| Age | Commit message (Collapse) | Author |
|---|
|
Two improvements to the SSH support feature:
1. KNOWN_HOSTS DISCOVERY (packages/ssh):
Computers are now auto-discovered from ~/.ssh/known_hosts (every hostname
you've ever connected to) in ADDITION to ~/.ssh/config (explicit Host
aliases). Config entries take precedence (full params); known_hosts entries
get defaulted params (User=defaultUser, IdentityFile=null→pool probes
default keys, Port from [host]:port or 22, knownHost=true). Zero-config —
no ~/.ssh/config file needed; hosts just appear.
Reject list: dispatch.toml [ssh].reject = [...] (glob patterns like
github.com, *.ts.net) filters noise from the catalog. Read from both
the global ~/.config/dispatch/dispatch.toml and the project dispatch.toml.
Parsed with Bun.TOML.parse (zero deps). Only filters discovery (catalog);
specific lookups (getComputer/getStatus/test/connect) ignore the reject
list (it's a visibility filter, not access control).
New pure functions: parseKnownHosts(), isRejected(), globMatch().
+26 tests. tsc EXIT 0, biome clean, 1756 tests pass.
2. REMOTE SYSTEM-PROMPT AWARENESS (packages/system-prompt):
When a conversation has a computerId set (remote turn), the system prompt
now resolves system:os, system:hostname, git:branch/git:status, and
file: reads against the REMOTE machine — not the local host. Previously
the prompt always said 'Arch Linux (WSL)' + local hostname even when the
agent was connected to a remote Artix Linux machine.
The ResolverAdapters' hostname()/platform() are now async (so a remote
adapter can run 'hostname'/'uname -s' over SSH). The system-prompt
extension builds remote adapters from the ExecBackend (readFile→SFTP,
spawn→SSH exec). Cache invalidation now checks computerId (switching
computers rebuilds the prompt). The compaction path also threads
computerId. @dispatch/system-prompt now depends on @dispatch/exec-backend.
|
|
barrel
Wave 5c (final wiring) of transparent SSH support.
- host-bin: register exec-backend + ssh in CORE_EXTENSIONS (exec-backend before
the tool extensions that dependsOn it; ssh after, provides the remote-backend
factory + ComputerService at boot). +@dispatch/exec-backend/@dispatch/ssh deps +
tsconfig refs.
- transport-http: CR-5 — re-export computerServiceHandle + ComputerService type
from the package barrel (src/index.ts), mirroring lsp/mcp handles, so ssh imports
the typed symbol cleanly (no more dist/seam.js subpath workaround).
- orchestrator: added the @dispatch/exec-backend dep the host-bin agent missed +
bun install.
LIVE-VERIFIED: bun packages/host-bin/src/main.ts boots clean ('Dispatch booted',
no disabled extensions) — exec-backend + ssh + all tool extensions load together.
Verified: tsc -b EXIT 0, biome clean, 1690 vitest pass (+6 sshd-integration skipped).
DEFERRED (CR-6): listComputers usageCount stays 0 until a conversation-store
count-by-alias helper is added (non-blocking).
Refs: notes/ssh-support-plan.md. No merge or push.
|
|
Wave 5b of transparent SSH support. NEW standard extension @dispatch/ssh makes
remote execution actually work over SSH, transparently. ssh2 verified to run under
Bun (load-bearing decision #1 confirmed: connects to local sshd :22 + execs).
- config.ts: ~/.ssh/config reader via ssh-config -> Computer[]/ComputerEntry[]
(read-only discovery; resolves hostName/port/user/identityFile/knownHost).
- hostkey.ts: known_hosts auto-trust-and-pin (present->verify/reject-on-mismatch,
absent->accept+append; the accept-new analog).
- errors.ts: pure ssh2/SFTP -> node:fs-style .code error mapping (so tools'
existing ENOENT branches work unchanged).
- pool.ts: SshConnectionPool (per-alias ssh2.Client, lazy connect, keep-alive,
idle reap ~15m); key-only auth from ~/.ssh (config IdentityFile or default
id_ed25519/id_rsa); no agent-forwarding, no PTY.
- backend.ts: SshExecBackend implements ExecBackend (spawn via client.exec with
shell-quoted cwd; fs via SFTP).
- service.ts + extension.ts: activate provides BOTH handles the other units
consume — remoteExecBackendFactoryHandle (exec-backend: computerId->SshExecBackend)
AND computerServiceHandle (transport-http: listComputers/getComputer/getStatus/test).
- orchestrator: added packages/ssh to root tsconfig.json refs + bun install.
Tests: 45 pass + 6 sshd-integration skipped (it.skipIf(!process.env.SSH_TEST_HOST)).
Verified: tsc -b EXIT 0, biome clean, 1690 vitest pass (was 1641, +49).
CRs for wave 5c: host-bin registration; CR-5 transport-http barrel re-export;
CR-6 usageCount wiring (deferred-ok, defaults to 0).
Refs: notes/ssh-support-plan.md (decisions §0.5/§13). No merge or push.
|
|
Wave 4 of transparent SSH support (3 parallel owner-agents on disjoint packages).
- transport-http: computer routes — GET /computers, GET /computers/:alias,
GET /computers/:alias/status, POST /computers/:alias/test (all delegate to a
new ComputerService seam, graceful []/disconnected when ssh not loaded);
GET/PUT/DELETE /conversations/:id/computer; PUT /workspaces/:id/default-computer
(mirror the cwd/default-cwd routes); /chat threads computerId into the
orchestrator. Defines ComputerService interface + computerServiceHandle
(defineService<ComputerService>('ssh')) in seam.ts — the seam the ssh package
provides via host.provideService in wave 5.
- transport-ws: chat.send + chat.queue thread computerId onto the route result
(mirrors cwd/workspaceId), forwarded to the orchestrator input.
