@fungi.computer/oyster 0.13.1 → 0.14.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,397 +1,152 @@
1
1
  ---
2
2
  title: Agent quick start
3
3
  description:
4
- Hand this guide to a coding agent to install Oyster from npm, run it locally,
5
- install an App, connect over MCP and deploy.
4
+ Run Oyster locally with one command, open Hello, then edit and restart.
5
+ Connect MCP, build another App or deploy when you are ready.
6
6
  ---
7
7
 
8
- You are an agent setting up Oyster for a person. Follow these steps in order.
9
- Run every command exactly as written, from the directory each step names.
10
- Compare its output with the expected output before you go on. When a step fails,
11
- read the "If it fails" notes under it before trying anything else.
12
-
13
- Not sure Oyster fits the job?
14
- [Is Oyster for you?](https://docs.fungi.computer/is-oyster-for-you/) compares it
15
- with Cloudflare Workers Builds before you install anything.
16
-
17
- ## Before you start
18
-
19
- - **Cloudflare plan:** SQLite-backed Durable Objects are available on Workers
20
- Free; they do not by themselves require Paid. Cloudflare's SQLite-backed
21
- Durable Objects limits list 30 seconds of CPU per request by default,
22
- configurable to 5 minutes. See Cloudflare's
23
- [Durable Objects limits](https://developers.cloudflare.com/durable-objects/platform/limits/)
24
- and
25
- [Durable Objects pricing](https://developers.cloudflare.com/durable-objects/platform/pricing/).
26
- - **Agent chat and AI Gateway:** By default, Agent chat uses the Workers AI
27
- binding; it needs no separate provider key or Gateway setup. Prints uses your
28
- account's default AI Gateway (created automatically); set `AI_GATEWAY_ID` to
29
- use another. The routed models include `openai/gpt-image-2` and video IDs
30
- `bytedance/seedance-2.0`, `bytedance/seedance-2.0-fast`,
31
- `bytedance/seedance-2.0-mini`, `google/veo-3.1`, `google/veo-3.1-fast` and
32
- `minimax/h3-max`.
33
-
34
- - **Deployment plan:** Step 7 needs Workers Paid because this deployment uses
35
- Containers for the Computer image; Cloudflare lists Containers as available on
36
- the
37
- [Workers Paid plan](https://developers.cloudflare.com/containers/platform/).
38
- SQLite-backed Durable Objects alone do not require Paid.
39
-
40
- ## What Oyster is
41
-
42
- - Oyster hosts Apps on one Cloudflare Worker, or on celld, a Workers-compatible
43
- runtime you host yourself.
44
- - Each App is an immutable Release built from an exact git commit. Its backend
45
- runs in its own Durable Object with its own SQLite, and it gets only the
46
- bindings it declares.
47
- - People use Apps in a browser. Agents call the same App methods over HTTP or
48
- through one MCP endpoint, `/mcp`.
49
-
50
- ## 1. Check prerequisites
51
-
52
- You need Node 24 or newer, pnpm 11 or newer, Git, curl and openssl. You do not
53
- need a Cloudflare account until step 7. Docker is needed only in step 7, where
54
- Wrangler builds the Computer's container image.
8
+ You are an agent setting up Oyster for a person. Start with a local desktop and
9
+ one working App. You need **Node 24 or newer and Git**, an internet connection
10
+ for the first install, and a browser to see the result. No Cloudflare account,
11
+ Docker or payment is needed for this trial.
12
+
13
+ ## 1. Start Oyster
14
+
15
+ In the directory where the person wants to keep their project, check the tools:
55
16
 
56
17
  ```sh
57
18
  node --version
58
- pnpm --version
59
19
  git --version
60
20
  ```
61
21
 
62
- Expected output: a `v24.` or later Node version, pnpm 11 or newer, then any
63
- `git version 2.x` line.
64
-
65
- If it fails:
22
+ Expect Node `v24` or newer and a Git version. If either is missing or Node is
23
+ older, install it before continuing.
