mirror of
https://github.com/deepseek-ai/deepseek-harness
synced 2026-08-15 21:04:50 +00:00
Merge origin/master into codex/truncated-design
This commit is contained in:
Dudu-0223
@@ -1,6 +1,6 @@
|
||||
# Bash Executor
|
||||
|
||||
The bash execution seam — the canonical [capability seam](../rfc/implemented/architecture/2026-06-13-capability-seams.md) example, split across three packages: interface ([dsh-bash](../../packages/bash/bash), `ctx.bash`), implementation ([dsh-bash-local](../../packages/bash/bash-local), local subprocesses), and consumer ([dsh-tool-bash](../../packages/bash/tool-bash), the `bash`/`bash_output`/`bash_kill` tool schemas). Bash is **one optional capability**, not part of the agent-loop spine — so its vocabulary lives here, not in [core.md](core.md). A sandboxed, containerized, or remote backend is a sibling package implementing the same interface.
|
||||
The bash execution seam is split across interface ([dsh-bash](../../packages/bash/bash), `ctx.bash`), implementations ([dsh-bash-local](../../packages/bash/bash-local) and [dsh-bash-sandbox](../../packages/bash/bash-sandbox)), and consumer ([dsh-tool-bash](../../packages/bash/tool-bash), the `bash` schema). Generic background-task ids, ownership, and controls live in [tasks.md](tasks.md); this seam returns a task-free process handle.
|
||||
|
||||
Source: [`packages/bash/bash/src/types.ts`](../../packages/bash/bash/src/types.ts)
|
||||
|
||||
@@ -42,15 +42,6 @@ interface BashExecRequest {
|
||||
* uses shell syntax like `FOO=bar cmd`).
|
||||
*/
|
||||
env?: Record<string, string> | undefined
|
||||
/**
|
||||
* Opaque OWNER token for a background task — the consumer's isolation key
|
||||
* (the tool layer passes the owning agent's `session.header.id`). The
|
||||
* executor stores it on the task and exposes it via {@link BashExecutor.ownerOf};
|
||||
* the executor itself NEVER interprets it (no access policy lives in the
|
||||
* seam — that is the consumer's job). Absent for foreground runs and for an
|
||||
* ownerless background start (a non-agent caller).
|
||||
*/
|
||||
owner?: OwnerToken | undefined
|
||||
/**
|
||||
* Explicit per-call sandbox-policy input, overriding the executor's
|
||||
* configured default mode for THIS call. Never a silent default: a
|
||||
@@ -82,10 +73,8 @@ interface BashExecSpec {
|
||||
signal?: AbortSignal | undefined
|
||||
/**
|
||||
* Bytes to write to the command's stdin (then close it), carried through
|
||||
* verbatim from {@link BashExecRequest.stdin}. OPTIONAL on the resolved spec
|
||||
* (unlike `owner`): it has no config default, so a missing one means "no
|
||||
* stdin" — the safe, ordinary case — not a silent footgun, so it stays a
|
||||
* plain optional rather than required-but-nullable (see the request field).
|
||||
* verbatim from {@link BashExecRequest.stdin}. It has no config default, so
|
||||
* a missing value means "no stdin" and remains an ordinary optional.
|
||||
*/
|
||||
stdin?: string | undefined
|
||||
/**
|
||||
@@ -97,19 +86,10 @@ interface BashExecSpec {
|
||||
*/
|
||||
env?: Record<string, string> | undefined
|
||||
/**
|
||||
* Opaque owner token, REQUIRED-but-nullable (mirrors `workdir`/`timeoutMs`
|
||||
* being required on the resolved spec): {@link BashExecutor.resolve} carries
|
||||
* the request's `owner` through, defaulting a missing one to `undefined`. A
|
||||
* required field makes a forgotten owner a VISIBLE `undefined` rather than a
|
||||
* silently-absent property that yields an unowned (cross-session-readable)
|
||||
* task. `start()` stores it; `run()` (foreground) ignores it.
|
||||
*/
|
||||
owner: OwnerToken | undefined
|
||||
/**
|
||||
* The sandbox mode this call executes under, REQUIRED-but-nullable for the
|
||||
* same visibility reason as `owner`. A sandboxing executor's `resolve()`
|
||||
* stamps the effective mode (the request's explicit override, else its
|
||||
* configured default) so `run()`/`start()` read the spec, never the config;
|
||||
* The sandbox mode this call executes under, required-but-nullable so every
|
||||
* resolved spec states its policy. A sandboxing executor's `resolve()` stamps
|
||||
* the effective mode (the request's explicit override, else its configured
|
||||
* default) so `run()`/`start()` read the spec, never the config;
|
||||
* a non-sandboxing executor carries the request value through verbatim and
|
||||
* ignores it (`undefined` under such an executor means what its README says:
|
||||
* unconfined execution).
