@fugood/buttress-server 2.25.7 → 2.25.9

package/README.md

@@ -69,14 +69,14 @@ The `bricks` CLI is the tool that performs the binding and writes the local stat
 ### Bind a server to a workspace
 
 ```bash
-# Pair the local machine's buttress-server with the workspace of the current bricks-cli profile
+# Pair the local server and provision a workspace DevTools token using the current CLI profile
 bricks buttress bind
 
 # Override the auto-detected server id, give it a friendly name, or write to a custom state dir
 bricks buttress bind --server-id buttress-mac-studio --name "Studio LLM" --state-dir /etc/buttress
 
 # For headless/remote setups: emit state.json to stdout instead of writing to disk
-bricks buttress bind --print > /etc/buttress/state.json
+(umask 077; bricks buttress bind --print > /etc/buttress/state.json)
 ```
 
 The state file (`~/.bricks-cli/buttress/state.json` by default, or `$BRICKS_BUTTRESS_STATE_DIR`) stores:
@@ -84,6 +84,15 @@ The state file (`~/.bricks-cli/buttress/state.json` by default, or `$BRICKS_BUTT
 - `workspace.id` / `workspace.name` — which workspace this server belongs to
 - `workspace.serverId` — the server's stable id (defaults to `buttress-<machineId>`)
 - `workspace.issuerPublicKey` + `workspace.kid` — Ed25519 SPKI used to verify access tokens
+- `devtools` — workspace-scoped DevTools JWT (`k: "da"`), workspace id and expiry,
+  provisioned automatically by `bind` for agent device tools
+
+Normal bind/status output reports only token expiry, never the token. The CLI writes
+state with mode `0600`; `--print` includes credentials and must be treated as a secret.
+Rebinding provisions a fresh token (default lifetime: 30 days) and rotates the announce
+key. Issuance failure stops binding before changing the remote announce key. Rebinding
+or unbinding does not revoke previously issued DevTools JWTs; they remain valid until
+expiry or issuer-key rotation.
 
 **Restart `bricks-buttress` after binding** for the change to take effect — the state file is read once at startup.
 
@@ -207,6 +216,8 @@ Most ggml-llm `[generators.model]` keys can also live in `[runtime]` as defaults
 | `no_extra_bufts`             | boolean                       | Disable extra compute buffer types                                     |
 | `cpu_mask`, `cpu_strict`     | string / boolean              | CPU affinity (advanced)                                                |
 | `devices`                    | string[]                      | Restrict to specific GGML devices                                      |
+| `split_mode`                 | `"none"` / `"layer"` / `"row"` / `"tensor"` | ggml-llm GPU splitting: single GPU, layers + KV (default), weight rows, or weights + KV (experimental tensor parallelism) |
+| `tensor_split`               | number[]\|string              | ggml-llm per-GPU proportions, e.g. `[3, 1]` or `"3,1"`; unset uses the native automatic split |
 | Speculative keys             | various                       | `speculative`, `spec_type`, `spec_draft_n_max/n_min/p_min/p_split`, plus draft-model GPU/cache settings |
 
 ### `[autodiscover]`
@@ -268,6 +279,10 @@ Additional keys honored by **ggml-llm**, **ggml-stt** and **ggml-tts** (mlx-llm
 
 Loads a GGUF LLM. Runtime keys above can be overridden per-generator under `[generators.model]`; `[generators.backend]` only controls backend selection and resource planning.
 
+GPU splitting (`split_mode`, `tensor_split`) is currently configurable only in Buttress.
+TODO: Expose these options in the LLM (GGML) generator. The experimental `tensor` mode
+requires a compatible GPU backend and model architecture.
+
 **`[generators.backend]`**
 
 | Key                   | Type     | Default                                       | Notes                                                            |
@@ -283,6 +298,7 @@ Loads a GGUF LLM. Runtime keys above can be overridden per-generator under `[gen
 | ----------------------------------------------------------------------------- | ---------------- | -------------------------------------------------------------------- |
 | `n_ctx`                                                                       | number           | Context window. Auto-capped at the model's training context.         |
 | `n_gpu_layers`                                                                | number\|`"auto"` | Layers offloaded to GPU (default `"auto"`)                           |
+| `split_mode`, `tensor_split`                                                   | string / number[]\|string | Per-model overrides for the GPU split defaults in `[runtime]`       |
 | `n_batch`                                                                     | number           | Prompt batch size (default `512`)                                    |
 | `embedding`                                                                   | boolean          | Enable llama.cpp embedding mode for a dedicated embedding model (default `false`) |
 | `pooling_type`                                                                | string           | Optional embedding pooling override such as `mean`                   |
@@ -355,12 +371,21 @@ spec_draft_cache_type_v = "f16"
 
 Loads a Whisper GGML model for speech-to-text.
 