- mcp: CR-1 fix — filterMcpTools now preserves computerId on the returned
ToolAssembly (mirrors cwd preservation), so the filter chain stays consistent.
- orchestrator: added @dispatch/wire dep to transport-http (build/config, my lane)
so its seam.ts Computer/ComputerEntry import resolves.
Verified: tsc -b EXIT 0, biome clean, 1641 vitest pass (was 1620, +21).
Refs: notes/ssh-support-plan.md (decisions §0.5/§13). No merge or push.
|
|
Wave 2 of transparent SSH support (4 parallel owner-agents on disjoint
tool packages). The tools now resolve an ExecBackend per-call from
ctx.computerId and call backend.spawn / backend.readFile / etc. instead of
node:fs and node:child_process directly — so they are transport-agnostic
(local now; remote over SSH later, transparent to the agent). Still LOCAL-ONLY
this wave (computerId always undefined -> LocalExecBackend, behavior-identical).
- tool-shell: factory takes resolveBackend; execute calls backend.spawn.
spawn.ts DELETED (realSpawn was a verbatim duplicate of exec-backend's
LocalExecBackend.spawn — logic moved to the sanctioned shared package).
manifest dependsOn:[exec-backend]; host.getService at activation.
- tool-read-file: readFile/stat/readdir -> backend.* (pure logic untouched;
ENOENT .code branches kept).
- tool-write-file: exists/stat/writeFile -> backend.* (pure logic untouched).
- tool-edit-file: readFile/writeFile -> backend.* + forward-compatible REMOTE
diagnostics skip (ctx.computerId set -> skip LSP, return empty — plan §6.1;
local path byte-identical to today). LSP lookup stays lazy.
- orchestrator: pre-wired @dispatch/exec-backend dep into the 4 tool
package.jsons + bun install (build/config, my lane) so isolated verify
resolved cleanly; agents added the ../exec-backend tsconfig ref.
Verified: tsc -b EXIT 0, biome clean, 1599 vitest pass (was 1592).
Refs: notes/ssh-support-plan.md (decisions §0.5/§13). No merge or push.
|
|
Wave 1 of transparent SSH support (parallel owner-agents on disjoint packages,
plus the orchestrator-authored kernel contract seam from wave 0):
- packages/wire: + Computer/ComputerEntry (read-only view over ~/.ssh/config
Host aliases) + Workspace.defaultComputerId (string|null, null=local). Types
only; 3 conformance tests.
- packages/exec-backend (NEW core extension): the ExecBackend abstraction
(spawn + minimal fs surface) the bundled tools will program against instead
of node:fs/child_process. LocalExecBackend wraps today's node calls
(behavior-identical; node:fs-style .code errors). execBackendHandle +
ExecBackendResolver (sync; computerId undefined -> local; set -> throws until
the ssh package wires remote resolution in wave 5). 20 tests.
- packages/kernel (runtime only): thread computerId through dispatch.ts +
run-turn.ts exactly as cwd is threaded (opaque, forwarded to
ToolExecuteContext; absent = local = byte-identical to today). +2 tests.
- packages/conversation-store: computer (SSH alias) assignment + resolution
mirroring cwd — WorkspaceRow.defaultComputerId + setWorkspaceDefaultComputerId
+ getComputerId/setComputerId/clearComputerId + getEffectiveComputer
(override -> per-conv -> workspace default -> null/local). Fixes the 3
Workspace literal sites the new required wire field broke. +18 tests.
- orchestrator: root tsconfig.json ref for exec-backend + bun install.
Verified: tsc -b EXIT 0, biome clean, 1592 vitest pass (was 1549, +43).
Refs: notes/ssh-support-plan.md (decisions §0.5/§13). No merge or push.
|
|
Mirrors the existing GET /conversations/:id/lsp route exactly: gates on the
persisted then effective cwd (null → empty servers), returns 503 when the
MCP service isn't loaded, and maps McpServerStatus → McpServerInfo
(conditionally including `error` per exactOptionalPropertyTypes).
Wires mcpService into CreateServerOptions + extension activate via a plain
host.getService (mirroring lspService; "mcp" added to dependsOn, route added
to contributes.routes), adds the @dispatch/mcp workspace dep, and re-exports
mcpServiceHandle / McpService / McpServerStatus from seam.ts. Adds 4 tests
mirroring the LSP status tests.
|
|
New `mcp` standard extension (`packages/mcp/`) that makes Dispatch an MCP
host: spawns configured MCP servers (stdio child processes), performs the MCP
handshake (initialize → notifications/initialized), discovers tools via
tools/list, and registers each as a first-class Dispatch ToolContract via
host.defineTool. When the model calls an MCP tool, the extension proxies the
call to tools/call on the MCP server and returns the flattened result.
Architecture (sibling of `lsp` extension):
- Config: .dispatch/mcp.json (servers key) → opencode.json mcp key fallback,
resolved per-cwd (mirrors LSP config resolution)
- Transport: StdioTransport (spawn child, Content-Length framing + JSON-RPC 2.0)
- Client: initialize → tools/list → tools/call; handles list_changed
notifications for dynamic tool updates
- Registry: tool name namespacing (<serverId>__<toolName>), ToolContract
adapter that proxies execute → callTool, content flattening (text/image/
resource → string)
- Manager: one client per server, lazy-spawn, status(), shutdownAll()
- Extension: manifest (dependsOn session-orchestrator, capabilities spawn),
registers tools + a toolsFilter (drops disconnected server's tools),
mcpServiceHandle, deactivate kills all child processes
Phase 1 scope: stdio only, Tools only (no Resources/Prompts/HTTP/sampling).
Hand-rolled JSON-RPC + framing (zero external deps, adapts LSP patterns).
Wave 1 (agent): 12 source + 8 test files, 69 new tests.