66
24
 
67
- - Wrong Node: install Node 24 with your version manager, for example
68
- `mise use node@24` or `nvm install 24`.
69
- - No pnpm, or an older version: install pnpm 11 with `npm install -g pnpm@11`.
70
-
71
- ## 2. Create a project
72
-
73
- Make a new directory, install Oyster, its Computer and Wrangler, and copy the
74
- example Worker out of the package. The local pnpm policy allows esbuild and
75
- workerd to install their native binaries and disables unneeded optional
76
- dependency scripts. Write it before installing so neither command stops for
77
- unapproved build scripts:
25
+ Keep the next command running in a persistent terminal or your agent runtime's
26
+ managed process session:
78
27
 
79
28
  ```sh
80
- mkdir my-oyster
81
- cd my-oyster
82
- cat > pnpm-workspace.yaml <<'YAML'
83
- allowBuilds:
84
- "@google/genai": false
85
- "@mongodb-js/zstd": false
86
- esbuild: true
87
- exifreader: false
88
- node-liblzma: false
89
- protobufjs: false
90
- workerd: true
91
- YAML
92
- pnpm add @fungi.computer/oyster @fungi.computer/computer react@^19.2.0 react-dom@^19.2.0
93
- pnpm add -D wrangler
94
- cp node_modules/@fungi.computer/oyster/examples/standalone/worker.ts .
95
- cp node_modules/@fungi.computer/oyster/examples/standalone/wrangler.jsonc .
29
+ npx --yes @fungi.computer/oyster try --no-open
96
30
  ```
97
31
 
98
- pnpm 11 holds back any version published in the last day, so a brand-new release
99
- may not be the one it installs. Run
100
- `pnpm ls @fungi.computer/oyster @fungi.computer/computer` to see which versions
101
- you got, and compare them with
102
- [Package versions in each Fungi release](https://docs.fungi.computer/reference/packages/#package-versions-in-each-fungi-release).
103
- To install a specific release, name its versions, as in
104
- [Upgrade Oyster](https://docs.fungi.computer/guides/run-your-own-oyster/#upgrade-oyster).
105
-
106
- Expected output: pnpm lists the added package after each `add`, and `ls` now
107
- shows `node_modules`, `package.json`, `pnpm-workspace.yaml`, `worker.ts` and
108
- `wrangler.jsonc`.
32
+ It creates `./my-oyster`, installs dependencies, starts Oyster on
33
+ `http://127.0.0.1:8787`, installs Hello and builds Tourtois, the desktop, from
34
+ its public source. `--no-open` lets you open the sign-in link yourself; omit it
35
+ on a person's computer to open their browser automatically.
109
36
 
110
- Every later command runs from this `my-oyster` directory unless the step says
111
- otherwise.
37
+ On a fresh install, expect these phases, with timing and progress alongside:
112
38
 
113
- If it fails:
114
-
115
- - `ERR_PNPM_IGNORED_BUILDS`: check that the policy file above exists before
116
- running `pnpm add`. If it names a new dependency not listed there, stop and
117
- review that package before allowing its scripts; do not treat exit 1 as
118
- success.
119
- - `ERR_PNPM_UNEXPECTED_STORE`: this directory was installed from a different
120
- pnpm store than your current default. Run the same `pnpm add` again with
121
- `--store-dir` set to the store the error says the dependencies are linked
122
- from; do not delete `node_modules` to get past it.
123
- - `cp: cannot stat …/examples/standalone/…`: the install did not finish. Run
124
- `pnpm add @fungi.computer/oyster @fungi.computer/computer react@^19.2.0 react-dom@^19.2.0`
125
- again and read its first error.