+With `@fugood/whisper.node` 1.2.0-rc.0, `snapdragon` enables experimental
+Whisper Hexagon NPU acceleration on Linux arm64 with a working Qualcomm
+HTP/FastRPC runtime. This variant does not include OpenCL. Install it with
+`--ggml-variant=snapdragon` (the default CPU package remains available for fallback).
+The platform package includes the HTP libraries; the loader defaults
+`ADSP_LIBRARY_PATH` to that package directory and `GGML_HEXAGON_NDEV` to `16`,
+without overriding explicit environment values. Hexagon always enables flash
+attention, even when `use_flash_attn` is off.
+
 **`[generators.backend]`**
 
 | Key                   | Type     | Default                       | Notes                              |
 | --------------------- | -------- | ----------------------------- | ---------------------------------- |
-| `variant`             | string   | auto                          | `cuda` / `vulkan` / `default`      |
-| `variant_preference`  | string[] | `["cuda","vulkan","default"]` | Probe order                        |
+| `variant`             | string   | auto                          | `cuda` / `vulkan` / `snapdragon` / `default` |
+| `variant_preference`  | string[] | `["cuda","vulkan","snapdragon","default"]` | Probe order                  |
 | `gpu_memory_fraction` | number   | `0.85`                        |                                    |
 | `cpu_memory_fraction` | number   | `0.5`                         |                                    |
 
@@ -385,7 +410,7 @@ Loads a Whisper GGML model for speech-to-text.
 [[generators]]
 type = "ggml-stt"
 [generators.backend]
-variant_preference = ["cuda", "vulkan", "default"]
+variant_preference = ["cuda", "vulkan", "snapdragon", "default"]
 [generators.model]
 repo_id = "BricksDisplay/whisper-ggml"
 filename = "ggml-large-v3-turbo-q8_0.bin"
@@ -867,6 +892,7 @@ model = "buttress/ggml-org/gpt-oss-20b-GGUF"  # split on the FIRST slash: provid
 # model = "anthropic/claude-sonnet-5"         # any pi-supported provider; API key from env
 system_prompt_file = "./prompts/ops.md"       # or inline: system_prompt = "..."
 tools = ["get_server_status", "restart_service"]  # local function names (explicit; no wildcard)
+local_devices = false                        # opt in to LAN scan + device DevTools tools
 max_turns = 30                                 # assistant↔tool round-trips per run (default 30)
 # max_tokens_per_run = 200000                 # per-run token budget; unset = unlimited
 # temperature = 0.2                           # unrecognized keys pass through to generation
@@ -901,6 +927,38 @@ improvise around it. MCP tools get server-qualified names
 (`mcp__github__create_issue`); MCP servers connect lazily on the first run and
 fail closed unless marked `optional = true`.
 
+**Local devices.** Set `local_devices = true` on an individual `[[agents]]` entry
+(default `false`) to add just two built-in tools, independently of local functions/MCP:
+
+- `devices_scan`: the same UDP + HTTP LAN discovery as `bricks devtools scan`,
+  including project previews found relative to the Buttress config directory.
+- `devtools`: one action-based tool for screenshots (inline images), brick tree/query,
+  tap/key/text input, console capture, JavaScript evaluation, storage overview, and
+  raw CDP requests for other supported operations. Reuses the CLI CDP transport and
+  shared DevTools operations; no CLI subprocess or separate tool per device/action.
+
+For example, scan with `{}` then call `devtools` with
+`{ "action": "tree", "address": "192.168.1.42", "port": 19851 }`.
+After `bricks buttress bind` and a restart, the workspace credential is injected
+server-side automatically when the device advertises the same workspace and issuer.
+The token is validated for signature, kind and expiry before use; inbound Buttress
+JWTs (`k: "ba"`) are never forwarded. Older bindings without the credential, or expired
+credentials, require rebinding and a restart. The server does not read the CLI login
+or renew tokens itself. Local previews can still use advertised inspect credentials;
+explicit `passcode` / `accessToken` arguments remain optional overrides. Calls open/close their own connection;
+avoid parallel calls or other inspectors on the same device (one CDP client at a time).
+
+**Security:** enabling this option grants network discovery and device control,
+including arbitrary JavaScript evaluation on reachable authenticated devices. It is
+not read-only or a network sandbox. Automatic authentication requires a **trusted LAN**:
+workspace/issuer hints from device info are not cryptographic peer authentication.
+The stored token is never inserted into model arguments and is redacted from text
+results/errors before truncation. Explicit credentials supplied in arguments still
+enter session transcripts, as do console output and device data; prefer automatic
+authentication and only grant this to trusted agents/callers. Disable with `local_devices = false` and
+restart Buttress. Aborting closes active CDP connections; discovery completes its
+bounded scan before returning the abort result.
+
 **Sessions.** Each run returns a `sessionId`; pass it back to continue the
 conversation, or add `fork: true` to branch it into a fresh session. Sessions
 are JSONL files under `sessions_dir`, scoped by agent name (renaming an agent

package/config/sample.toml

@@ -55,6 +55,7 @@ enabled = true
 # model = "buttress/ggml-org/gpt-oss-20b-GGUF"  # provider/model-id, split on the FIRST slash
 # system_prompt = "You are an ops automation agent. Use the provided tools."
 # tools = ["host-info"]                # local function names (explicit; no wildcard)
+# local_devices = false               # opt in: devices_scan + devtools (inspection AND control)
 # max_turns = 30
 # [agents.mcp_servers.example]
 # url = "https://example.com/mcp/"
@@ -77,6 +78,12 @@ cache_type_k = "q8_0"
 cache_type_v = "q8_0"
 n_threads = 6
 