|
||||
@@ -118,13 +98,9 @@ interface BashExecSpec {
|
||||
}
|
||||
```
|
||||
|
||||
The `owner` token is the isolation key: the executor stores it but never interprets it (access policy is the consumer's job), so a background task started by one agent isn't readable cross-session. A required-but-nullable field makes a forgotten owner a visible `undefined` rather than a silently-unowned task.
|
||||
`stdin` and `env` are trusted in-process plugin inputs and are not exposed by `dsh-tool-bash`. The local executor scrubs ambient credentials before merging explicit caller-supplied env. See [the bash-stdin-env RFC](../rfc/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md).
|
||||
|
||||
`stdin` and `env` are set by in-process plugins (the hooks bridges, native plugins) to feed a hook command its JSON payload on stdin and its `CLAUDE_PROJECT_DIR`/`CLAUDE_PLUGIN_ROOT` env. The model-facing `dsh-tool-bash` tool does not expose them as parameters — its request is built from `command`/`workdir`/`timeoutMs`/`signal`/`owner` only — because a model already has equivalent power through shell syntax (`FOO=bar cmd`, a heredoc), so duplicating them as tool params would be redundant. This is NOT a security boundary: the credential scrub in `dsh-bash-local` is what stops the harness's ambient secrets reaching a spawned command, and it works regardless of these fields (a model cannot read a value the scrub removed, and tool-call args are static JSON, never shell-evaluated). A guard test asserts the tool doesn't forward model `env`/`stdin`/`stdoutMaxBytes` — to catch a future `...args` spread, not to defend a trust wall. `env` is merged AFTER the scrub so an explicit caller entry (a value it already holds) wins even on a credential-shaped name. See [the bash-stdin-env RFC](../rfc/implemented/architecture/2026-06-30-bash-stdin-env-trusted-plugin-surface.md).
|
||||
|
||||
`stdoutMaxBytes` is also trusted-plugin-only. It lets a foreground consumer ask the executor to retain complete stdout up to a bounded parser budget without changing stderr, background tasks, or the model-facing bash tool's ordinary `maxOutputBytes` behavior.
|
||||
|
||||
Both ids the seam handles are [branded](core.md) (zero-cost `string` brands, the same machinery as `SessionId`/`AgentId`): `BashTaskId` (a tracked background task, generated `bash-N` by the local executor) and `OwnerToken` (the opaque isolation key). `OwnerToken` is deliberately a DISTINCT brand from `SessionId`, not an alias: the bash seam is a capability seam that must not know what an owner token *means*, so it never imports `dsh-session`'s vocabulary — the `dsh-tool-bash` consumer is the single boundary that casts the owning agent's `SessionId` into an `OwnerToken`. Branding both stops a raw `string` (or a `BashTaskId` where an `OwnerToken` is expected, or vice versa) from slipping through the type checker on the model-facing `task_id` path.
|
||||
`stdoutMaxBytes` is also trusted-plugin-only. It lets a foreground consumer request complete stdout up to a bounded parser budget without changing stderr, background tasks, or the model-facing bash tool's ordinary output cap.
|
||||
|
||||
## Foreground runs: `BashRunResult`
|
||||
|
||||
@@ -168,9 +144,9 @@ interface CollectedOutput {
|
||||
|
||||
## File sandbox: `BashSandboxInfo`
|
||||
|
||||
A sandbox-consuming executor (`dsh-bash-sandbox`) exposes its configured fallback through `BashExecutor.sandboxMode`. The tool layer folds each agent session's durable `bash/sandbox-mode` override, stamps the effective mode onto the request, states it in the per-agent prompt, and may replace it for one user-approved strictly wider call. The mode/enforcement vocabulary is owned and cataloged by the [`@deepseek-ai/dsh-sandbox` seam](sandbox.md), whose provider wraps the executor's argv; modes govern FILE effects only, not network or process visibility.
|
||||
A sandbox-consuming executor exposes its configured fallback through `BashExecutor.sandboxMode`. The tool layer folds each session's durable `bash/sandbox-mode` override and may replace it for one user-approved strictly wider call. The mode/enforcement vocabulary is owned by the [`@deepseek-ai/dsh-sandbox` seam](sandbox.md); modes govern file effects only.