Wave 2 (orchestrator): root tsconfig ref, host-bin CORE_EXTENSIONS
registration + package.json dep, bun install.
Verified: tsc -b EXIT 0, biome clean, 1537 vitest pass (was 1468, +69).
|
|
LSP extension:
- Multi-server aggregation: query ALL connected servers matching the
file's extension (not just the first), merge diagnostics tagged by source
- Incremental sync: capture each server's textDocumentSync.change during
initialize; compute prefix/suffix diff ranges for change:2 servers;
full content for change:1 (generic, works for any LSP)
- New diff.ts: pure computeChangeRange + offsetToPosition (O(n), tested)
- Buffer sync: change(filePath, newText) sends didChange with post-edit
in-memory content; openWithText for first open; tracks open doc text
- languageId mapping: extended with .rb/.rbs/.c/.cpp/etc. (was 'unknown')
- waitForDiagnostics: accepts text override + timeoutMs; returns
{ formatted, slow, timedOut }; polls for publishDiagnostics push
- DiagnosticsStore: hasReceivedPush/clearReceived tracking; formatFiltered
with minSeverity (1=Error, 2=Warning) for edit_file integration
- LspService.getDiagnostics: service method for cross-extension use
tool-edit-file:
- After successful edit, calls LSP getDiagnostics with post-edit buffer
- Only appends diagnostics with severity ≤ 2 (errors+warnings, no noise)
- Appends slow warning (>10s): 'LSP is taking unusually long...'
- 60s timeout; graceful degradation when no LSP available
- Optional dep on @dispatch/lsp (getService pattern, not manifest depOn)
1468 vitest pass (was 1453, +15 new diff tests).
|
|
Register @dispatch/system-prompt in CORE_EXTENSIONS (after skills, before
cache-warming). Add dep + tsconfig ref. 1396 vitest pass, typecheck + biome clean.
|
|
session-orchestrator:
- Wire systemPromptService as optional dep (lazy via host.getService)
- Regular turn: construct on first turn (new conversation), get on subsequent
turns, set on providerOpts.systemPrompt (cache-safe)
- Compaction: construct (fresh resolve) + append COMPACTION_SYSTEM_PROMPT
- 12 new tests (construct/get/service-unavailable/compaction)
transport-http:
- GET /system-prompt (returns template or DEFAULT_TEMPLATE)
- PUT /system-prompt (validate + setTemplate, 503 when unavailable)
- GET /system-prompt/variables (static catalog, always available)
- 6 new tests
system-prompt service: added getTemplate/setTemplate to interface + impl.
1396 vitest pass. typecheck + biome clean.
|
|
New @dispatch/system-prompt extension (standard tier):
- Pure parser: [type:name] variables, [if]/[else]/[endif] conditionals,
negated [if !...], nested blocks, unmatched-tag pass-through.
- Variable resolver (injected adapters): system:time/date/os/hostname,
prompt:cwd/model/conversation_id, git:branch/status, file:<path> (dynamic).
- Service handle: construct (resolve+persist) + get (cached, cache-safe).
- Default template: persona + AGENTS.md if exists + cwd.
- 52 tests (parser 29, resolver 12, catalog 3, service 8).
transport-contract 0.17.0→0.18.0: SystemPromptTemplateResponse,
SetSystemPromptTemplateRequest, SystemPromptVariable, SystemPromptVariablesResponse.
Design: notes/system-prompt-design.md (caching constraint, compaction
integration, wave plan). 1384 vitest pass.
|
|
Wire 0.12.0: Workspace, WorkspaceEntry, ConversationMeta.workspaceId
Transport-contract 0.16.0: workspaceId on ChatRequest/QueueRequest/ChatQueueMessage;
workspace endpoint types (EnsureWorkspaceRequest, WorkspaceResponse, etc.)
Kernel: re-export Workspace/WorkspaceEntry from contracts
Conversation-store: workspace persistence + service methods (getWorkspace,
ensureWorkspace, setWorkspaceTitle, setWorkspaceDefaultCwd, deleteWorkspace,
listWorkspaces, getWorkspaceId, setWorkspaceId, getEffectiveCwd, isValidWorkspaceSlug);
listConversations filter by workspaceId; forkHistory/replaceHistory preserve
workspaceId. 111 tests pass.
FE handoff: frontend-workspaces-handoff.md (courier doc)
18 typecheck errors in session-orchestrator/transport-http/cli test fakes
(expected fan-out — fixed in Wave 2+3).
|
|
New standard tool extension backed by a self-hosted transcriber service
(http://100.102.55.49:41090, Tailscale, no API key). One tool
youtube_transcript — fetches transcripts for YouTube videos. Returns
completed (full text + timestamped segments), queued/processing (position
+ ETA + .youtube_subtitles_pending retry convention), or failed (error).
Pure core: validateUrl + format* functions + truncateOutput. Injected
edge: TranscriptClient (injectable fetchFn, AbortSignal.any for
cancellation). concurrencySafe true, capabilities network. 30 tests.
Verified: tsc EXIT 0, 1152 vitest, biome clean (327 files). Boot smoke
clean.
|
|
New standard tool extension with a single todo_write tool (opencode
todowrite pattern: full-list replace, returns JSON, no business-rule
enforcement — the description guides the model). Per-conversation in-memory
state + per-conversation surface (rendererId: todo, scope: conversation)
via subscriber-notify (message-queue pattern).
Wave 0 (kernel contract): added conversationId?: string to ToolExecuteContext
(additive, backward-compatible). Wired in dispatch.ts — the kernel already
had it but wasn't passing it through to tools.
Wave 1 (todo extension): pure core (validateTodos — shape only; getTodos/
setTodos/clearTodos; buildTodoSpec; formatTodoResult). Shell:
createTodoWriteTool + surface provider. Tool description matches opencode's
todowrite.txt depth (when-to-use, examples, task states). Priority field
removed (bloats the tool with little value). 25 tests.