+# Optional ggml-llm multi-GPU defaults (also supported under [generators.model]).
+# split_mode = "layer"   # none / layer (default) / row / tensor (experimental)
+# tensor_split = [3, 1]  # per-GPU proportions; also accepts "3,1"
+# Tensor mode requires a compatible GPU backend and model architecture.
+# TODO: Expose these Buttress-only options in the LLM (GGML) generator.
+
 # Delay before releasing context when refCount reaches 0 (default: 10000ms)
 # Set to 0 for immediate release
 context_release_delay_ms = 30000

package/lib/agent/device-auth.d.ts

@@ -0,0 +1,3 @@
+import type { WorkspaceState } from '../utils/workspaceState';
+/** Select only the bound workspace's credential; never reuse an inbound Buttress (k=ba) JWT. */
+export declare const resolveDeviceAccessToken: (address: string, port: number, state?: WorkspaceState, signal?: AbortSignal) => Promise<string | undefined>;

package/lib/agent/device-tools.d.ts

@@ -0,0 +1,5 @@
+import type { AgentFunctionTool } from './tools';
+import type { AgentDefinition } from './types';
+import type { WorkspaceState } from '../utils/workspaceState';
+/** Two tools regardless of device count; no remote tool-list expansion or persistent sockets. */
+export declare const buildLocalDeviceTools: (agent: Pick<AgentDefinition, 'localDevices'>, configDir: string, workspaceState?: WorkspaceState) => AgentFunctionTool[];

package/lib/agent/types.d.ts

@@ -1,7 +1,7 @@
 /**
  * Agent feature types: config-defined agents that run pi-agent-core loops
- * inside the buttress-server process, with local functions (and MCP servers)
- * as their tools and config-scoped JSONL sessions.
+ * inside the buttress-server process, with local functions, MCP servers and
+ * opt-in device tools, plus config-scoped JSONL sessions.
  */
 export type AgentMcpServerConfig = {
     /** StreamableHTTP endpoint; mutually exclusive with `command`. */
@@ -27,6 +27,8 @@ export type AgentDefinition = {
     systemPromptFile: string | null;
     /** Local function names exposed as tools (explicit list, no wildcard). */
     tools: string[];
+    /** Opt-in LAN discovery and device DevTools access (default false). */
+    localDevices: boolean;
     mcpServers: Record<string, AgentMcpServerConfig>;
     /** Assistant↔tool round-trips per run. */
     maxTurns: number;

package/lib/{cli-DrbWX4ea.mjs → cli-CtBbHMYQ.mjs}

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import{a as e,i as t,n,r,t as i}from"./client-BCBBen9i.mjs";import a from"node:readline";const o=`\x1B[2m`,s=`\x1B[33m`,c=`\x1B[31m`,l=`\x1B[0m`,u=e=>{let t={name:null,configArg:null,url:null,token:null,sessionId:null,fork:!1,listSessions:!1,plain:!1};for(let n=0;n<e.length;n+=1){let r=e[n];r===`-h`||r===`--help`?(console.log(`
+import{a as e,i as t,n,r,t as i}from"./client-DzfRSFcJ.mjs";import a from"node:readline";const o=`\x1B[2m`,s=`\x1B[33m`,c=`\x1B[31m`,l=`\x1B[0m`,u=e=>{let t={name:null,configArg:null,url:null,token:null,sessionId:null,fork:!1,listSessions:!1,plain:!1};for(let n=0;n<e.length;n+=1){let r=e[n];r===`-h`||r===`--help`?(console.log(`
 bricks-buttress agent — chat with a configured agent on a running server
 