|
||||
|
||||
A sandboxed run always reports the facts it executed under on `BashRunResult.sandbox`: `denied` is the executor's conservative classification of a failure as sandbox-caused (a failed exit whose stderr carries a filesystem-permission signature — never a clean exit or a signal kill), read from the collected stderr tail; `enforcement` reports how completely the selected backend governs the mode's file effects (`SandboxEnforcement = 'full' | 'partial'` — `partial` when an older Landlock ABI governs only a subset of the requested accesses; absent under `danger-full-access`, where nothing is confined); `runnerFailed` marks the opposite of a denial — the sandbox RUNNER itself failed and the command never ran (stamped only on settled background tasks; a foreground run surfaces the same condition as the thrown `SANDBOX_UNAVAILABLE` error):
|
||||
A sandboxed run reports its mode, conservative denial classification, and enforcement completeness. `runnerFailed` marks a sandbox runner failure before the command ran; foreground execution throws `SANDBOX_UNAVAILABLE`, while a settled background process has only its facts channel.
|
||||
|
||||
```ts type-equiv
|
||||
interface BashSandboxInfo {
|
||||
@@ -207,41 +183,42 @@ interface BashSandboxInfo {
|
||||
}
|
||||
```
|
||||
|
||||
One more piece completes the vocabulary: the `SANDBOX_UNAVAILABLE` error code (owned by the [sandbox seam](sandbox.md)) is what the `ctx.sandbox` provider throws — and the executor propagates — when a confined mode has no usable backend. A selected runner refusing its profile reaches the same fail-closed foreground error; a settled background task records `runnerFailed`. The model sees the current effective mode in the prompt, receives denial/runner facts in results, and can request a one-shot strictly wider retry through `sandbox_permissions` plus `justification`; `ctx.approval` must grant that exact call before anything executes. The complete policy and switching design is the [sandbox RFC](../rfc/implemented/feature/2026-07-06-sandbox.md).
|
||||
One more piece completes the vocabulary: the `SANDBOX_UNAVAILABLE` error code (owned by the [sandbox seam](sandbox.md)) is what the `ctx.sandbox` provider throws — and the executor propagates — when a confined mode has no usable backend. A selected runner refusing its profile reaches the same fail-closed foreground error; a settled background task records `runnerFailed`. The model receives denial/runner facts in results, learns the effective mode only when a denial marker names it, and can request a one-shot strictly wider retry through `sandbox_permissions` plus `justification`; `ctx.approval` must grant that exact call before anything executes. The complete policy and switching design is the [sandbox RFC](../rfc/implemented/feature/2026-07-06-sandbox.md).
|
||||
|
||||
## Background tasks: `BashTask`
|
||||
## Background processes: `BashProcess`
|
||||
|
||||
A long-running command started with `start()` is tracked as a `BashTask`. `BashTaskStatus` is `'running' | 'completed' | 'killed'`; `done` resolves when the underlying process closes and never rejects. A sandboxing executor stamps `sandbox` once the task settles — classification runs against the settled task's collected stderr — so the field is absent while running and under an unsandboxed executor.
|
||||
`start()` returns a handle with no id or owner. `dsh-tool-bash` adapts it into `ctx.tasks.start()` hooks; the generic runtime then owns task identity and lifecycle. `done` resolves when the process closes and never rejects, reads remain valid after settlement, and sandbox facts are stamped before `done` resolves.
|
||||
|
||||
```ts type-equiv
|
||||
interface BashTask {
|
||||
readonly id: BashTaskId
|
||||
readonly command: string
|
||||
status: BashTaskStatus
|
||||
interface BashProcess {
|
||||
/** Process lifecycle state (settled exactly once). */
|
||||
status: BashProcessStatus
|
||||
/** Exit code once finished (null = killed by signal / still running). */
|
||||
exitCode: number | null
|
||||
/** Terminating signal name, when signal-killed. */
|
||||
signal: NodeJS.Signals | null
|
||||
/** Resolves when the underlying process closes (never rejects). */
|
||||
/** Resolves when the underlying process closes (never rejects — a spawn failure settles as `killed` with the error on stderr). */
|
||||
readonly done: Promise<void>
|
||||
/**
|
||||
* Sandbox facts for this task's execution, stamped by a sandboxing executor
|
||||
* once the task settles and BEFORE completion listeners are notified — an
|
||||
* `onTaskDone` consumer and a `done` awaiter both see it. Denial
|
||||
* classification runs against the settled task's collected stderr, so the
|
||||
* field cannot exist earlier: absent while the task is running and under an
|
||||
* executor that does not sandbox. See {@link BashSandboxInfo} for the
|
||||
* `denied` semantics.
