Skip to content

What I’m working on

Purpose: Single public-facing briefing for humans and LLMs. Keep to ~1–2 screens; expand detail in wiki/projects/ and link out.

New Grok chat (blank thread): Paste wiki/Grok-Brief.md once (self-contained; no wikilinks), or paste only the capsule for the project you name. Then open with e.g. “Another idea for Crusty” / “Early childhood education app” / “Facebook campaign for Rusty Cage” / “Brainstorm Aye Robot” / “App Store copy for LunarCast.” See Grok-Brief.

Snapshot (2026-07-16)

Top priority = product building (active now → 2026-07-18): Team products and agents you're actively building — not life rhythms (coaching, band), wiki upkeep as its own line item, or shipped apps waiting on demand (LunarCast). Both Crusty waypoints concluded; executable next steps live on each project page. Verify against local repo, Cursor, and live sites before shipping.

# Line Status Next (one line)
1 Early childhood education app Active — small team, AI-first product Ship and iterate with human + non-human agent collaboration — reinvent how the team builds → projects/early-childhood-education-app
2 Crusty / Hermes — daily loops Aye Robot + digest + Darwin robot + ideas + projects/nostalgia-loop live Nostalgia: multi-user PoC (friends → separate Telegram chats) → projects/nostalgia-loop; Aye Robot quality tune → projects/crusty#aye-robot-daily-post--live-quality-tuning-2026-06
3 Operator Agent Node 1 (EV L2 POC) killed; creative lane live (Darwin / daily use-case robot video) Monitor creative lane; next physical node TBD at 2026-07-18 review → projects/operator-agent#creative-evolution--super-grok-video--use-case-robots-2026-06-07--live-2026-06-09

Detail: Each project page leads with counterpoint / risk where relevant, then phased steps (days/weeks), not aspirations.

Two waypoints on the same learning curve (revised 2026-04-18, late-day): "Crusty runs a real automated workflow end-to-end with a Telegram human-gate, reliably, day after day." Waypoint 1 is the easier, lower-stakes proof; Waypoint 2 is the advanced version of the same pattern against real money.

  • Waypoint 1 — Aye Robot comic autonomous loop (concluded 2026-05-03): Gate pattern proven end-to-end (brainstorm → Telegram approve → xurl post). Experiment stopped intentionally — idea quality drifted, growth path is manual-only, and the proof-of-concept served its purpose. Learnings (Telegram gate, xurl, Crusty cron) carry forward. → projects/ayerobot-comic, projects/aye-robot-crusty-paused-x-automation
  • Follow-on — Aye Robot Crusty vs Grok task (active mini experiment 2026-05-10): Crusty scheduled job vs grok.com scheduled task for the same comic lane — compare creative quality and delivery reliability (not a Waypoint 1 reopen; informal bake-off). Posting still xurl when automated. → projects/ayerobot-comic#mini-experiment-crusty-vs-grok-scheduled-task-2026-05-10
  • Waypoint 2 — Operator Agent / EV L2 POC (Node 1 killed 2026-05-03): EV L2 Shoreline POC did not survive economic stress-test — net returns (~$50–$100/mo per charger) insufficient for the capital, Year 1 time cost, and illiquidity vs existing equity. Capital stays in stocks. Long-term Operator Agent thesis (laundromat → RaaS → own hardware) remains intact; next node TBD at 2026-07-18 review. → projects/operator-agent, projects/ev-charging-shoreline-poc
  • Both waypoints concluded. The 90-day focus window (→ 2026-07-18) continues; direction now open — parallel execution on the product-building priorities in the snapshot above. Seat 2b (productized-agent / AaaS) lane remains on goals. → goals

Life & identity (active, not top priority):

Shipped products — demand-gated (no feature work until demand):

Context / other lines (not top-priority during focus window):