Wave 2 (host-bin): registered todo in CORE_EXTENSIONS + dep + root tsconfig ref.
Verified: tsc EXIT 0, 1123 vitest, biome clean (314 files). Boot smoke clean.
FE handoff: frontend-todo-handoff.md.
|
|
New standard tool extension with one tool web_search supporting 4 modes
(search, scrape, crawl, map) against a self-hosted Firecrawl instance.
Pure core: validateArgs (discriminated union by mode) + format* functions
+ truncateOutput. Injected edge: FirecrawlClient (injectable fetchFn/sleep/now,
AbortSignal.any for per-request timeout + caller cancellation). concurrencySafe
true, capabilities network. 38 tests, zero vi.mock.
Live-verified: umans-glm-5.2 called web_search → real Firecrawl results (also
the first live Umans API call).
|
|
Extract a generic @dispatch/openai-stream library from provider-openai-compat
(convert-messages, convert-tools, parse-sse, listModels, stream, provider),
parameterizing createOpenAICompatProvider with uid=1000(tradam) gid=1000(tradam) groups=1000(tradam),966(docker),968(ollama),998(wheel) + hook.
Refactor provider-openai-compat to import from the lib (byte-identical behavior).
New @dispatch/provider-umans extension wraps the Umans OpenAI-compatible backend
(https://api.code.umans.ai/v1). Self-contained: reads UMANS_API_KEY from env
directly (no auth-apikey dep). transformBody maps reasoningEffort →
reasoning_effort (capping xhigh/max → high). Dynamic listModels via GET /v1/models.
host-bin: registered provider-umans in CORE_EXTENSIONS + umans credential
(gated on UMANS_API_KEY — the credential is the model-catalog index).
Verified: tsc EXIT 0, 1059 vitest, biome clean (293 files). Boot smoke:
umans models appear in GET /models (7 models live).
|
|
A per-conversation message queue (new message-queue extension) holds user
messages enqueued while a turn generates; delivered mid-turn as steering at the
tool-result boundary (or carried to a new turn if no tool call fires).
- kernel: RunTurnInput.drainSteering callback (generic; kernel stays pure)
- wire 0.7.0->0.8.0: QueuedMessage, QueuePayload, TurnSteeringEvent (additive)
- transport-contract 0.11.0->0.12.0: POST /conversations/:id/queue + chat.queue WS op
- message-queue ext: queue state + per-conversation custom surface (rendererId message-queue)
- session-orchestrator: enqueue facade + drainSteering wiring + post-seal carry
- transport-http/ws: queue endpoint + chat.queue op (fixes WsClientMessage exhaustive switch)
- host-bin: register message-queue
1043 vitest + 199 transport bun pass; tsc/biome clean; boot smoke clean.
FE courier: frontend-message-queue-handoff.md.
|
|
cache bust
LSP + per-conversation CWD feature:
- new bundled `lsp` extension: hand-rolled JSON-RPC codec (framing/rpc), lazy
one-server-per-(serverID,root), per-cwd config resolution, on-demand `lsp` tool
- `conversation-store`: getCwd/setCwd (cwdKey); `session-orchestrator` defaults a
turn's cwd from the store
- `transport-http`: cwd + lsp status endpoints; wire types in transport-contract
- host-bin: register lsp; config wiring
Cache-warming fix (the warm read 0% on the first reheat after a message):
- warm assembled tools under a different cwd than the real turn (a reheat sends no
cwd, and the warm service had no store fallback). The skills filter rewrites the
cwd-sensitive `load_skill` description, so the tools block — the first bytes of
the prompt-cache prefix — diverged and the cache missed entirely. Warm now
resolves cwd as opts.cwd ?? conversationStore.getCwd(), mirroring handleMessage.
- capture warm sends as `provider.request` spans flagged `warm:true` (thread a
child logger into providerOpts) so warm vs real bodies are diffable (obs §3.1).
- kernel logger: span-close now merges child-bound attrs like span-open, so a
`warm:true` query finds the closed span (with usage/status), not just the open.
Tests: warm forwards a warm-flagged logger; warm falls back to stored cwd; logger
open/close attr consistency. Full suite green (873).
|
|
Backend-driven warming targeting whatever provider a conversation uses (incl. the
external Claude provider-anthropic). Core engine + on/off + last-cache-% done;
interval-as-view-control pending a ui-contract NumberField (surface-system gap).
Mechanism:
- kernel: expose HostAPI.emit (typed bus event emit; counterpart of on)
- session-orchestrator: turnStarted/turnSettled event hooks (conversationId/cwd/model);
warm() service (cacheWarmHandle) reusing the real-turn assembly (byte-identical prefix,
provider-agnostic), refuses mid-turn, never persists/emits, returns Usage
- cache-warming (new ext): per-conversation timers (arm on settle, cancel on start,
in-flight invalidation), calls warm(), pct=round(clamp(cacheRead/input,0,1)*100),
persists {enabled,intervalMs} (default on/240s), registers a controls surface
- host-bin: register cache-warming; transport-http: HostAPI stub +emit (fan-out)
Honors old-code invariants. 760 vitest + 109 bun = 869 tests; tsc -b EXIT 0; biome clean.
|
|
Skills are markdown in .skills/ dirs (~/.skills + <cwd>/.skills, cwd shadows home;
name = filename). Format: line1 summary, line2 ---, body line3+; load strips the
first two lines; malformed = no summary but still loadable.