|
||||
*/
|
||||
/** Sandbox facts, stamped once a confined process settles. */
|
||||
sandbox?: BashSandboxInfo
|
||||
/**
|
||||
* Read output produced since the previous read (consuming — consecutive
|
||||
* reads never re-deliver). Reads that lost data flag `lossy` and point at
|
||||
* full-stream spill files when available.
|
||||
*/
|
||||
readOutput(): BashProcessRead
|
||||
/**
|
||||
* Kill the process group. Returns false when it had already finished
|
||||
* (no-op); idempotent.
|
||||
*/
|
||||
kill(): boolean
|
||||
}
|
||||
```
|
||||
|
||||
`readOutput()` returns an incremental `BashTaskRead` — the output produced since the previous read, with a `lossy` flag when truncation dropped unread bytes:
|
||||
`readOutput()` returns the incremental delta and spill recovery facts:
|
||||
|
||||
```ts type-equiv
|
||||
interface BashTaskRead {
|
||||
task: BashTask
|
||||
interface BashProcessRead {
|
||||
/** Output produced since the previous read (stderr in a marked section). */
|
||||
delta: string
|
||||
/** True when truncation dropped unread bytes the delta cannot include. */
|
||||
@@ -255,4 +232,4 @@ interface BashTaskRead {
|
||||
|
||||
## The service
|
||||
|
||||
`BashExecutor` (`ctx.bash`, abstract — defined in [`packages/bash/bash/src/index.ts`](../../packages/bash/bash/src/index.ts)) mirrors the `LlmService`/`LlmAdapter` split: `resolve` (request → spec), `run` (foreground), `start` (background), `get`/`ownerOf`/`list`/`readOutput`/`kill`, and `onTaskDone` (a `BashTaskListener` completion callback). Spawned commands get a **scrubbed env** (dropping `*KEY*`/`*SECRET*`/`*TOKEN*`) and spill files use a private 0700 dir with random names and owner-only opens — model output never gets the ambient environment or a predictable path. The implementation that provides all this is `dsh-bash-local`; the model-facing `bash`/`bash_output`/`bash_kill` schemas that call it are in `dsh-tool-bash` (and present as terminals via the [tool-presentation vocabulary](tools.md#tool-presentation-ui-vocabulary)).
|
||||
`BashExecutor` owns `resolve`, foreground `run`, background-process `start`, and the `sandboxMode` capability fact. `dsh-bash-local` owns process groups, timeout/abort handling, bounded collectors, spill files, credential scrubbing, and disposal quiescence. `dsh-tool-bash` owns model-facing rendering and adapts background handles into the [generic task runtime](tasks.md).
|
||||
|
||||
@@ -39,8 +39,8 @@ interface CodeRunResult {
|
||||
* or value-less run leaves this absent.
|
||||
*/
|
||||
value?: unknown
|
||||
/** Everything the program emitted, in order (capped by the implementation). */
|
||||
logs: CodeLogEntry[]
|
||||
/** Text the program emitted, in order (capped by the implementation). */
|
||||
logs: string[]
|
||||
/** Present iff the run failed; see {@link CodeRunFailure} for the taxonomy. */
|
||||
error?: CodeRunFailure
|
||||
}
|
||||
@@ -65,18 +65,7 @@ type CodeBindingFunction = (args: unknown) => Promise<unknown>
|
||||
|
||||
## Captured output and the failure taxonomy
|
||||
|
||||
Logs arrive in emission order, attributed to their channel (the runtime's `console` shim, or stray writes to the underlying streams):
|
||||
|
||||
```ts type-equiv
|
||||
interface CodeLogEntry {
|
||||
/** Which channel produced the text. */
|
||||
source: 'console' | 'stdout' | 'stderr'
|
||||
/** The console method used; present only when `source` is `'console'`. */
|
||||
level?: 'log' | 'info' | 'warn' | 'error' | 'debug'
|
||||
/** The captured text (possibly truncated by the implementation's caps, marked in-band). */
|
||||
text: string
|
||||
}
|
||||
```
|
||||
Logs are plain strings in emission order. The runtime captures the program's console and stream output, but channel and console-method metadata are not part of the seam because consumers render only the text. Implementations cap the aggregate output and mark truncation in-band.