  • This repo: Personal projects wiki — raw/ + wiki/ + SCHEMA.md; https://wiki.darcymenard.com. Grok upkeep: grok-wiki-maintainer-setup. projects/operator-agent (2026-06-09): daily use-case robot video live — replicated Aye Robot comic job; latest: Moon Shots-styled mini music video (style test → approve / extend / regenerate) → projects/operator-agent#creative-evolution--super-grok-video--use-case-robots-2026-06-07--live-2026-06-09. projects/luna: CBT web app paused; may pivot sleep/apnea. projects/personal-context-ios: background — provider memory improving.
  • Learning — AI model distillation: Phase 1 done (Moonshot / Wissner-Gross, paper; Boxy generic on-device video-audio POC — see projects/ai-model-distillation). Phases 2–4 queued: LLM generalization doc → 8–12 readings + notes → narrow distilled model + case study. If I say “distillation” / “distillation project” with this wiki as context, open raw/ai-model-distillation-learning-plan.md (ai-model-distillation-learning-plan). → projects/ai-model-distillation
  • Crusty (Hermes live 2026-05-20): Primary agent on a dedicated, clean laptop with minimal permissions and no personal-account linkage. Hermes Agent replaced OpenClaw; SuperGrok subscription via browser OAuth (xai-oauth) for core Grok workloads. Live daily jobs (2026-07-16): email summary, Aye Robot, Darwin robot, idea generation, projects/nostalgia-loop (~6:30 AM era YouTube clip in Telegram; like/dislike; on-demand). Next: nostalgia multi-user PoC with a couple friends. Anthropic Claude Code stays dev-time only on that machine. 2026-05-10: rustycagetrio mail summary + 3 pm meal ideas still on grok.com (not Hermes). → projects/crusty, projects/nostalgia-loop, projects/operator-agent, projects/aye-robot-crusty-paused-x-automation, projects/openclaw-autonomy-org, recurring-issues
  • Music — Rusty Cage: See Life & identity above; full detail on projects/rusty-cage. Band Gmail summary on grok.com (2026-05-10), not Crusty.
  • Music — practice (harmonica + voice): Metronaut — live on App Store (2026-06-21); second live iOS app with projects/lunarcast → projects/metronaut. Harmonica: Low F, G, A, Bb, C; cross + slant deepening. Complements projects/rusty-cage and projects/music-production.
  • Health / diet: Keto-flex routine (detail lives upstream in USER.md). → projects/personal-health-diet (review before public publish)
  • projects/the-gathering (new 2026-04-22): Recurring small-group forum for entrepreneurial and technically-curious peers; Seattle-area, in-person preferred; host-rotation and/or a shared third-place venue; possible recorded / published format. Not an income project — a people / signal-exchange surface that sits alongside the band and Happy Body group as a real-world collaboration graph. Stage: naming + seeding (the people are not yet assembled).
  • projects/creative-space (new 2026-04-24) — scoping, Level 0 only during focus window: Recurring mental picture of a dedicated multi-purpose space (training + coaching + music tracking + agent building) with audio isolation, imagined at the scale of a converted music studio. Logged as a project page rather than letting it re-cost attention each time it resurfaces. Explicitly not a fleet-roster seat on goals — environmental infrastructure, not agent capability; sits alongside projects/the-gathering, projects/rusty-cage, projects/happy-body as a life / identity surface. Staged four-level plan: Level 0 (adapt current shared space, <$1k, one weekend — the only level active during the focus window), Level 1 (convert outbuilding, $5–30k), Level 2 (rent commercial space monthly, often the right final answer), Level 3 (build/buy full studio, explicitly deferred past 2027 and gated on fleet income actually landing). Focus-window deliverable is a friction log (empty table on the page, populate as the shared-space constraint actually blocks something) that gives the 2026-07-18 review real data rather than a re-run of the same open feeling. Escalation triggers for Level 2 (≥ 6 friction-log entries, or a concrete Happy Body coaching engagement whose revenue covers rent, or a Gathering chemistry signal) documented on the page. Explicitly not pulled into the top-priority briefing — same pattern as projects/the-gathering and projects/buddyline.
  • projects/buddyline — background (2026-06-07): Meela.ai already does the imagined product (phone-first senior AI companion, family signup, scheduled calls). Not building — page kept as thesis reference.

Direction: Solo entrepreneur / producer (music, Happy Body coaching); Seattle-area network with shared-purpose friends (values aligned with e.g. Musk; avoid politics). Mission (2026-06-14, see goals): experiment with and build truth-seeking and useful AI — mechanisms, agents, and products that help humans solve problems — the scoped agent fleet (Crusty first, successors via projects/openclaw-autonomy-org) is one implementation layer: shared wiki context, draft → human-gate → execute → verify on every state-changing action. Long-term thesis (2026-04-20, see goals): the agent era rewards context depth + trusted access, not UI polish — Crusty + this wiki + the Operator Agent experiment are all explicitly chosen to build that stack. 2026-04-18 (revised late-day): the work during the 90-day focus window is a single learning curve with two waypoints — Crusty runs a real automated workflow end-to-end with a Telegram human-gate. Waypoint 1 is projects/ayerobot-comic autonomous loop (quicker win, lower stakes — core image + X API 2026-04-21; morning image → Telegram 2026-04-27; Aye Robot gate unlocked 2026-04-27 with live xurl post after Telegram approval; next concrete step = mentions listen path (see projects/ayerobot-comic); ~1 week to harden the post path daily). Waypoint 2 is projects/operator-agent (the advanced version against real payouts + landlord contracts + driver support, Node 1 = EV L2 Shoreline POC per projects/ev-charging-shoreline-poc, long-term ladder through scale EV → laundromat → RaaS → own hardware). Day job stays stable; other projects paused or kept as identity-only during the 90-day focus window (→ 2026-07-18).