 Usage:
@@ -19,4 +19,4 @@ In the chat: /exit quits, /new starts a fresh session, Ctrl+C aborts the
 current run (a second Ctrl+C quits).
 `),process.exit(0)):r===`-c`||r===`--config`?t.configArg=e[++n]??null:r===`--url`?t.url=e[++n]??null:r===`--token`?t.token=e[++n]??null:r===`--session`?t.sessionId=e[++n]??null:r===`--fork`?(t.sessionId=e[++n]??null,t.fork=!0):r===`--sessions`?t.listSessions=!0:r===`--plain`?t.plain=!0:!r.startsWith(`-`)&&!t.name?t.name=r:(console.error(`Unknown argument: ${r}`),process.exit(1))}return t},d=e=>{console.error(`${c}${e}${l}`),process.exit(1)},f=async(t,n,r,i,a,u)=>{let d=i,f=!1,p=!1,m=()=>{f&&=(process.stdout.write(`${l}\n`),!1)},h=await e(t,n,{prompt:r,sessionId:i??void 0,fork:a},{signal:u,onFrame:({event:e,payload:t})=>{if(e===`session`){t.sessionId&&t.sessionId!==d&&(d=t.sessionId,i||console.log(`${o}session ${d}${l}`));return}if(e!==`agent`)return;let n=t?.event;if(t?.type===`message_update`&&n)n.type===`thinking_delta`?(f||=(process.stdout.write(`${o}`),!0),process.stdout.write(n.delta??``),p=!0):n.type===`text_delta`&&(m(),process.stdout.write(n.delta??``),p=!0);else if(t?.type===`tool_execution_start`){m(),p&&process.stdout.write(`
 `);let e=JSON.stringify(t.args??{});console.log(`${s}⚙ ${t.toolName}${l}${o}(${e.length>160?`${e.slice(0,160)}…`:e})${l}`),p=!1}else t?.type===`tool_execution_end`?console.log(t.isError?`${c}✗ ${t.toolName} failed${l}`:`${o}✓ ${t.toolName}${l}`):t?.type===`tool_emit`&&console.log(`${o}  ${t.toolName} → ${t.event}${l}`)}});if(m(),h.kind===`result`){let e=h.result;d=e.sessionId??d;let t=e.usage||{};return process.stdout.write(`
-`),console.log(`${o}— ${e.stopReason} · ${t.totalTurns??`?`} turn(s) · ${t.input??0} in / ${t.output??0} out · session ${d}${l}`),{sessionId:d,ok:!0}}return h.kind===`aborted`?(console.log(`\n${s}(aborted)${l}`),{sessionId:h.sessionId??d,ok:!1}):h.kind===`error`?(console.error(`\n${c}Run failed: ${h.message}${l}`),{sessionId:h.sessionId??d,ok:!1}):(console.error(`\n${c}Connection lost mid-run.${l}`),{sessionId:h.sessionId??d,ok:!1})},p=async(e,t,n,r)=>{let s=n,u=r;console.log(`${t}${l} @ ${e.baseUrl}`+(s?` ${o}(${u?`forking`:`continuing`} ${s})${l}`:``)),console.log(`${o}/exit to quit, /new for a fresh session, Ctrl+C aborts a running turn${l}`);let d=a.createInterface({input:process.stdin,output:process.stdout}),p=!1;d.on(`close`,()=>{p=!0});let m=e=>new Promise(t=>{if(p){t(null);return}d.question(e,t),d.once(`close`,()=>t(null))}),h=null;for(d.on(`SIGINT`,()=>{h?(h.abort(),h=null):(d.close(),process.exit(0))});;){let n=await m(`> ${l}`);if(n==null)break;let r=n.trim();if(!r)continue;if(r===`/exit`||r===`/quit`)break;if(r===`/new`){s=null,u=!1,console.log(`${o}Started a fresh session.${l}`);continue}h=new AbortController;let a=s;try{let n=await f(e,t,r,s,u,h.signal);n.sessionId&&(s=n.sessionId),i(a,s)&&(u=!1)}catch(e){console.error(`${c}${e?.message||e}${l}`)}finally{h=null}}d.close()},m=async e=>{let i=u(e),{config:a,agentsConfig:s}=n(i.configArg),c=a.server.port||2080,f=(i.url||`http://127.0.0.1:${c}`).replace(/\/$/,``),m={baseUrl:f,token:i.token||(s?r(s.runtimeTokenFile):null)||process.env.BUTTRESS_AGENT_TOKEN||null},h=await t(m,`/agents`).catch(e=>d(`Cannot reach ${f}: ${e?.message||e} (is the server running?)`));if(!h.ok){let e=await h.text().catch(()=>``);d(`Server rejected the request (${h.status}): ${e.slice(0,300)}`)}let{agents:g}=await h.json();if(!i.name){console.log(`Configured agents on ${f}:`);for(let e of g)console.log(`- ${e}`);g.length===0&&console.log(`(none — add [[agents]] to the server config)`);return}if(g.includes(i.name)||d(`Unknown agent "${i.name}" (configured: ${g.join(`, `)||`none`})`),i.listSessions){let{sessions:e}=await(await t(m,`/agents/${encodeURIComponent(i.name)}/sessions`)).json();if(!e?.length){console.log(`No sessions yet.`);return}for(let t of e)console.log(`${t.sessionId}  ${o}${t.updatedAt}${l}${t.parentSession?`  ${o}(fork of ${t.parentSession})${l}`:``}`),t.preview&&console.log(`  ${o}${t.preview}${l}`);return}if(process.stdin.isTTY&&process.stdout.isTTY&&!i.plain){let{runAgentTui:e}=await import(`./tui-7B7x6A08.mjs`);await e({connection:m,agentName:i.name,sessionId:i.sessionId,fork:i.fork});return}await p(m,i.name,i.sessionId,i.fork)};export{m as runAgentCommand};
+`),console.log(`${o}— ${e.stopReason} · ${t.totalTurns??`?`} turn(s) · ${t.input??0} in / ${t.output??0} out · session ${d}${l}`),{sessionId:d,ok:!0}}return h.kind===`aborted`?(console.log(`\n${s}(aborted)${l}`),{sessionId:h.sessionId??d,ok:!1}):h.kind===`error`?(console.error(`\n${c}Run failed: ${h.message}${l}`),{sessionId:h.sessionId??d,ok:!1}):(console.error(`\n${c}Connection lost mid-run.${l}`),{sessionId:h.sessionId??d,ok:!1})},p=async(e,t,n,r)=>{let s=n,u=r;console.log(`${t}${l} @ ${e.baseUrl}`+(s?` ${o}(${u?`forking`:`continuing`} ${s})${l}`:``)),console.log(`${o}/exit to quit, /new for a fresh session, Ctrl+C aborts a running turn${l}`);let d=a.createInterface({input:process.stdin,output:process.stdout}),p=!1;d.on(`close`,()=>{p=!0});let m=e=>new Promise(t=>{if(p){t(null);return}d.question(e,t),d.once(`close`,()=>t(null))}),h=null;for(d.on(`SIGINT`,()=>{h?(h.abort(),h=null):(d.close(),process.exit(0))});;){let n=await m(`> ${l}`);if(n==null)break;let r=n.trim();if(!r)continue;if(r===`/exit`||r===`/quit`)break;if(r===`/new`){s=null,u=!1,console.log(`${o}Started a fresh session.${l}`);continue}h=new AbortController;let a=s;try{let n=await f(e,t,r,s,u,h.signal);n.sessionId&&(s=n.sessionId),i(a,s)&&(u=!1)}catch(e){console.error(`${c}${e?.message||e}${l}`)}finally{h=null}}d.close()},m=async e=>{let i=u(e),{config:a,agentsConfig:s}=n(i.configArg),c=a.server.port||2080,f=(i.url||`http://127.0.0.1:${c}`).replace(/\/$/,``),m={baseUrl:f,token:i.token||(s?r(s.runtimeTokenFile):null)||process.env.BUTTRESS_AGENT_TOKEN||null},h=await t(m,`/agents`).catch(e=>d(`Cannot reach ${f}: ${e?.message||e} (is the server running?)`));if(!h.ok){let e=await h.text().catch(()=>``);d(`Server rejected the request (${h.status}): ${e.slice(0,300)}`)}let{agents:g}=await h.json();if(!i.name){console.log(`Configured agents on ${f}:`);for(let e of g)console.log(`- ${e}`);g.length===0&&console.log(`(none — add [[agents]] to the server config)`);return}if(g.includes(i.name)||d(`Unknown agent "${i.name}" (configured: ${g.join(`, `)||`none`})`),i.listSessions){let{sessions:e}=await(await t(m,`/agents/${encodeURIComponent(i.name)}/sessions`)).json();if(!e?.length){console.log(`No sessions yet.`);return}for(let t of e)console.log(`${t.sessionId}  ${o}${t.updatedAt}${l}${t.parentSession?`  ${o}(fork of ${t.parentSession})${l}`:``}`),t.preview&&console.log(`  ${o}${t.preview}${l}`);return}if(process.stdin.isTTY&&process.stdout.isTTY&&!i.plain){let{runAgentTui:e}=await import(`./tui-DbQ0zW-C.mjs`);await e({connection:m,agentName:i.name,sessionId:i.sessionId,fork:i.fork});return}await p(m,i.name,i.sessionId,i.fork)};export{m as runAgentCommand};