|
||||
|
||||
Failure kinds are **orthogonal outcomes reported independently** (per [defensive-patterns](../defensive-patterns.md)): a budget expiry is not an exception, an abort is not a timeout, and a substrate death (e.g. OOM) is neither:
|
||||
|
||||
|
||||
@@ -50,6 +50,6 @@ interface CompactionResult {
|
||||
|
||||
## The service
|
||||
|
||||
`CompactService` (`ctx.compact`, abstract — defined in [`packages/compact/compact/src/index.ts`](../../packages/compact/compact/src/index.ts)) declares two abstract methods: `compactIfNeeded(agent, fullSystemPrompt, sessionPrefix, signal)` checks token pressure and compacts an older range if the history is too large (returning `null` when nothing needs it), and `compactRegion(session, start, end, agent, signal?)` forcibly summarizes surface nodes `[start, end]` into a single replacement node. `compactIfNeeded`'s parameters are all required — the loop's `agent/pre-step` checkpoint supplies the agent, the assembled `fullSystemPrompt`, the instance's composed `sessionPrefix` (request-only messages the derived history omits, so the pressure estimate must count them), and the turn `signal`. A backend summarizing via `ctx.llm.stream()` must forward `signal` into the call's `GenerateOptions.signal`, so an abort or dispose tears down the in-flight summarization. The entire strategy — token estimation, retention policy, event sequencing, summarization — is a HOW decision owned by the implementation.
|
||||
`CompactService` exposes `compactIfNeeded(...)` for pressure-triggered compaction, returning `null` when no compaction is needed, and `compactRegion(...)` for an explicit inclusive surface range. The pre-step caller supplies the agent, full prompt, session prefix, and abort signal; implementations must forward that signal to summarization. Estimation, retention, event sequencing, and summarization remain backend policy.
|
||||
|
||||
Auto-compaction runs on the serial `agent/pre-step` loop seam (fired once per step, after `turn/start` and BEFORE the step opens and its request history is derived), not the `agent/request` waterfall: compaction mutates the session surface in place — with its log-only `compact/*` records landing cleanly outside any step — and the loop derives the request from the already-compacted surface. Retention is turn-agnostic — the only structural guard is tool-pairing balance (a compacted region's edges are balanced cuts on the surface, so it never splits a step's tool-calls from their results), so a single runaway turn that alone exceeds the window compacts its own early closed steps rather than being retained verbatim. The backend that ships this (`dsh-compact-basic`) documents the retention walk, summary shrink validation, bounded re-compaction, and the crash/recoverable failure taxonomy.
|
||||
Auto-compaction runs at serial `agent/pre-step`, before the step and request derivation, so it can replace surface nodes while keeping trace events outside the step. Region boundaries preserve tool-call/result pairing but do not preserve whole turns, allowing early closed steps of one oversized turn to compact. `dsh-compact-basic` owns the retention and failure details.
|
||||
|
||||
@@ -24,14 +24,14 @@ Everything else is documented on a **sub-page**, not here. The rule that draws t
|
||||
| [tools.md](tools.md) | `ToolDefinition` full fields, the schema DSL, `ToolExecution`/`ToolResult`, tool-presentation UI types, and the guarded execution pipeline |
|
||||
| [user-interaction.md](user-interaction.md) | the UI-backed human question/answer seam: `AskUserQuestionRequest`, answer/options vocabulary, provider API, error taxonomy |
|
||||
| [approval.md](approval.md) | the one-shot user-approval seam: `ApprovalRequest`, `ApprovalOutcome`, per-session policy, audit and answerer contracts |
|
||||
| [bash.md](bash.md) | the bash executor seam: `BashExecRequest`/`Spec`, `BashRunResult`, background `BashTask`s |
|
||||
| [bash.md](bash.md) | the bash executor seam: `BashExecRequest`/`Spec`, `BashRunResult`, background `BashProcess` handles |
|
||||
| [sandbox.md](sandbox.md) | the process-confinement seam: file-effect modes, `SandboxPolicy`, `ConfinedArgv`, enforcement and fail-closed errors |
|
||||