126
-
127
- ## 3. Run a local Oyster
128
-
129
- `worker.ts` and `wrangler.jsonc` are the whole Oyster. Run it under
130
- `wrangler dev` on your machine, without its Computer container. Oyster serves an
131
- HTTPS origin; `--local-upstream` presents your loopback requests to it at
132
- `https://oyster.test`. Choose an owner key and keep this terminal open:
133
-
134
- ```sh
135
- export OYSTER_OWNER_KEY=$(openssl rand -hex 24)
136
- export OYSTER_URL=http://127.0.0.1:8787
137
- echo "$OYSTER_OWNER_KEY" > .oyster-owner-key
138
- npx wrangler dev --local --enable-containers=false --ip 127.0.0.1 --port 8787 \
139
- --local-upstream oyster.test --upstream-protocol https \
140
- --var OYSTER_OWNER_KEY:$OYSTER_OWNER_KEY --var OYSTER_ORIGIN:https://oyster.test
39
+ ```text
40
+ Creating my-oyster
41
+ Installing dependencies with npm
42
+ Starting Oyster
43
+ Installing Hello and building Tourtois
44
+ Preparing desktop sign-in
45
+ Oyster is running
141
46
  ```
142
47
 
143
- Expected output: warnings that the AI and Artifacts bindings reach remote
144
- resources, a bindings table listing `env.REGISTRY (OysterRegistry)`,
145
- `env.INSTANCES (StandaloneInstance)`, `env.COMPUTER (ComputerDO)` and the rest,
146
- then `Ready on http://127.0.0.1:8787`. Locally there is no Computer, git storage
147
- or Workers AI; steps 3 to 5 need none of them.
48
+ Then Oyster prints `Open (works once):` with a local sign-in URL, the Oyster
49
+ address, and paths to the owner key and Hello source. The key is saved in
50
+ `my-oyster/.oyster-owner`; it is never printed or put in the sign-in URL. Treat
51
+ the one-time URL and the key as credentials: do not put them in reports.
148
52
 
149
- Keep it running. Open a second terminal in `my-oyster`, set the same two
150
- variables there, and check that Oyster answers:
151
-
152
- ```sh
153
- export OYSTER_OWNER_KEY=$(cat .oyster-owner-key)
154
- export OYSTER_URL=http://127.0.0.1:8787
155
- curl -s -o /dev/null -w '%{http_code}\n' $OYSTER_URL/login
156
- curl -s -o /dev/null -w '%{http_code}\n' $OYSTER_URL/api/installations
157
- curl -s -H "Authorization: Bearer $OYSTER_OWNER_KEY" $OYSTER_URL/api/installations
158
- ```
159
-
160
- Expected output: `200`, then `401` (no key), then `[]` (no Apps installed yet).
53
+ Keep this terminal running while you use the desktop. If there is no success or
54
+ failure after five minutes, inspect the logs under `my-oyster/.oyster/` and
55
+ report the last phase; do not silently leave a stalled process running.
161
56
 
162
57
  If it fails:
163
58
 
164
- - `Address already in use`: another process holds port 8787. Pick another port
165
- and use it in both `--port` and `OYSTER_URL`.
166
- - `curl` prints `000`: the server is not ready yet. Wait for the `Ready on`
167
- line.
168
- - `401` with the key: the two terminals hold different keys. Set
169
- `OYSTER_OWNER_KEY` from `.oyster-owner-key` in both.
170
- - An agent runtime that cannot keep a process in the foreground starts the same
171
- command in the background with its output in a log, then waits until the
172
- `/login` check prints `200`:
173
-
174
- ```sh
175
- npx wrangler dev --local --enable-containers=false --ip 127.0.0.1 --port 8787 \
176
- --local-upstream oyster.test --upstream-protocol https \
177
- --var OYSTER_OWNER_KEY:$OYSTER_OWNER_KEY --var OYSTER_ORIGIN:https://oyster.test \
178
- > oyster-dev.log 2>&1 &
179
- ```
180
-
181
- To stop the server, press Ctrl-C in its terminal. A background server runs as
182
- several processes (`npm exec`, `wrangler.js` and `workerd`); stop them all by
183
- ending `wrangler.js`, then check that `/login` prints `000`:
184
-
185
- ```sh
186
- pkill -f 'wrangler[.]js dev .*--port 8787'
187
- ```
188
-
189
- ## 4. Add an App
59
+ - **Port in use:** leave that listener alone. Follow the suggested command with
60
+ another port, such as `--port 8788`, and use the address Oyster prints.