package/lib/{client-BCBBen9i.mjs → client-DzfRSFcJ.mjs}

@@ -1,5 +1,5 @@
 #!/usr/bin/env node
-import{a as e,n as t}from"./config-lP89VahD.mjs";import n from"node:path";import r from"node:fs";import i from"@iarna/toml";const a=a=>{let o={},s=process.cwd();a&&(r.existsSync(a)?(o=i.parse(r.readFileSync(a,`utf8`)),s=n.dirname(n.resolve(a))):o=i.parse(a));let c=e(o);return{config:c,configDir:s,agentsConfig:t(c,{configDir:s})}},o=e=>{try{return r.readFileSync(e,`utf8`).trim()||null}catch{return null}},s=async(e,t,n)=>fetch(`${e.baseUrl}${t}`,{...n,headers:{"content-type":`application/json`,...e.token?{authorization:`Bearer ${e.token}`}:{},...n?.headers}});async function*c(e){let t=e.getReader(),n=new TextDecoder,r=``;try{for(;;){let{done:e,value:i}=await t.read();if(e)break;r+=n.decode(i,{stream:!0});let a=r.indexOf(`
+import{a as e,n as t}from"./config-DbRjQnNp.mjs";import n from"node:path";import r from"node:fs";import i from"@iarna/toml";const a=a=>{let o={},s=process.cwd();a&&(r.existsSync(a)?(o=i.parse(r.readFileSync(a,`utf8`)),s=n.dirname(n.resolve(a))):o=i.parse(a));let c=e(o);return{config:c,configDir:s,agentsConfig:t(c,{configDir:s})}},o=e=>{try{return r.readFileSync(e,`utf8`).trim()||null}catch{return null}},s=async(e,t,n)=>fetch(`${e.baseUrl}${t}`,{...n,headers:{"content-type":`application/json`,...e.token?{authorization:`Bearer ${e.token}`}:{},...n?.headers}});async function*c(e){let t=e.getReader(),n=new TextDecoder,r=``;try{for(;;){let{done:e,value:i}=await t.read();if(e)break;r+=n.decode(i,{stream:!0});let a=r.indexOf(`
 