| [code-runtime.md](code-runtime.md) | the code-execution seam: `CodeRunRequest`/`Result`, binding namespaces, captured logs, the `CodeRunFailure` taxonomy |
|
||||
| [filesystem.md](filesystem.md) | the filesystem seam: `FsTarget`, read/write/edit outcomes, observed-file state, `FsErrorCode` |
|
||||
| [skills.md](skills.md) | the skill service: discovery priority, `SkillSummary`/`SkillDefinition`, session-prefix catalog, model-facing `skill` loading |
|
||||
| [compaction.md](compaction.md) | the compaction seam: the `compact/*` session events, `CompactionResult`, the `CompactService` interface |
|
||||
| [subagent.md](subagent.md) | the subagent seam: the named-provider registry, `SubagentStartRequest`/`Result`/`Run`, the start-time-vs-runtime capability split |
|
||||
| [web.md](web.md) | the web access seam: `WebSearchRequest`/`Result`, `WebFetchRequest`/`Result`, `WebFetchBody`, provider/capability status, `WebError` |
|
||||
| [web.md](web.md) | the web access seam: `WebSearchRequest`/`Result`, `WebFetchRequest`/`Result`, `WebFetchBody`, provider availability, `WebError` |
|
||||
| [spill.md](spill.md) | the spill storage seam: `SaveTextSpill`, `SpillOwner`/`SpillSource`, `SpillRef`, the branded `SpillLocator` |
|
||||
| [workflow.md](workflow.md) | the workflow seam: `WorkflowStartRequest`, `WorkflowMeta`, `WorkflowRun`/`Result`, the `workflow/*` event payloads, `WorkflowError` fatality |
|
||||
|
||||
@@ -77,7 +77,7 @@ Two large discriminated unions are the ones consumers `switch` over most: **`Str
|
||||
|
||||
IDs that cross package boundaries are **branded** — structurally strings, but non-interchangeable at the type level (an `AgentId` can't be passed where a `CallId` is expected). Construction goes through a per-type factory; comparison, logging, and JSON behave as ordinary strings.
|
||||
|
||||
The `Branded<B>` primitive lives in its own type-only package, [dsh-brand](../../packages/util/brand) (no runtime code, no harness-package dependency), so any package can brand the ids it owns without depending on an unrelated capability package (e.g. dsh-bash brands `BashTaskId`/`OwnerToken` via dsh-brand alone, never pulling in dsh-llm).
|
||||
The `Branded<B>` primitive lives in its own type-only package, [dsh-brand](../../packages/util/brand) (no runtime code, no harness-package dependency), so any package can brand the ids it owns without depending on an unrelated capability package.
|
||||
|
||||
Source: [`packages/util/brand/src/index.ts`](../../packages/util/brand/src/index.ts)
|
||||
|
||||
@@ -85,7 +85,7 @@ Source: [`packages/util/brand/src/index.ts`](../../packages/util/brand/src/index
|
||||
type Branded<B extends string> = string & { readonly [BRAND]: B }
|
||||
```
|
||||
|
||||
The three core IDs: `CallId` (correlates a tool call with its result; dsh-llm), `SessionId` (dsh-session), `AgentId` (dsh-agent). Each is `Branded<'CallId'>` etc. plus a same-named factory function. Capability seams brand their own ids too — see `BashTaskId`/`OwnerToken` in [bash.md](bash.md).
|
||||
The three core IDs are `CallId`, `SessionId`, and `AgentId`. Capability packages brand their own ids too, such as `TaskId` in [tasks.md](tasks.md).
|
||||
|
||||
## Content blocks and messages
|
||||
|
||||
@@ -158,17 +158,8 @@ interface GenerateOptions {
|
||||
stop?: string[]
|
||||
signal?: AbortSignal
|
||||
/**
|
||||
* The id of the session this request belongs to — stamped by the agent loop
|
||||
* from `agent.session.id`. Adapters ignore it; it lets an `llm/stream` listener
|
||||
* route a call by WHICH session issued it (the replay adapter keys its per-call
|
||||
* cursor by session, so a parent and its in-process subagent — each with its
|
||||
* own session on one context — replay from their own recorded scripts).
|
||||
*
|
||||
* Typed as `Branded<'SessionId'>` rather than importing `SessionId` from
|
||||
* `dsh-session`: that package imports `Message` from here, so importing its
|
||||
* `SessionId` back would cycle. `SessionId` IS `Branded<'SessionId'>`, so a
|
||||
* real session id assigns with no cast. (A future ids package could own the
|
||||
* brand and dissolve this note.)
|
||||
* Session identity stamped by the loop for listener routing. Adapters ignore
|
||||
* it; replay uses it to keep concurrent parent and child cursors independent.