61
+ - **Directory already has files:** choose a new directory with
62
+ `npx --yes @fungi.computer/oyster try another-oyster --no-open`. Do not delete
63
+ someone else's project.
64
+ - **Install or startup error:** read the diagnostic and the log path it names.
65
+ Keep the project, fix the reported cause and rerun the same command. Oyster
66
+ preserves your key and source; `--verbose` shows install and server logs.
67
+ - **Oyster needs desktop repair:** the host is available for recovery, but the
68
+ desktop is not ready. Follow the printed repair diagnostic and report it; this
69
+ is not a successful trial.
190
70
 
191
- An App is a git repository with a `package.json` that declares it under
192
- `fungi.app`. Make the smallest one, a single committed HTML page:
71
+ ## 2. Open Hello
193
72
 
194
- ```sh
195
- mkdir -p hello-app/bundle
196
- cat > hello-app/package.json <<'EOF'
197
- {
198
- "name": "hello-app",
199
- "private": true,
200
- "fungi": { "app": { "displayName": "Hello", "capabilities": [] } }
201
- }
202
- EOF
203
- cat > hello-app/bundle/index.html <<'EOF'
204
- <!doctype html>
205
- <title>Hello</title>
206
- <h1>Hello from my first Oyster App</h1>
207
- EOF
208
- git -C hello-app init -q
209
- git -C hello-app add .
210
- git -C hello-app -c user.name=Oyster -c user.email=oyster@example.invalid \
211
- commit -qm "Hello App"
212
- ```
213
-
214
- Then pack that commit into a Release and install it as installation `hello` with
215
- `@fungi.computer/oyster/pack`:
73
+ Open the printed one-time link in a browser on the same machine as Oyster. From
74
+ the Tourtois launcher, choose **Hello**. Success means the App visibly reads
75
+ **Hello from my first Oyster App**, without a loading or timeout overlay. An
76
+ HTTP 200 alone does not prove that the desktop or App opened.
216
77
 
217
- ```sh
218
- cat > install-hello.mjs <<'EOF'
219
- import { execFileSync } from "node:child_process";
220
- import { mkdtemp, rm } from "node:fs/promises";
221
- import { tmpdir } from "node:os";
222
- import { join, resolve } from "node:path";
223
- import process from "node:process";
224
- import { installRelease, packRelease } from "@fungi.computer/oyster/pack";
225
-
226
- const source = resolve("hello-app");
227
- const commit = execFileSync("git", ["-C", source, "rev-parse", "HEAD"], {
228
- encoding: "utf8",
229
- }).trim();
230
- const out = await mkdtemp(join(tmpdir(), "oyster-hello-"));
231
- try {
232
- await packRelease({
233
- appId: "app_hello",
234
- sourceDirectory: source,
235
- sourceRevision: `git:${commit}`,
236
- recipe: {
237
- kind: "bundled-html",
238
- entry: "bundle/index.html",
239
- assets: {
240
- maxManifestBytes: 65_536,
241
- maxObjects: 16,
242
- maxObjectBytes: 1_048_576,
243
- maxTotalBytes: 4_194_304,
244
- },
245
- assetBase: "./",
246
- backend: null,
247
- maxOutputBytes: 4_194_304,
248
- },
249
- outDirectory: join(out, "release"),
250
- });
251
- const response = await installRelease({
252
- directory: join(out, "release"),
253
- url: process.env.OYSTER_URL,
254
- key: process.env.OYSTER_OWNER_KEY,
255
- installationId: "hello",
256
- });
257
- console.log(`installed hello: ${response.status} ${await response.text()}`);
258
- } finally {
259
- await rm(out, { recursive: true, force: true });
260
- }
261
- EOF
262
- node install-hello.mjs
263
- curl -s -o /dev/null -w '%{http_code}\n' \
264
- -H "Authorization: Bearer $OYSTER_OWNER_KEY" $OYSTER_URL/apps/hello/
265
- ```
78
+ If the link has expired or was already used, open the printed Oyster address
79
+ with `/login` appended and paste the key from `my-oyster/.oyster-owner` into the
80
+ login form. If your agent cannot use a browser on that machine, give the person
81
+ these steps and report that the visible check is still unverified.
