From d424fd92fb7c9eec46b0be390db9527b4ab23492 Mon Sep 17 00:00:00 2001 From: aaron Date: Fri, 21 Aug 2026 09:36:14 -0400 Subject: [PATCH] docs(readme): recommend opencode go + extra usage with GLM-5.3 / Kimi K3; fix stale facts MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Model story (the recommendation, per team strategy): the recommended setup is an opencode go subscription — extra usage for long campaigns — running GLM-5.3 or Kimi K3, the two models the loop is tuned against. Model-agnostic framing intact (bring your own provider/key). Subtitle updated to match; Muse Spark dropped from the recommendation. Stale facts found by parsing the README against the repo: - public-skill count 37 → 38 (verified against the directory) - skills table listed the internal-only fleet skill (never shipped in the vsix) and omitted six shipped ones (amico-slack, autoresearch, problem-types, grill-with-docs, improve-codebase-architecture, teach) — table now matches packages/extension/skills/ exactly (38/38, machine-checked) - the Amicode-terminal sentence said 'the vendored, amicode-aware binary' — post-M4 the terminal's opencode is the managed, auto-updated canonical build, with opencode-amicode shimming the vendored fleet build Closes #490. --- README.md | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/README.md b/README.md index 96f043ef..a8abb934 100644 --- a/README.md +++ b/README.md @@ -11,7 +11,7 @@ Your vaults, your fleet, your devices, your pulses — composed by conversation. -A VS Code extension · built on [Piccolo.jl](https://github.com/harmoniqs/Piccolo.jl) · open [opencode](https://github.com/anomalyco/opencode) chat harness · telaio research harness by subscription (arriving) · model-agnostic — Kimi K3, Muse Spark, open models +A VS Code extension · built on [Piccolo.jl](https://github.com/harmoniqs/Piccolo.jl) · open [opencode](https://github.com/anomalyco/opencode) chat harness · telaio research harness by subscription (arriving) · model-agnostic — GLM-5.3 or Kimi K3 recommended @@ -21,7 +21,7 @@ Amicode is an **open autonomous research studio** that lives in your editor — Describe what you want in plain language — a gate, a state preparation, a calibration sweep — and Amicode designs the pulse, runs the solve, and shows you the result. Every run is captured, every pulse versioned for warm-start, and every session distilled into durable knowledge. The loop gets smarter as you use it. -Leveraging **Kimi K3** (Moonshot AI), **Muse Spark** (Meta), and other open models — on the open [opencode](https://github.com/anomalyco/opencode) ecosystem (vendored from `anomalyco/opencode`). Model-agnostic by design: the loop, not the model, is the product. +**Recommended setup:** an [opencode](https://github.com/anomalyco/opencode) **go** subscription — add extra usage if you're running long campaigns, because autonomous research sessions are token-hungry — running **GLM-5.3** or **Kimi K3** (Moonshot AI), the two models we tune the loop against. Model-agnostic by design: bring your own provider and key and the studio works the same — the loop, not the model, is the product. We start with **quantum control** because it is the hardest physical system to prove the loop on. If the studio works here — arbitrary Hamiltonians, hard constraints, **hardware in the loop via Strumento.jl / QICK** — it generalizes to any physical system you can model. Bring your own Hamiltonian; the loop is the same. That's **physical intelligence**: not one device or platform, but a composable way to do experimental science. Plenty will sell you a closed “superintelligence” that never touches the hardware. We ship open, down to the RFSoC. @@ -70,7 +70,7 @@ Amicode reads your **vaults** — the stack you mount under `~/.amico/vaults/` ( Your machines form one logical studio. Vault mounts sync via `armonia-sync-once` (launchd, every 15 min); the chat database stays canonical via an SSH mesh; WIP follows you between hosts with `leave`/`arrive`. No second writer ever touches the same SQLite file, no live `.git` is file-synced — the invariants are enforced, not assumed. Check it with `/fleet`; the skill is the playbook. Solo still works fully offline — the fleet simply means you never have to choose which machine holds the truth. -When the canonical is unreachable, `Amicode: Fleet — Enter Local Fallback` lets you keep working offline (local sessions, `Fleet: LOCAL FALLBACK` in the status bar) and `Rejoin` merges back when the tunnel is back. Drift is surfaced by `Amicode: Healthcheck` (`Fleet guard / settings / tunnel`) and one-command `pnpm sync` (`--check` CI twin, `--fix` also `git pull` + `fetch:opencode` + `Fleet — Repair`) — `Amicode: Repo Sync` runs it, and `Amicode: Open Amicode Terminal` opens a shell whose `opencode` is the vendored, amicode-aware binary (same `OPENCODE_CONFIG_CONTENT` as the chat) so `pnpm sync` there diagnoses the same panel you see. +When the canonical is unreachable, `Amicode: Fleet — Enter Local Fallback` lets you keep working offline (local sessions, `Fleet: LOCAL FALLBACK` in the status bar) and `Rejoin` merges back when the tunnel is back. Drift is surfaced by `Amicode: Healthcheck` (`Fleet guard / settings / tunnel`) and one-command `pnpm sync` (`--check` CI twin, `--fix` also `git pull` + `fetch:opencode` + `Fleet — Repair`) — `Amicode: Repo Sync` runs it, and `Amicode: Open Amicode Terminal` opens a shell whose `opencode` is the amicode-aware binary — the managed, auto-updated canonical build (same `OPENCODE_CONFIG_CONTENT` as the chat; `opencode-amicode` reaches the vendored fleet build) — so `pnpm sync` there diagnoses the same panel you see. ### Open system management — from pulse to device @@ -100,16 +100,16 @@ catalog shared catalog shared catalog, warm-starts ## Skills are the product -Skills are not configuration — they are the capability surface. The **37 public skills** in `packages/extension/skills/` ship in the vsix, versioned with the product. Additional skills load from your own vault mounts and from co-located Julia packages behind entitlements. +Skills are not configuration — they are the capability surface. The **38 public skills** in `packages/extension/skills/` ship in the vsix, versioned with the product. Additional skills load from your own vault mounts and from co-located Julia packages behind entitlements. | Surface | What it covers | |---|---| | Physics | `transmon`, `fluxonium`, `atoms` (Rydberg), `bosonic`, `ions` — Hamiltonians, drives, construction patterns | -| Lab + catalog + vault | `amico-lab`, `amico-catalog`, `amico-vault`, `amico-strategy`, `amico-schema-check` | -| Analysis + synthesis | `analyze`, `structural-analysis`, `hypothesis-review`, `dream-reflect` | -| System | `fleet`, `setup`, `solve`, `simulate`, `warm-start`, `constraints`, `objectives` | +| Lab + catalog + vault | `amico-lab`, `amico-catalog`, `amico-vault`, `amico-strategy`, `amico-schema-check`, `amico-slack` | +| Analysis + synthesis | `analyze`, `structural-analysis`, `hypothesis-review`, `dream-reflect`, `autoresearch` | +| System | `setup`, `solve`, `simulate`, `warm-start`, `constraints`, `objectives`, `problem-types` | | Delivery | `demo`, `pasqal`, `plot`, `compose`, `multistart` | -| Engineering | `debugging`, `tdd`, `verification`, `brainstorming`, `deliberate`, `grill-me`, `report-a-bug` | +| Engineering | `debugging`, `tdd`, `verification`, `brainstorming`, `deliberate`, `grill-me`, `grill-with-docs`, `improve-codebase-architecture`, `teach`, `report-a-bug` | The extension stages the union of the public bundle and your vault mounts at startup; mount presence is the eligibility proof.