 `);for(;a!==-1;){let e=r.slice(0,a);r=r.slice(a+2),a=r.indexOf(`
 

package/lib/{config-lP89VahD.mjs → config-DbRjQnNp.mjs}

@@ -1,2 +1,2 @@
 #!/usr/bin/env node
-import e from"node:path";import t from"node:os";import n from"bytes";import r from"ms";import i from"node-machine-id";const a=(e={},t={})=>{let n=Array.isArray(e)?[...e]:{...e};return Object.entries(t||{}).forEach(([e,t])=>{t&&typeof t==`object`&&!Array.isArray(t)?n[e]=a(n[e]||{},t):n[e]=t}),n},o=e=>e&&typeof e==`object`?structuredClone(e):null,s=(e,t)=>{let n=o(e)||{},r=o(t)||{};return a(n,r)},c=(e,t)=>a(structuredClone(e.global),t||{}),l=(e,t,n,r)=>{if(e.generators.length>0){let i=e.generators.filter(e=>e?.type===n);if(i.length>0&&r){let a=i.find(e=>t.getModelIdentifier(n,e)===r);if(a)return c(e,a)}}return Object.keys(e.global).length>0?c(e,{}):null},u={udp:{port:8089,announcements:{enabled:!0,interval:5e3},requests:{enabled:!0,responseDelay:100}},http:{enabled:!0,path:`/buttress/info`,cors:!0}},d=e=>e?e===!0?{...u}:a(u,e):null,f=(e,t)=>{if(!e.generators||e.generators.length===0)return t.map(e=>({type:e}));let n=new Set;return e.generators.forEach(e=>{e.type&&n.add(e.type)}),n.size===0?t.map(e=>({type:e})):Array.from(n).map(e=>({type:e}))},p=(e,t,n)=>e===void 0?n:typeof e==`number`?e:t(e)??n,m=6e4,h=1024*1024*50,g=s=>{let c=i.machineIdSync(),l={server:{id:`buttress-${c}`,name:`Buttress Server (${c.slice(-8)})`,port:2080,temp_file_dir:e.join(t.tmpdir(),`.buttress`),session_timeout:m,max_body_size:h},autodiscover:!1},u=a(l,o(s)||{}),f=Array.isArray(u.generators)?u.generators:[],{server:g,generators:_,autodiscover:v,...y}=u;return{autodiscover:d(v),server:{id:g.id,name:g.name,port:g.port,log_level:g.log_level,temp_file_dir:g.temp_file_dir,max_body_size:p(g.max_body_size,n.parse,h),session_timeout:p(g.session_timeout,e=>r(e),m)},global:y,generators:f}},_=/^[A-Za-z0-9_-]{1,64}$/,v=n=>n===`~`||n.startsWith(`~/`)?e.join(t.homedir(),n.slice(1)):n;var y=class extends Error{constructor(e){super(`[Agents] ${e}`),this.name=`AgentsConfigError`}};const b=(e,t,n)=>{if(t==null)return n;if(t===0||t===!1)return null;if(typeof t==`number`){if(Number.isFinite(t)&&t>0)return t}else if(typeof t==`string`){let e;try{e=r(t)}catch{e=void 0}if(typeof e==`number`&&e>0)return e}throw new y(`agents_options.${e} must be a duration string (e.g. "30d") or a positive number of milliseconds; use 0 to disable it`)},x=(e,t)=>{if(t==null)return{};if(typeof t!=`object`||Array.isArray(t))throw new y(`agent "${e}": mcp_servers must be a table`);let n={};for(let[r,i]of Object.entries(t)){if(!_.test(r))throw new y(`agent "${e}": MCP server name "${r}" must match ${_}`);let t=!!(typeof i?.url==`string`&&i.url),a=!!(typeof i?.command==`string`&&i.command);if(t===a)throw new y(`agent "${e}": MCP server "${r}" needs exactly one of url / command`);n[r]={url:t?i.url:void 0,headers:i?.headers&&typeof i.headers==`object`?i.headers:void 0,command:a?i.command:void 0,args:Array.isArray(i?.args)?i.args.map(String):void 0,env:i?.env&&typeof i.env==`object`?i.env:void 0,optional:i?.optional===!0}}return n},S=(e,t,n)=>{let r=e?.name;if(typeof r!=`string`||!_.test(r))throw new y(`[[agents]] #${t+1}: name is required and must match ${_}`);let i=e?.model;if(typeof i!=`string`||!i.includes(`/`))throw new y(`agent "${r}": model must be "provider/model-id" (e.g. "buttress/ggml-org/gpt-oss-20b-GGUF")`);let a=i.indexOf(`/`),o=i.slice(0,a),s=i.slice(a+1);if(!o||!s)throw new y(`agent "${r}": invalid model reference "${i}"`);if(o===`buttress`&&!n.generators.filter(e=>e.type===`ggml-llm`||e.type===`mlx-llm`).some(e=>(e.model?.repo_id||e.model?.repository)===s))throw new y(`agent "${r}": model "${s}" does not match any configured [[generators]] ggml-llm/mlx-llm repo_id`);if(e?.system_prompt!=null&&e?.system_prompt_file!=null)throw new y(`agent "${r}": set system_prompt or system_prompt_file, not both`);let c=[];if(e?.tools!=null){if(!Array.isArray(e.tools))throw new y(`agent "${r}": tools must be an array of function names`);for(let t of e.tools){if(typeof t!=`string`||!_.test(t))throw new y(`agent "${r}": tool name "${t}" must match ${_} (local function names only)`);if(c.includes(t))throw new y(`agent "${r}": duplicate tool "${t}"`);c.push(t)}}let l=Number(e?.max_turns??30);if(!Number.isInteger(l)||l<1)throw new y(`agent "${r}": max_turns must be a positive integer`);let u=null;if(e?.max_tokens_per_run!=null&&(u=Number(e.max_tokens_per_run),!Number.isInteger(u)||u<1))throw new y(`agent "${r}": max_tokens_per_run must be a positive integer`);let{name:d,model:f,system_prompt:p,system_prompt_file:m,tools:h,mcp_servers:g,max_turns:v,max_tokens_per_run:b,thinking:S,...C}=e;return{name:r,provider:o,modelId:s,modelRef:i,systemPrompt:typeof p==`string`?p:null,systemPromptFile:typeof m==`string`?m:null,tools:c,mcpServers:x(r,g),maxTurns:l,maxTokensPerRun:u,thinking:typeof S==`string`?S:void 0,generation:C}},C=(t,{configDir:n=process.cwd()}={})=>{let r=t.global||{},i=r.agents;if(i==null)return null;if(!Array.isArray(i))throw new y(`agents must be an array of tables ([[agents]])`);if(i.length===0)return null;let a=i.map((e,n)=>S(e,n,t)),o=new Set;for(let e of a){if(o.has(e.name))throw new y(`duplicate agent name "${e.name}"`);o.add(e.name)}let s=r.agents_options||{},c=typeof s.sessions_dir==`string`&&s.sessions_dir?s.sessions_dir:`./.buttress-agent/sessions`,l=v(c),u=e.isAbsolute(l)?l:e.resolve(n,l);for(let t of a)if(t.systemPromptFile){let r=v(t.systemPromptFile);t.systemPromptFile=e.isAbsolute(r)?r:e.resolve(n,r)}let d=Number(s.max_depth??2);if(!Number.isInteger(d)||d<1)throw new y(`agents_options.max_depth must be a positive integer`);let f=s.session_max_count,p=500;if(f===0||f===!1)p=null;else if(f!=null&&(p=Number(f),!Number.isInteger(p)||p<1))throw new y(`agents_options.session_max_count must be a positive integer`);return{agents:a,configDir:n,sessionsDir:u,runtimeTokenFile:e.join(e.dirname(u),`runtime-token`),sessionMaxAgeMs:b(`session_max_age`,s.session_max_age,2592e6),sessionMaxCount:p,maxDepth:d,allowUnauthenticated:s.allow_unauthenticated===!0}};export{g as a,o as i,C as n,l as o,s as r,f as s,y as t};