266
82
 
267
- Expected output: one line starting
268
- `installed hello: 201 {"installationId":"hello",` and containing
269
- `"status":"enabled"`, then `200`. Running `node install-hello.mjs` again prints
270
- `installed hello: 200`: the installation already exists, and Oyster switches it
271
- to the Release just packed.
83
+ The local trial has no Computer, hosted git storage or Workers AI. Opening Hello
84
+ does not prove a cloud deployment or a model reply.
272
85
 
273
- A person can open `$OYSTER_URL/login`, sign in with the owner key, and open
274
- Hello from the installation list at `$OYSTER_URL/`.
86
+ ## 3. Stop, restart and edit
275
87
 
276
- If it fails:
277
-
278
- - `installed hello: 401`: `OYSTER_OWNER_KEY` in this terminal differs from the
279
- server's. Set it from `.oyster-owner-key`.
280
- - `fetch failed` or `ECONNREFUSED`: the server from step 3 is not running.
281
- - The build reads committed files only. To install an edit, commit it in
282
- `hello-app` and run `node install-hello.mjs` again; Oyster switches the
283
- installation to the new Release and keeps its data.
284
-
285
- An App with a backend, typed methods and its own SQLite adds
286
- `fungi.backendDefinition` and a `backend` entry to its recipe. The
287
- [Notes example App](https://github.com/fungi-computer/oyster/tree/main/examples/notes-app)
288
- is one, and the Oyster README's "Pack an App" and "Build App Releases" sections
289
- describe the recipe.
290
-
291
- ## 5. Connect over MCP
292
-
293
- `/mcp` serves every installed App to MCP clients through one `execute` tool, in
294
- which each App's methods are `env.<APP>`. The owner key signs in. For Claude
295
- Code:
296
-
297
- ```sh
298
- claude mcp add --transport http oyster "$OYSTER_URL/mcp" \
299
- --header "Authorization: Bearer $OYSTER_OWNER_KEY"
300
- ```
301
-
302
- To check `/mcp` without an MCP client, open a session and list its tools:
303
-
304
- ```sh
305
- MCP=(-H "Authorization: Bearer $OYSTER_OWNER_KEY" -H 'content-type: application/json' -H 'accept: application/json, text/event-stream')
306
- SESSION=$(curl -s "${MCP[@]}" -D - -o /dev/null \
307
- -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-11-25","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}' \
308
- $OYSTER_URL/mcp | grep -i '^mcp-session-id' | cut -d' ' -f2 | tr -d '\r')
309
- curl -s "${MCP[@]}" -H "mcp-session-id: $SESSION" -H 'mcp-protocol-version: 2025-11-25' \
310
- -d '{"jsonrpc":"2.0","method":"notifications/initialized"}' $OYSTER_URL/mcp
311
- curl -s "${MCP[@]}" -H "mcp-session-id: $SESSION" -H 'mcp-protocol-version: 2025-11-25' \
312
- -d '{"jsonrpc":"2.0","id":2,"method":"tools/list"}' $OYSTER_URL/mcp
313
- ```
314
-
315
- Expected output: the last command prints an `event: message` block whose `data:`
316
- line contains `"name":"execute"`.
317
-
318
- If it fails:
319
-
320
- - `SESSION` is empty: the `initialize` call was refused. Check the key as in
321
- step 3.