Mechanism (first use of the context-assembly filter chain, §3.2):
- kernel: expose HostAPI.applyFilters (delegates to bus.applyFilters)
- session-orchestrator: define/export toolsFilter + ToolAssembly; apply once per turn
before runTurn (cache-stable across steps), threading cwd + conversationId
- skills (new ext): pure parse/merge/render + load_skill tool (live read, path-contained)
+ a toolsFilter filter rewriting load_skill's description + name enum per cwd
- host-bin: register skills in CORE_EXTENSIONS
- transport-http: fix HostAPI test stub for the new applyFilters method (fan-out)
734 vitest + 109 bun = 843 tests; tsc -b EXIT 0; biome clean; clean live boot.
|
|
Four standard-tier tool extensions (one tool per extension, zero ABI change):
- tool-read-file: read_file now lists directory contents (sorted, /-suffixed subdirs)
- tool-shell: run_shell (foreground, streamed, cancellable, cwd, timeout + output cap)
- tool-edit-file: edit_file (oldString/newString/replaceAll; errors on absent/non-unique)
- tool-write-file: write_file (explicit overwrite flag)
Registered in host-bin CORE_EXTENSIONS. Live boot clean (shell capability accepted).
686 vitest + 89 bun = 775 tests; tsc -b EXIT 0; biome clean.
|
|
New throughput-store extension records one token-weighted sample per turn
(model, output tokens, pure generation time = Σ step genTotalMs) into a
day-bucketed KV store, and aggregates per-model tok/s = Σtokens / Σgen-seconds
over a day/week/month (server-local boundaries; week = ISO Mon–Sun).
transport-http records a sample per turn (logged) and serves
GET /metrics/throughput?period=day|week|month&date=<...>. The response is typed
as transport-contract's ThroughputResponse, so store/wire drift is a compile
error. Pure period + aggregate logic fully unit-tested.
|
|
tool calls
Expose a per-step grouping key so a client can render a model's batched/parallel
tool calls (those emitted in one step) as one unit, on both the live stream and
replayed history. Key = branded StepId, derived turnId#stepIndex (0-based).
- [email protected]: required stepId on Turn{Tool,ToolResult}Event; optional stepId on
Tool{Call,Result}Chunk (generation provenance on the chunk, not the StoredChunk
envelope — StoredChunk unchanged). [email protected] (re-export bump).
- kernel-runtime: mint stepId per step; stamp on tool chunks + tool events.
- conversation-store: chunk-carried stepId round-trips append/load/loadSince for
free; reconcile copies it onto synthesized (interrupted) results.
- cli: stepId added to event test fixtures (renderer unchanged).
typecheck clean; 509 vitest + 89 bun; biome 0/0. FE courier reply + reference
snapshots regenerated in ../dispatch-web.
|
|
Add chat WS ops (chat.send / chat.delta / chat.error) + unified
WsClientMessage/WsServerMessage unions to @dispatch/transport-contract
(imports ui-contract; surface protocol unchanged — additive non-colliding
type variants, no channel wrapper). transport-ws drives
sessionOrchestrator.handleMessage, streaming each AgentEvent as chat.delta
over the same connection that carries surface ops; per-connection
AbortController cancels in-flight turns on socket close; error-isolated.
Verified live: one WS connection delivered the surface catalog AND a real
flash chat turn (chat.delta stream, reply 'Hello my friend').
Completes the FE Slice 2 backend prereqs. typecheck clean, 485 vitest + 80 bun,
biome clean.
Discovered (separate, pre-existing): runtime does not emit
turn-start/done/turn-sealed on either transport — needed for FE cache-commit;
tracked in tasks.md.
|
|
endpoint
Incremental rehydration endpoint for long-lived clients. Returns
ConversationHistoryResponse { chunks: StoredChunk[], latestSeq } — the RAW,
append-order, seq-filtered slice from conversation-store.loadSince, NOT
reconciled (reconcile conflicts with the per-chunk seq cursor, so it stays on
the turn path; the read path is a pure sync primitive).
- transport-contract: add ConversationHistoryResponse + StoredChunk re-export.
- transport-http: GET /conversations/:id route reaching the log directly via
conversationStoreHandle (dependsOn conversation-store); pure parseSinceSeq
(absent->0, invalid->400).
- build wiring: conversation-store dep + project ref.
FE Slice 2 backend prereq (read-side). typecheck clean, 481 vitest, biome clean.
|
|
split (B2)
FE slice 1 — backend-declared, frontend-agnostic surface system (verified live): new types-only @dispatch/ui-contract (SurfaceSpec / field kinds / region / ActionRef / catalog), surface-registry (typed service handle), transport-ws (Bun WS :24205, path-agnostic upgrade), surface-loaded-extensions (first real surface); kernel HostAPI.getExtensions; host-bin wiring; bin/up. Harness: retire AGENTS 'backend only', ORCHESTRATOR §3/§7/§8, frontend-design.md locked.
B2 — wire-types split (chat-slice prerequisite): new types-only @dispatch/wire single-sources the wire ABI (AgentEvent + 11 variants; conversation model Chunk/ChatMessage/Role/TurnId/StepId + 6 chunk variants; Usage) with zero @dispatch/* deps. @dispatch/kernel re-exports via shims so its public surface is byte-identical (zero consumer blast radius). transport-contract re-exports AgentEvent from @dispatch/wire and drops its @dispatch/kernel dependency, so HTTP clients (the web frontend) consume the wire without the kernel runtime.
tsc -b + biome clean; 460 vitest + 77 bun pass.
|
|
transport-contract wire package
|
|
env-gated capture + hermetic fixture tests (331 tests)
provider-openai-compat now consumes @dispatch/trace-replay. (A) Opt-in record mode: when DISPATCH_RECORD_FIXTURE is set, the fetch edge wraps recordFetch and saves a fixture of the verbatim post-transform request + raw SSE response, self-redacting the auth header in the provider's OWN code (reuses its existing maskSecret graduated-tier mask — no shared helper, isolation over DRY). Zero overhead when unset; fail-safe. (B) Hermetic replay tests: stream.test.ts drives the provider off committed SSE fixtures via replayFetch (chunk-split to exercise SSE parsing across boundaries), asserting ProviderEvents + that the outgoing request still matches the recorded one (transform-drift regression). Injectable fetch via an internal StreamConfig.fetchFn — NO kernel contract change.