+import e from"node:path";import t from"node:os";import n from"bytes";import r from"ms";import i from"node-machine-id";const a=(e={},t={})=>{let n=Array.isArray(e)?[...e]:{...e};return Object.entries(t||{}).forEach(([e,t])=>{t&&typeof t==`object`&&!Array.isArray(t)?n[e]=a(n[e]||{},t):n[e]=t}),n},o=e=>e&&typeof e==`object`?structuredClone(e):null,s=(e,t)=>{let n=o(e)||{},r=o(t)||{};return a(n,r)},c=(e,t)=>a(structuredClone(e.global),t||{}),l=(e,t,n,r)=>{if(e.generators.length>0){let i=e.generators.filter(e=>e?.type===n);if(i.length>0&&r){let a=i.find(e=>t.getModelIdentifier(n,e)===r);if(a)return c(e,a)}}return Object.keys(e.global).length>0?c(e,{}):null},u={udp:{port:8089,announcements:{enabled:!0,interval:5e3},requests:{enabled:!0,responseDelay:100}},http:{enabled:!0,path:`/buttress/info`,cors:!0}},d=e=>e?e===!0?{...u}:a(u,e):null,f=(e,t)=>{if(!e.generators||e.generators.length===0)return t.map(e=>({type:e}));let n=new Set;return e.generators.forEach(e=>{e.type&&n.add(e.type)}),n.size===0?t.map(e=>({type:e})):Array.from(n).map(e=>({type:e}))},p=(e,t,n)=>e===void 0?n:typeof e==`number`?e:t(e)??n,m=6e4,h=1024*1024*50,g=s=>{let c=i.machineIdSync(),l={server:{id:`buttress-${c}`,name:`Buttress Server (${c.slice(-8)})`,port:2080,temp_file_dir:e.join(t.tmpdir(),`.buttress`),session_timeout:m,max_body_size:h},autodiscover:!1},u=a(l,o(s)||{}),f=Array.isArray(u.generators)?u.generators:[],{server:g,generators:_,autodiscover:v,...y}=u;return{autodiscover:d(v),server:{id:g.id,name:g.name,port:g.port,log_level:g.log_level,temp_file_dir:g.temp_file_dir,max_body_size:p(g.max_body_size,n.parse,h),session_timeout:p(g.session_timeout,e=>r(e),m)},global:y,generators:f}},_=/^[A-Za-z0-9_-]{1,64}$/,v=n=>n===`~`||n.startsWith(`~/`)?e.join(t.homedir(),n.slice(1)):n;var y=class extends Error{constructor(e){super(`[Agents] ${e}`),this.name=`AgentsConfigError`}};const b=(e,t,n)=>{if(t==null)return n;if(t===0||t===!1)return null;if(typeof t==`number`){if(Number.isFinite(t)&&t>0)return t}else if(typeof t==`string`){let e;try{e=r(t)}catch{e=void 0}if(typeof e==`number`&&e>0)return e}throw new y(`agents_options.${e} must be a duration string (e.g. "30d") or a positive number of milliseconds; use 0 to disable it`)},x=(e,t)=>{if(t==null)return{};if(typeof t!=`object`||Array.isArray(t))throw new y(`agent "${e}": mcp_servers must be a table`);let n={};for(let[r,i]of Object.entries(t)){if(!_.test(r))throw new y(`agent "${e}": MCP server name "${r}" must match ${_}`);let t=!!(typeof i?.url==`string`&&i.url),a=!!(typeof i?.command==`string`&&i.command);if(t===a)throw new y(`agent "${e}": MCP server "${r}" needs exactly one of url / command`);n[r]={url:t?i.url:void 0,headers:i?.headers&&typeof i.headers==`object`?i.headers:void 0,command:a?i.command:void 0,args:Array.isArray(i?.args)?i.args.map(String):void 0,env:i?.env&&typeof i.env==`object`?i.env:void 0,optional:i?.optional===!0}}return n},S=(e,t,n)=>{let r=e?.name;if(typeof r!=`string`||!_.test(r))throw new y(`[[agents]] #${t+1}: name is required and must match ${_}`);let i=e?.model;if(typeof i!=`string`||!i.includes(`/`))throw new y(`agent "${r}": model must be "provider/model-id" (e.g. "buttress/ggml-org/gpt-oss-20b-GGUF")`);let a=i.indexOf(`/`),o=i.slice(0,a),s=i.slice(a+1);if(!o||!s)throw new y(`agent "${r}": invalid model reference "${i}"`);if(o===`buttress`&&!n.generators.filter(e=>e.type===`ggml-llm`||e.type===`mlx-llm`).some(e=>(e.model?.repo_id||e.model?.repository)===s))throw new y(`agent "${r}": model "${s}" does not match any configured [[generators]] ggml-llm/mlx-llm repo_id`);if(e?.system_prompt!=null&&e?.system_prompt_file!=null)throw new y(`agent "${r}": set system_prompt or system_prompt_file, not both`);let c=[];if(e?.tools!=null){if(!Array.isArray(e.tools))throw new y(`agent "${r}": tools must be an array of function names`);for(let t of e.tools){if(typeof t!=`string`||!_.test(t))throw new y(`agent "${r}": tool name "${t}" must match ${_} (local function names only)`);if(c.includes(t))throw new y(`agent "${r}": duplicate tool "${t}"`);c.push(t)}}if(e?.local_devices!==void 0&&typeof e.local_devices!=`boolean`)throw new y(`agent "${r}": local_devices must be a boolean`);let l=Number(e?.max_turns??30);if(!Number.isInteger(l)||l<1)throw new y(`agent "${r}": max_turns must be a positive integer`);let u=null;if(e?.max_tokens_per_run!=null&&(u=Number(e.max_tokens_per_run),!Number.isInteger(u)||u<1))throw new y(`agent "${r}": max_tokens_per_run must be a positive integer`);let{name:d,model:f,system_prompt:p,system_prompt_file:m,tools:h,local_devices:g=!1,mcp_servers:v,max_turns:b,max_tokens_per_run:S,thinking:C,...w}=e;return{name:r,provider:o,modelId:s,modelRef:i,systemPrompt:typeof p==`string`?p:null,systemPromptFile:typeof m==`string`?m:null,tools:c,localDevices:g,mcpServers:x(r,v),maxTurns:l,maxTokensPerRun:u,thinking:typeof C==`string`?C:void 0,generation:w}},C=(t,{configDir:n=process.cwd()}={})=>{let r=t.global||{},i=r.agents;if(i==null)return null;if(!Array.isArray(i))throw new y(`agents must be an array of tables ([[agents]])`);if(i.length===0)return null;let a=i.map((e,n)=>S(e,n,t)),o=new Set;for(let e of a){if(o.has(e.name))throw new y(`duplicate agent name "${e.name}"`);o.add(e.name)}let s=r.agents_options||{},c=typeof s.sessions_dir==`string`&&s.sessions_dir?s.sessions_dir:`./.buttress-agent/sessions`,l=v(c),u=e.isAbsolute(l)?l:e.resolve(n,l);for(let t of a)if(t.systemPromptFile){let r=v(t.systemPromptFile);t.systemPromptFile=e.isAbsolute(r)?r:e.resolve(n,r)}let d=Number(s.max_depth??2);if(!Number.isInteger(d)||d<1)throw new y(`agents_options.max_depth must be a positive integer`);let f=s.session_max_count,p=500;if(f===0||f===!1)p=null;else if(f!=null&&(p=Number(f),!Number.isInteger(p)||p<1))throw new y(`agents_options.session_max_count must be a positive integer`);return{agents:a,configDir:n,sessionsDir:u,runtimeTokenFile:e.join(e.dirname(u),`runtime-token`),sessionMaxAgeMs:b(`session_max_age`,s.session_max_age,2592e6),sessionMaxCount:p,maxDepth:d,allowUnauthenticated:s.allow_unauthenticated===!0}};export{g as a,o as i,C as n,l as o,s as r,f as s,y as t};