322
- - The copied Worker already includes `POST /api/keys`. Use the owner key to
323
- create a separate key for each agent; no extra Worker is needed.
324
-
325
- ## 6. Add Git
326
-
327
- Git is already part of the host: people and agents `clone`, `fetch` and `push`
328
- App repositories with ordinary `git` at `/git/<installation>/<repo>.git`, with
329
- repositories in Cloudflare Artifacts. Artifacts runs only once the Worker is
330
- deployed in step 7. There, give each agent its own key:
88
+ Press Ctrl-C in the terminal that owns this trial to stop it. In an agent
89
+ runtime, stop your managed session and its children; do not kill processes by a
90
+ shared name or port. From the **same parent directory** as step 1, restart:
331
91
 
332
92
  ```sh
333
- curl -s -X POST -H "Authorization: Bearer $OYSTER_OWNER_KEY" \
334
- -H 'content-type: application/json' \
335
- -d '{"name":"morel","grants":[{"capability":"apps.use","targets":{"kind":"all"}}]}' \
336
- $OYSTER_URL/api/keys
93
+ npx --yes @fungi.computer/oyster try my-oyster --no-open
337
94
  ```
338
95
 
339
- Expected output: JSON with a `plaintext` field starting `oyk_`. Oyster shows it
340
- once; that key signs in to `git` and to `/mcp`.
341
-
342
- ## 7. Deploy to Cloudflare Workers
343
-
344
- This step needs a Cloudflare account on Workers Paid with Containers, Artifacts,
345
- Workers AI and Workflows enabled, and Docker running on this machine. Ask the
346
- person before you deploy anything.
347
-
348
- `wrangler.jsonc` names the Worker `oyster`, its R2 bucket `oyster-releases` and
349
- its Artifacts namespace `oyster`. The Worker's address is
350
- `https://oyster.<your-subdomain>.workers.dev`, or a custom domain:
96
+ Or, from **inside `my-oyster`**, run:
351
97
 
352
98
  ```sh
353
- npx wrangler login
354
- npx wrangler r2 bucket create oyster-releases
355
- export OYSTER_URL=https://oyster.<your-subdomain>.workers.dev
356
- npx wrangler deploy --var OYSTER_ORIGIN:$OYSTER_URL
357
- printf '%s' "$OYSTER_OWNER_KEY" | npx wrangler secret put OYSTER_OWNER_KEY
99
+ npx --yes @fungi.computer/oyster try . --no-open
358
100
  ```
359
101
 
360
- Expected output: `wrangler deploy` builds the Computer image, uploads it and
361
- prints the deployed `https://` URL, and `wrangler secret put` reports the secret
362
- uploaded.
102
+ Use the directory and `--port` you chose if they differ. Expect
103
+ `Restarting the Oyster in …`, then `Installing Hello and checking Tourtois`,
104
+ followed by a new sign-in link. Your owner key and Hello source stay in place.
363
105
 
364
- Then install Fungi's seven starter Apps, each built on this Oyster's Computer
365
- from its own repository:
106
+ To make the first edit, change the heading in
107
+ `my-oyster/hello-app/bundle/index.html`. Keep the `main.js` script: it tells
108
+ Oyster that the App is ready. Save it, then from the **parent directory** commit
109
+ the change in the `hello-app` Git repository:
366
110
 
367
111
  ```sh
368
- npx oyster install-apps
112
+ git -C my-oyster/hello-app add bundle/index.html
113
+ git -C my-oyster/hello-app -c user.name="Oyster learner" \
114
+ -c user.email="oyster@example.invalid" commit -m "Edit Hello"
369
115
  ```
370
116
 
371
- Expected output: a `source recorded` line per App, `Building … on the Computer`
372
- and `Installed …` lines (two builds at a time, Tourtois first), a timing table,
373
- and `Installed 7 Apps`. Open `$OYSTER_URL/login` with the owner key to see the
374
- Tourtois desktop.