2 committed fixtures (text-turn + tool-call, currently hand-authored-faithful; a real flash text-turn swap follows). Verified: tsc -b clean, 331 vitest (327 -> +4: 2 replay + 2 redaction), biome 0/0. Provider 44 -> 48.
|
|
library (39 tests)
New standalone package @dispatch/trace-replay: replayFetch (pure — fixture -> fetch double + captured request, optional chunking to simulate streaming), recordFetch (tees a real fetch into a fixture WITHOUT consuming the caller's stream), and serialize/parse + save/load fixture I/O. Redaction-free by design: calling extensions self-redact in their OWN code before saving (isolation over DRY, D5/§9). Zero @dispatch/* deps, no bun:sqlite (runs under vitest). The shared unit realizing the §7/D5 replay affordance for hermetic provider tests; provider-openai-compat will consume it next.
Root tsconfig ref wired. Verified: tsc -b clean, 327 vitest (288 -> +39: replay 12 / record 8 / fixture 19), biome 0/0. Agent stayed in lane (packages/trace-replay only).
|
|
collector + trace CLI (345 tests)
trace-store (bun:sqlite): records+bodies schema (thin/fat split), idempotent insertRecords (FNV-1a id + INSERT OR IGNORE), getTurn/getBody, pure renderEasyView (D8 timeline skeleton), trace CLI. Its own DB, isolated from storage-sqlite.
observability-collector: out-of-process bin — tail journal -> splitLines/drainOnce -> trace-store.insertRecords; offset sidecar; at-least-once + idempotent; fail-safe; clean SIGINT/SIGTERM drain.
Build-config (orchestrator): root tsconfig refs; both excluded from vitest + added to test:bun (bun:sqlite); bun install.
Verified: tsc -b clean, 345 tests (273 vitest + 72 bun), biome 0 warnings/0 infos. Pipeline proven end-to-end: app -> journal -> collector -> SQLite -> 'trace <turnId>' easy-view. Known follow-up (next commit): kernel spans are flat (parent=ROOT) — run-turn nesting fix.
|
|
sink (250 tests)
Structured, agent-first logging captured durably to an append-only journal file.
Kernel (contracts/logging.ts): leveled/attributed Logger + Span, auto-scoped per extension (host stamps manifest.id, unspoofable), incremental span records (open/close) for crash-reconstructable traces, injected LogSink (pure record-builder). ctx.log on ToolContract; runTurn opens turn/step/tool-call spans and captures the verbatim pre-mutation prompt (the 'before') on the step span.
journal-sink (new package, bootstrap dep — not an extension): LogSink appending NDJSON to a rotating journal; pure serialize + thin fs edge; fail-safe drop, never blocks a turn. host-bin injects it via HostDeps; session-orchestrator threads host.logger (childed per turn) into runTurn.
Redaction is per-extension self-redaction (no shared helper — isolation over DRY). The out-of-process collector + SQLite store + the verbatim 'after' provider.request capture are Phase B / next (notes/observability-design.md §10/§11).
Verified: tsc -b clean, 250 tests (218→+32), biome clean. Live boot: a turn's journal holds host logs + turn/step spans (open+close) + the prompt:before record with the verbatim messages array.
Harness: ORCHESTRATOR §3 rule-scoping map; .dispatch/rules/isolation-over-dry.md; notes/observability-design.md (design D1–D10 + Phase A/B plan).
|
|
First TOOL extension (standard tier, fs capability). Pure-core/shell split with
workdir containment (realpath symlink guard). host-bin registers it in
CORE_EXTENSIONS; flows into runTurn via session-orchestrator's resolveTools.
Verified: typecheck clean, 214 tests pass (was 185), biome clean. Live curl
against flash produced a real tool-call + tool-result round-trip with correct
final answer. Proves the kernel tool-dispatch loop end-to-end (plan §3.3).
|
|
Bun.serve; full-fidelity wiring (178 tests)
|
|
build graph (164 tests)
|
|
- storage-sqlite: bun:sqlite StorageNamespace backend + migrations (21 bun tests)
- auth-apikey: pure resolver from env → ApiKeyCredentials (4 tests)
- provider-openai-compat: OpenAI-compatible SSE stream → ProviderEvents
- orchestrator fixes: provider imports (@dispatch/kernel), missing dep,
exactOptionalPropertyTypes (omit-when-undefined), root tsconfig refs
- vitest excludes storage-sqlite (bun:sqlite); test:bun runs it under bun
|
|
stub)
|
|
|
|
Add Language Server Protocol integration modeled on opencode's, wired for
this codebase's plain-TypeScript tool/agent architecture.
Core (@dispatch/core):
- lsp/client.ts: LSP/JSON-RPC client over stdio (vscode-jsonrpc) with the
initialize handshake, didOpen/didChange sync, push + pull diagnostics
(textDocument/diagnostic, workspace/diagnostic), and a generic request()
passthrough for hover/definition/references/documentSymbol.
- lsp/server.ts: resolves dispatch.toml [lsp] entries into spawn specs.
Config-driven only — no builtin registry, no auto-download.
- lsp/manager.ts: process-wide LspManager owning client lifecycles, keyed
by root+serverID, lazy spawn + reuse + graceful shutdown.
- lsp/language.ts: extension->languageId map incl. .luau -> "luau".
- lsp/diagnostic.ts: error-only <diagnostics> block formatting (1-based).
- tools/lsp.ts: on-demand 'lsp' tool (1-based coords -> 0-based wire).
- write-file.ts: optional onAfterWrite hook for diagnostics-on-write.