|
||||
*/
|
||||
sessionId?: Branded<'SessionId'>
|
||||
}
|
||||
@@ -203,7 +194,9 @@ The model-facing `ToolSchema` is the wire shape; the registered `ToolDefinition`
|
||||
|
||||
### The request envelope: `LlmCallConfig` and the logged header
|
||||
|
||||
Requests are built by the loop, not shaped per call: the non-history half of a request — the `EpochHeader`: this call configuration plus the rendered system prompt, the tool schemas in the authoritative returned assembly order (initially canonicalized by dsh-system-prompt's `toolOrder` config, or lexicographically when unset), and the session prefix — is logged session state (`request/header` snapshot and delta events, [session.md](session.md#the-request-header-events-requestheader-and-requestheader-delta)), so every conversation request is a pure function of the session log ([reconstructability RFC](../rfc/implemented/architecture/2026-07-05-reconstructable-requests.md)). The `agent/request` waterfall receives a frozen `LlmCallConfig` seed and a listener returns a replacement to switch model or sampling; the `agent/session-prefix` waterfall — fired once per loop instance — composes the request-only messages fronting the derived history (recorded as the header's `messagePrefix`) — the loop logs whatever the request actually uses. Loop-built requests arrive at `llm/stream` deep-frozen; mutation throws.
|
||||
The loop builds each request from logged state. `EpochHeader` records call config, rendered prompt, authoritative returned tool order (configured by `toolOrder`, or lexicographic when unset), and session prefix through `request/header` snapshots and deltas. Together with derived history, this makes the request reconstructable from the session log. See [session.md](session.md#the-request-header-events-requestheader-and-requestheader-delta) and the [reconstructability RFC](../rfc/implemented/architecture/2026-07-05-reconstructable-requests.md).
|
||||
|
||||
`agent/request` receives a frozen call-config seed and may return a replacement. `agent/session-prefix` composes request-only prefix messages once per loop instance, and the header records the exact result used. Requests reaching `llm/stream` are deep-frozen, so mutation throws.
|
||||
|
||||
On the wire, a loop-built request reads in this order: the `system` slot (the rendered prompt assembly) → `messagePrefix` (the frozen session prefix) → the derived history — the boundary snapshot, whose tail is the newest `user/message` on a turn's first step and the previous step's tool results on later steps. The prefix never enters the derived history; its durable record is the header events, and the dev invariant recomputes exactly this equation against every loop-built request.
|
||||
|
||||
@@ -254,6 +247,15 @@ The fifteen event variants (`turn/start`, `turn/end`, `step/start`, `step/end`,
|
||||
|
||||
Source: [`packages/core/agent/src/types.ts`](../../packages/core/agent/src/types.ts)
|
||||
|
||||
`InjectOptions` extends ordinary message attribution with context-only framing and durable model-hidden JSON metadata:
|
||||
|
||||
```ts type-equiv
|
||||
interface InjectOptions extends SendOptions {
|
||||
envelope?: ContextEnvelope
|
||||
meta?: JsonValue
|
||||
}
|
||||
```
|
||||
|
||||
```ts type-equiv
|
||||
interface Agent {
|
||||
readonly id: AgentId
|
||||
@@ -290,8 +292,10 @@ interface Agent {
|
||||
/**
|
||||
* Inject in-session context (file-change notices, skill content, cron
|
||||
* notifications, …): appends a `context/message` session event the next model
|
||||
* request sees at its chronological position, rendered as tagged synthetic
|
||||
* context rather than a user prompt. Does not run the model.
|
||||
* request sees at its chronological position, rendered as synthetic context
|
||||
* rather than a user prompt. The default uses the canonical context tag;
|
||||
* `options.envelope: 'raw'` preserves caller-owned framing. Does not run the
|
||||
* model.
|
||||
*
|
||||
* Turn-enclosure (the turn-enclosure RFC): an inject while a turn is open joins that turn;
|
||||
* an inject while idle wraps its `context/message` in a one-shot `injection`
|
||||
@@ -301,11 +305,11 @@ interface Agent {
|
||||
* (inject is synchronous): a failing flush is reported via `agent/error`
|
||||
* (step `0`) and the logger, never thrown into the caller.
|
||||
*
|
||||
* Live-adapter review has validated the tagged-envelope rendering against
|
||||
* current DeepSeek behavior; provider-specific mismatches belong in that
|
||||
* adapter, not in the canonical session vocabulary.
|
||||
* Live-adapter review has validated the canonical tagged-envelope rendering
|
||||
* against current DeepSeek behavior; provider-specific mismatches belong in
|
||||
* that adapter, not in the canonical session vocabulary.
