a private, unpublished Echo skill to this bot through a self-written Cloudflare Worker relay, with optional Voice Monkey announcements and Ring/Govee home hooks.
Not specified0 bookmarks on XShared 25 Oct 2026
Knowledge Graph Keeper
Keeps one shared knowledge graph that all your bots read, builds a one-way Obsidian vault from it, and flags notes that drift from their source code or files. For people running several bots that need the same accurate…
Not specifiedObsidian0 bookmarks on XShared 26 Oct 2026
LinkedIn Agent
A LinkedIn ghostwriter and networking assistant. It finds sourced topics in your lane, drafts posts and outreach in your voice, and sends nothing until you approve the exact text.
Drafts for your approvalLinkedIn0 bookmarks on XShared 7 Oct 2026
Plaud Notes
Intake, classify, index, and route voice-recorder notes. Uses a spreadsheet as the only index, a 16-section context pack, and a sensitivity gate for personnel, HR, and health. Does not invent a recording list.
Not specified0 bookmarks on XShared 7 Oct 2026
Research PM
Senior PM for a recurring research-intelligence pipeline. It reviews nightly claim runs, drives a weekly brief cycle, and hands coding fixes to coding agents, while you keep every publish, send, deploy and merge.
Can act for you0 bookmarks on XShared 7 Oct 2026
Research Project PM
A project manager for a research collaboration. It briefs coding agents, checks their proof itself, keeps the collaborator hub and review repo current, and stays quiet unless something real happens.
Not specified0 bookmarks on XShared 7 Oct 2026
Ship Evidence Auditor
Independently proves that each auto-ship from your coding agents met its gate, using evidence instead of agent summaries, and pauses a lane when it didn't. For teams letting AI agents merge and deploy.
Can act for you0 bookmarks on XShared 12 Oct 2026
Ship-State Tracker
Tracks every code change from proposed to deployed in an append-only, evidence-backed log, so nothing gets lost or quietly skips a step. For teams running coding agents.
Not specified0 bookmarks on XShared 22 Oct 2026
Supply-Chain Guard
Read-only supply-chain guard for teams with coding agents. Freezes installs to lockfile versions, pins checksums, waits a week before trusting brand-new releases, and scans machines for stealer tricks (.pth hijacks,…
Drafts for your approval0 bookmarks on XShared 26 Oct 2026
Test Cop
Reviews pull requests for quietly weakened tests (skips, loosened asserts, narrowed test config, softened CI) and returns a CLEAN, SUSPECT, or WEAKENED verdict with file:line proof.
Can act for you0 bookmarks on XShared 16 Oct 2026
Website Publisher
Senior web publisher for a public static site: gated deploys, SEO and indexing, docs, signup embeds, and featured-content drops — without inventing editorial.
Can act for you0 bookmarks on XShared 2 Oct 2026
Made this? Let people know
Badge code for a website
<a href="https://botshelf.dev/shelf/ryanthawks/"><img src="https://botshelf.dev/badge.svg" alt="On Bot Shelf" width="148" height="28"></a>