375
-
376
- To keep `OYSTER_ORIGIN` across later deploys, set `vars.OYSTER_ORIGIN` in
377
- `wrangler.jsonc`; every deploy replaces the Worker's variables.
378
-
379
- If it fails:
380
-
381
- - `R2 bucket … already exists`: use that bucket, or pick a new name and put it
382
- in `wrangler.jsonc`.
383
- - A secret prompt on stdin: pipe the key as shown; never type it into a log.
384
- - Bindings or migrations refused: deploy the copied `wrangler.jsonc` unchanged
385
- except for `vars` and the bucket name.
386
- - `docker` not found or not running: start Docker; Wrangler builds the Computer
387
- image from `node_modules/@fungi.computer/computer/Dockerfile`.
388
- - A build `failed` in `install-apps`: run it again; finished Apps are kept and
389
- replaced in place.
390
-
391
- ## Where to go next
392
-
393
- - The Oyster README (`node_modules/@fungi.computer/oyster/README.md`) explains
394
- the host, Apps, capabilities, MCP and the shell in depth.
395
- - Oyster is built from the other Fungi packages, each its own repository under
396
- [github.com/fungi-computer](https://github.com/fungi-computer) and on npm as
397
- `@fungi.computer/<name>`. The README's "Part of Fungi" section lists them.
117
+ Expect a commit summary with one changed file. This example supplies an identity
118
+ for just this commit, so it also works on a machine without global Git settings.
119
+ Use your own name and email if you prefer. Adjust `my-oyster` if you chose
120
+ another project directory.
121
+
122
+ Stop and restart Oyster as above. It builds committed source; saving or
123
+ refreshing alone does not install an edit. Open the new sign-in link and choose
124
+ Hello again to check the changed heading.
125
+
126
+ The [build-an-App recipe](https://docs.fungi.computer/agents/build-an-app.md)
127
+ has exact edit commands and the next App to build.
128
+
129
+ ## After the local trial
130
+
131
+ Stop here unless the person wants the next step. These are separate tasks:
132
+
133
+ - **Connect MCP:** use the
134
+ [MCP guide](https://docs.fungi.computer/oyster/concepts/#connect-an-agent-over-mcp)
135
+ for the `/mcp` endpoint and authentication. For this trial, use the printed
136
+ local Oyster address and the owner key in `my-oyster/.oyster-owner`. Hello is
137
+ a static page; connecting MCP does not give it backend methods.
138
+ - **Build another App:** follow the
139
+ [build-an-App recipe](https://docs.fungi.computer/agents/build-an-app.md),
140
+ then the [App guide](https://docs.fungi.computer/apps/) for backends and typed
141
+ methods.
142
+ - **Deploy:** follow
143
+ [Run your own Oyster](https://docs.fungi.computer/guides/run-your-own-oyster/#deploy-it).
144
+ That step needs your own Cloudflare account, paid resources and Docker for the
145
+ Computer image. Do it only when the person asks for hosting.
146
+
147
+ For manual setup, the ordinary
148
+ [quickstart](https://docs.fungi.computer/oyster/quickstart/) keeps the Wrangler
149
+ commands under **What it does under the hood**. The
150
+ [Oyster concepts](https://docs.fungi.computer/oyster/concepts/) and
151
+ [API reference](https://docs.fungi.computer/oyster/api/) cover custom hosts,
152
+ packing Releases and Git integration.