package/lib/functions/types.d.ts

@@ -1,6 +1,5 @@
 import type { AgentsFunctionApi, AgentsService } from '../agent/types';
-import type { Backend } from '../index';
-import type { Config } from '../types';
+import type { Backend, Config } from '../types';
 import type { VerifiedIdentity } from '../utils/buttressAuth';
 import type { WorkspaceState } from '../utils/workspaceState';
 /**

package/lib/index.d.ts

@@ -1,7 +1,6 @@
 import type { AnyElysia } from 'elysia';
-import * as backendCore from '@fugood/buttress-backend-core';
 import { AutodiscoverService } from './autodiscover';
-import type { Config } from './types';
+import type { Backend, Config } from './types';
 import { compareVersions } from './utils/update';
 import type { AgentsConfig, AgentsService } from './agent/types';
 import type { FunctionsConfig, FunctionsService } from './functions';
@@ -14,7 +13,7 @@ export declare const checkForUpdates: () => Promise<string | null>;
 export { compareVersions };
 export declare const logUpdateMessage: (latestVersion: string) => void;
 export declare const checkAndNotifyUpdates: () => Promise<void>;
-export type Backend = typeof backendCore;
+export type { Backend } from './types';
 export interface StartServerOptions {
     backend?: Backend;
     router?: AnyElysia;