- config schema: validate [lsp] block; DispatchConfig.lsp + LspServerConfig.
API (@dispatch/api):
- AgentManager owns one LspManager; per-working-directory server cache
cleared on config reload; diagnostics appended to write_file results;
'lsp' tool gated by new perm_lsp setting; shutdownAll on destroy().
Config:
- dispatch.toml: documented, commented [lsp.luau-lsp] Roblox example.
Tests: fake-lsp-server fixture + client/manager/server/diagnostic/schema/
tool/write-hook suites, plus an opt-in real-binary luau-lsp smoke test
(auto-skipped when luau-lsp is absent). 652 pass; biome + 3 typechecks green.
|
|
- Add phosphor-svelte ^3.1.0 to frontend deps.
- Wire phosphor-svelte/vite (sveltePhosphorOptimize) as a fallback so
stray named imports still tree-shake correctly. Per-icon imports like
'phosphor-svelte/lib/ListIcon' remain the preferred pattern; see the
comment in vite.config.ts.
- Replace the 'Sidebar' text in Header.svelte with a Phosphor List icon
styled as a square daisyUI button (btn btn-square btn-sm btn-neutral)
with aria-label preserved for screen readers.
|
|
reasoning round-trip + max-thinking budget audit
Migrates the LLM stack from [email protected] + @ai-sdk/[email protected] +
@ai-sdk/[email protected] to [email protected] + @ai-sdk/[email protected]
+ @ai-sdk/[email protected]. Full design in plan-v6-upgrade.md;
two rounds of Gemini code review captured in report.md.
Motivation: the recurring 'reasoning-signature without reasoning' error
on Claude Opus 4.7 was a v4 SDK artefact — @ai-sdk/[email protected] emitted
Anthropic signature_delta as a separate stream chunk that orphaned when
the model produced a signed-but-empty thinking block, and our chunk
store had no signature field so the round-trip back to Anthropic was
rejected on the next turn. In v6, signatures arrive inside
providerMetadata on the reasoning-end event, and the orphan-signature
class of bug is gone at the SDK level.
Core changes:
• ThinkingChunk gains optional metadata?: Record<string, unknown>
(the v6 providerMetadata blob). A non-undefined metadata 'seals'
the chunk: subsequent reasoning-delta opens a new chunk rather
than extending the sealed one.
• AgentEvent gains { type: 'reasoning-end'; metadata? } (replaces
the v4 reasoning-signature variant).
• toModelMessages (replaces toCoreMessages):
- returns ModelMessage[] (was CoreMessage[])
- thinking → { type: 'reasoning', text, providerOptions: metadata }
- tool-batch entries → { type: 'tool-call', input } (was 'args')
- tool results → { output: { type: 'text', value } } ToolResultOutput
• Claude OAuth uses createAnthropic({ authToken }) natively — no more
custom-fetch x-api-key → Bearer swap.
• rewriteBodyForOpus47 deleted — Opus 4.7 adaptive thinking is native
via providerOptions.anthropic.thinking = { type: 'adaptive' }.
• V1 middleware → V3 (specificationVersion: 'v3').
• v4-era normalizeMessages openai-compatible middleware deleted; the
v6 openai-compatible provider extracts reasoning_content natively
from { type: 'reasoning' } content parts.
• applyAnthropicStructuralNormalisations (mirrors opencode
provider/transform.ts:53-148): drops empty text/reasoning parts,
scrubs non-[a-zA-Z0-9_-] toolCallIds, splits [tool-call, non-tool]
assistant turns (Anthropic rejects tool_use followed by text).
• applyOpenAICompatibleReasoningNormalisation (mirrors opencode
transform.ts:217-249): lifts reasoning text into
providerOptions.openaiCompatible.reasoning_content (always, even
empty). Solves DeepSeek 'The reasoning_content in the thinking
mode must be passed back' — the v6 SDK skips emitting
reasoning_content when text is empty (dist/index.mjs:245), but
DeepSeek requires the field present once thinking was used.
• Tools: tool({ inputSchema: jsonSchema(zodToJsonSchema(...)) })
(was parameters: ZodSchema). AI SDK tools have no execute
callback — the agent runs tools manually for permission prompts
and shell-output streaming. New dep: zod-to-json-schema@^3.25.2.
• fullStream event loop rewritten for v6 event shape: text-delta
(text not textDelta), reasoning-start/delta/end, tool-input-*,
tool-call (input not args), tool-result, tool-error (new), abort
(new), start-step/finish-step, finish.
Max-thinking audit (matches opencode transform.ts:642-671 budgets):
• Claude enabled-thinking max budget 16000 → 31999 (Anthropic ceiling)
• Claude enabled-thinking high budget 10000 → 16000
• maxOutputTokens 'budget + 8000' → fixed 32000 (matches opencode's
OUTPUT_TOKEN_MAX; model self-allocates thinking vs response within)
• Opus 4.7 adaptive thinking gains display: 'summarized' and sibling
effort field (without these, thinking content is hidden by Anthropic
and the model barely thinks).