@@ -23,7 +23,7 @@
23
23
  "class_name": "ComputerDO",
24
24
  "image": "./node_modules/@fungi.computer/computer/Dockerfile",
25
25
  "instance_type": "standard-4",
26
- "max_instances": 1,
26
+ "max_instances": 4,
27
27
  "rollout_active_grace_period": 0,
28
28
  "rollout_step_percentage": [100],
29
29
  },
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@fungi.computer/oyster",
3
- "version": "0.13.1",
3
+ "version": "0.14.0",
4
4
  "author": "Levi <levi@technomancy.dev>",
5
5
  "description": "Run your own Apps on Cloudflare Workers or celld: installable App Releases, each with its own Durable Object, SQLite database and declared bindings",
6
6
  "type": "module",
@@ -344,37 +344,37 @@
344
344
  "dependencies": {
345
345
  "@cfworker/json-schema": "4.1.1",
346
346
  "@earendil-works/pi-ai": "1.3.0",
347
- "@fungi.computer/agar": "0.7.1",
348
- "@fungi.computer/bird-dog": "0.3.3",
349
- "@fungi.computer/braid": "0.4.3",
347
+ "@fungi.computer/agar": "0.7.2",
348
+ "@fungi.computer/bird-dog": "0.3.4",
349
+ "@fungi.computer/braid": "0.4.4",
350
350
  "@fungi.computer/burrow": "0.2.1",
351
351
  "@fungi.computer/cairn": "0.1.0",
352
- "@fungi.computer/capnweb": "0.12.0",
352
+ "@fungi.computer/capnweb": "0.12.0-fungi.0",
353
353
  "@fungi.computer/caps": "0.7.1",
354
- "@fungi.computer/carrier": "0.2.1",
355
- "@fungi.computer/computer": "0.6.1",
356
- "@fungi.computer/computer-contract": "0.0.10",
357
- "@fungi.computer/darkroom": "0.5.1",
354
+ "@fungi.computer/carrier": "0.2.2",
355
+ "@fungi.computer/computer": "0.6.2",
356
+ "@fungi.computer/computer-contract": "0.0.11",
357
+ "@fungi.computer/darkroom": "0.5.2",
358
358
  "@fungi.computer/ear-tag": "0.1.0",
359
- "@fungi.computer/forage": "0.3.1",
360
- "@fungi.computer/gills": "0.2.1",
361
- "@fungi.computer/knapsack": "0.4.5",
359
+ "@fungi.computer/forage": "0.3.2",
360
+ "@fungi.computer/gills": "0.2.2",
361
+ "@fungi.computer/knapsack": "0.4.6",
362
362
  "@fungi.computer/lantern": "0.4.1",
363
- "@fungi.computer/mycelium": "0.3.4",
364
- "@fungi.computer/nacre": "0.2.0",
365
- "@fungi.computer/parakeet": "0.8.2",
366
- "@fungi.computer/puffball": "0.5.1",
367
- "@fungi.computer/seam": "0.2.2",
368
- "@fungi.computer/shiitake": "0.11.1",
369
- "@fungi.computer/shiitake-extensions": "0.6.6",
363
+ "@fungi.computer/mycelium": "0.3.5",
364
+ "@fungi.computer/nacre": "0.2.1",
365
+ "@fungi.computer/parakeet": "0.9.0",
366
+ "@fungi.computer/puffball": "0.5.2",
367
+ "@fungi.computer/seam": "0.2.3",
368
+ "@fungi.computer/shiitake": "0.12.0",
369
+ "@fungi.computer/shiitake-extensions": "0.6.7",
370
370
  "@fungi.computer/sigils": "0.3.0",
371
371
  "@fungi.computer/sprite": "0.2.0",
372
372
  "@fungi.computer/sprite-zdog": "0.3.0",
373
- "@fungi.computer/veil": "0.6.1",
374
- "@fungi.computer/watchdog": "0.1.0",
373
+ "@fungi.computer/veil": "0.6.2",
374
+ "@fungi.computer/watchdog": "0.1.1",
375
375
  "@fungi.computer/whistle": "0.3.0",
376
- "@fungi.computer/workflows-wire": "0.4.3",
377
- "@fungi.computer/zookeeper": "0.0.13",
376
+ "@fungi.computer/workflows-wire": "0.4.4",
377
+ "@fungi.computer/zookeeper": "0.0.14",
378
378
  "@modelcontextprotocol/server": "2.0.0",
379
379
  "@pnpm/lockfile.fs": "1100.2.13",
380
380
  "@pnpm/lockfile.pruner": "1100.0.27",