How to use this page (for an LLM)

  1. Grok / external LLM with no repo access: point them at Grok-Brief (paste or public URL when hosted).
  2. Read this file first, then goals (long-term thesis), index, glossary, recurring-issues, tools-and-repos, personal-operating-system (constraints → delegation).
  3. Crusty / Hermes work: projects/crusty (canonical), then projects/openclaw-autonomy-org.
  4. Other projects: projects/early-childhood-education-app, projects/ayerobot-comic, projects/lunarcast, projects/rusty-cage, projects/happy-body, projects/triviabalance, projects/ai-model-distillation, etc.
  5. Raw sources are local-only (the raw/ tree is gitignored and not published) — a public reader should rely on the wiki/ pages; these are the curated, public-safe projection of the private material.
  6. After substantive chat or ingest, update this file, log, Grok-Brief (if triggers/capsules change), and index per SCHEMA.md (repo root).

Tools & stack (quick)

  • Cursor for wiki + agents; Markdown / optional Obsidian graph; MkDocs + Vercel for https://wiki.darcymenard.com (canonical public render). Human mirror: darcymenard.com/projects (htdocs workspace) — refresh whenever this page changes materially.
  • Main Mac (local LLM): LM Studio with Google Gemma 4 (instruction-tuned MLX build, 4-bit quant) on Apple Silicon (M2 Pro, 16 GB unified memory — 2023 MacBook Pro class) for sensitive or local-only text Darcy wants off typical cloud chat/API paths — orthogonal to projects/crusty (Grok / Hermes); no shared credentials. Local inference ≠ blanket privacy (app/OS telemetry and paste discipline still matter).
  • Crusty: Hermes (in migration from OpenClaw), SuperGrok OAuth + legacy xurl / Telegram for Aye Robot surfaces (see projects/crusty).
  • Band: Manual booking; gog on rustycagetrio Gmail; grok.com for daily mail summary + meal ideas.

Wins / momentum

  • Nostalgia loop (2026-07-16): Morning 70s–mid-80s clip via Telegram (~5–7 min, in-app watch); ~20 personal sends, almost all liked. Next: friends multi-user PoC. → projects/nostalgia-loop
  • Hermes (2026-06-06 / 07-16): Gmail digest + Aye Robot + Darwin robot + idea gen + nostalgia on the clean, low-trust-surface Crusty machine; rustycagetrio mail summary + meals on grok.com (2026-05-10).
  • Metronaut (2026-06-21): Live on App Store — free practice metronome; second live iOS app (with LunarCast). → projects/metronaut
  • LunarCast (2026-05-21): Subscription caching + output device switching shipped — two-user household, daily-driver polish. → projects/lunarcast
  • Claude Code on Crusty (installed 2026-04-26): Debug-time tool now running on Crusty via Anthropic Console API key (Haiku 4.5 default, $10/mo spend cap, Keychain-only credential). Stops the "debug Crusty from main Mac" friction without breaking the dedicated-laptop posture. Runtime autonomous loop stays 100% Grok per the locked-in invariant. Now available to drive the Crusty model-routing decision in-place. 30-day spend revisit scheduled 2026-05-26. → projects/crusty (Claude Code on Crusty)
  • @ayerobotcomic: Crusty stack unblocked (2026-04-21) + morning image → Telegram (2026-04-27) + Aye Robot gate unlocked (same day): brainstorm → pick → Telegram → approve → xurl post to X verified end-to-end. Next: projects/ayerobot-comic#next-concrete-step--mentions-listen-path + post path reliability (~1 week). → projects/crusty, projects/aye-robot-crusty-paused-x-automation, projects/ayerobot-comic
  • LunarCast: Search fix through App Store review quickly (2026-04-16). → projects/lunarcast
  • Wiki bootstrapped; ingests from raw/projects.md + Crusty + LunarCast summaries.

Pain points / open loops

Product building:

Life & identity (tracked, not top priority):

Identity

  • Darcy Menard — see index and project pages for specifics.