Frontend mirrors:
• types.ts — ThinkingChunk.metadata?, AgentEvent reasoning-end
• tabs.svelte.ts — routes reasoning-end through applyChunkEvent
• ChatMessage.svelte — hides empty thinking chunks; hides the entire
assistant bubble when no chunk has renderable content
Gemini-review-driven fixes:
• tool-error and abort stream events now surface as error chunks
(were silently ignored)
• toolCallId scrubbing pass (opencode transform.ts:96-122 parity)
• Empty-reasoning-cull explicit test coverage for both Anthropic
structural normalisation and DeepSeek path
Test counts (223 tests across 3 packages, all green):
• tests/chunks/append.test.ts: 44 (was 38) — reasoning-end sealing,
orphan walk-back, multi-block interleaving
• tests/agent/agent.test.ts: 24 (was 5) — exhaustive v6 event
mappings, structural normalisations, signature/reasoning_content
round-trip, tool-error/abort branches, DeepSeek scenario, empty
reasoning edge case
• tests/llm/provider.test.ts: 9 (was 22) — dropped 13 obsolete v4
middleware tests; new minimal tests confirm no middleware wrapping
on default openai-compat path and that createAnthropic gets
authToken vs apiKey correctly for OAuth vs api-key flows
• tests/tools/registry.test.ts: 10 (was 4) — v6 tool() contract
(inputSchema, no execute, JSON Schema for nested zod)
• packages/api/tests/agent-manager.test.ts: 12 (was 7) — mock Agent
emits v6 reasoning events; reasoning-end broadcast + ordering
• packages/frontend/tests/chat-store.test.ts: 35 (was 32) —
reasoning-end flow through Svelte $state store
typecheck clean (tsc --noEmit on core + api, svelte-check on frontend),
biome clean across 124 files.
|
|
batching, error/system chunks
|
|
SDK compat
- Implement Anthropic prompt caching: first system message + last 2 non-system messages get cache_control: ephemeral, mirroring OpenCode's applyCaching strategy. Move system prompt inline into messages array so providerOptions can attach.
- Add opencode-anthropic provider variant routing MiniMax/Qwen models through the /messages endpoint with x-api-key auth, distinct from the Claude OAuth flow's Bearer auth and Claude Code mimicry.
- Split isAnthropic into isClaudeOAuth (billing header, mcp_ tool prefix, thinking config) and usesAnthropicSDK (cache markers) so non-OAuth Anthropic-format gateways get the right treatment.
- Pin @ai-sdk/anthropic to ^1.2.12: v3 returns LanguageModelV3-spec models that ai v4's streamText rejects at runtime ('AI SDK 4 only supports models that implement specification version v1'). Drop unnecessary V1 casts.
- Restore Opus 4.7 extended thinking by rewriting the outgoing /messages body in the Claude OAuth fetch interceptor: inject thinking: { type: 'adaptive' } (v1 SDK can't emit it), strip temperature/top_p/top_k (Anthropic rejects them with thinking enabled). Gated on max_tokens > 4096 so effort=none still works.
- Bump MAX_STEPS from 10 to 50 to align with AI SDK's stepCountIs(20) default and reduce mid-task halts.
- Fix pre-existing typecheck errors in agent-manager.ts (entry/nextEntry narrowing), app.ts (agentModels body field), KeyUsage.svelte (m guards), and a TS2742 in provider.ts via explicit ModelFactory return type.
- buildFallbackSequence now always returns at least one entry so processMessage runs the agent loop even without keyId/modelId (fixes 4 broken agent-manager tests).
|
|
filter
- Added Google (Gemini) as a provider: add-key UI, env var resolution via resolveApiKey, usage tracking via native models endpoint + gemini.google.com cookie scraping
- @ai-sdk/anthropic upgraded to v3 (adaptive thinking support) with LanguageModelV1 cast for ai v4 compat
- Claude Opus 4.7 uses adaptive thinking (type: adaptive); all other models keep explicit budget tokens
- Model selector modal: search filter with space matching dash/underscore
- Copy button: all tool results truncated at 300 chars
- Sidebar layout fix: Claude Reset panel removed from flex-1 fill to prevent overlap
|
|
and Windows exe build
- Add Electron wrapper (main.cjs, preload.cjs) for desktop frontend
- Add systemd service unit, env config, and sysusers for backend API
- Add PKGBUILD and .install for Arch package (makepkg -si)
- Add desktop entry, SVG/PNG icon, and wrapper scripts
- Add bin/build-pkg, bin/install-pkg, bin/windows-pkg scripts
- Make API port configurable via PORT env var (default 3000, prod 18390)
- Add electron-builder config for Windows exe cross-compilation
- Set vite base to './' for Electron file:// loading
|
|
effort, and dynamic model listing
|
|
- Config system: TOML-based dispatch.toml with hot-reload via chokidar
- Model/key resolution: tag-based model selection, key fallback chains
- Skills system: directory loader with TOML frontmatter, agent mappings
- Task list tool: add/update/list/get operations with WebSocket events
- API routes: GET /config, /skills, /skills/:name, /models, /models/resolve
- Frontend: sidebar with model status, task list, config viewer, skills browser, permission log
- Sliding sidebar animation using CSS transitions (not Svelte transitions)
|
|
|
|
|
|
Permission engine:
- Rule-based engine: wildcard matching, last-match-wins, reject cascade
- PermissionService with pending/approved state, PermissionChecker interface
- dispatch.yaml config loader with per-permission pattern rules
Shell tool:
- run_shell tool with child_process spawn, timeout, streaming output
- Tree-sitter static analysis (web-tree-sitter + tree-sitter-bash WASM)
- BashArity command normalization for 'always allow' patterns
- FILE_COMMANDS set: rm, cp, mv, mkdir, ls, find, grep, cat, etc.
Agent loop refactored:
- Removed maxSteps, manual step loop with tool execution
- Permission checks on shell commands (external_directory only)
- Permission checks on file tools outside workspace boundary
- Symlink bypass fix (realpathSync), .. false positive fix
- Shell output streaming via Promise.race + setImmediate polling
API layer:
- PermissionManager wraps PermissionService, broadcasts via WebSocket
- WebSocket handles permission-reply messages from frontend
- Config loaded from dispatch.yaml, converted to ruleset
Frontend:
- Permission prompt modal (native dialog, focus trap, ARIA)
- Always-allow confirmation flow with pattern preview
- Shell output display (live streaming + final parsed result)
- Permission log panel (fixed bottom-right overlay)
- Exit code badge (green 0, red non-zero)
134 tests, typecheck clean on all 3 packages
|