|
||||
*/
|
||||
inject(content: ContentBlock[], options?: SendOptions): void
|
||||
inject(content: ContentBlock[], options?: InjectOptions): void
|
||||
|
||||
/**
|
||||
* Cancel ALL pending work for the agent. `cancel()`:
|
||||
@@ -355,11 +359,13 @@ interface Agent {
|
||||
}
|
||||
```
|
||||
|
||||
`AgentStatus` is `'idle' | 'running' | 'disposed'`. `AgentId` is a branded string. `AgentOptions` (`model?`) is merge-extensible — plugins add creation options by declaration merging. Persona is not an agent option: the `dsh-system-prompt` config supplies the global default, and an agent-scoped `deployment:persona` section may shadow it. The `agent/*` event taxonomy (lifecycle emits incl. `agent/session-start`, serial `agent/pre-step`/`agent/turn-stop` checkpoints, and the `agent/prompt-submit`/`agent/request`/`agent/session-prefix`/`agent/step-result`/`agent/turn-continuation` waterfalls) is in [architecture.md § Event taxonomy](../architecture.md#event-taxonomy); turn/step boundaries are durable `session/event` records, not `agent/*` emits.
|
||||
`AgentStatus` is `'idle' | 'running' | 'disposed'`, and `AgentId` is branded. `AgentOptions` is merge-extensible and currently includes `model?`. Persona belongs to `dsh-system-prompt`: an agent-scoped `deployment:persona` may shadow the global default.
|
||||
|
||||
The [event taxonomy](../architecture.md#event) owns the `agent/*` lifecycle, checkpoint, and waterfall contracts. Turn and step boundaries are durable session events rather than agent emits.
|
||||
|
||||
## Interception decisions
|
||||
|
||||
Each `agent/*` interception waterfall returns a small, seam-specific typed union — the unified Decision idiom (the tool seams' `PreToolDecision`/`PostToolDecision` in [tools.md](tools.md) follow the same shape). A CC/Codex hook bridge maps its `permissionDecision`/`decision`/`continue`/`additionalContext` fields onto these; a native plugin returns them directly. They share one envelope for model-facing context, `HookContext`, which is `inject()`ed as a `context/message` and so carries a REQUIRED `source` (a missing source would default to `{kind:'user'}` and mislabel plugin context as a user prompt).
|
||||
Each `agent/*` interception waterfall returns a small, seam-specific typed union — the unified Decision idiom (the tool seams' `PreToolDecision`/`PostToolDecision` in [tools.md](tools.md) follow the same shape). A CC/Codex hook bridge maps its `permissionDecision`/`decision`/`continue`/`additionalContext` fields onto these; a native plugin returns them directly. Prompt and post-tool decisions share one model-facing context shape, `HookContext`, which is `inject()`ed as a `context/message` and therefore carries a REQUIRED `source` (a missing source would default to `{kind:'user'}` and mislabel plugin context as a user prompt). Its optional `envelope` selects the canonical context tag or caller-owned raw framing, while JSON `meta` persists plugin state without exposing it to the model. Both decisions carry `additionalContexts[]` so every entry preserves its own provenance, framing, and metadata. Continuation reasons are steering messages instead and deliberately use the narrower content/source shape.
|
||||
|
||||
Source: [`packages/core/agent/src/types.ts`](../../packages/core/agent/src/types.ts)
|
||||
|
||||
@@ -367,23 +373,25 @@ Source: [`packages/core/agent/src/types.ts`](../../packages/core/agent/src/types
|
||||
interface HookContext {
|
||||
content: ContentBlock[]
|
||||
source: MessageSource
|
||||
envelope?: ContextEnvelope
|
||||
meta?: JsonValue
|
||||
}
|
||||
```
|
||||
|
||||
`agent/prompt-submit` returns a `PromptDecision` (allow a drained queued message — optionally rewriting its `content` or attaching `additionalContext` — or block it; a batch whose every prompt is blocked opens a zero-step turn that ends `rejected`):
|
||||
`agent/prompt-submit` returns a `PromptDecision` (allow a drained queued message — optionally rewriting its `content` or attaching `additionalContexts` — or block it; a batch whose every prompt is blocked opens a zero-step turn that ends `rejected`):
|
||||
|
||||
```ts type-equiv
|
||||
type PromptDecision =
|
||||
| { kind: 'allow'; content?: ContentBlock[]; additionalContext?: HookContext }
|
||||
| { kind: 'allow'; content?: ContentBlock[]; additionalContexts?: HookContext[] }
|
||||
| { kind: 'block'; reason: string }
|
||||
```
|
||||
|
||||
`agent/turn-continuation` returns a `ContinuationDecision` (the loop's default is `continue` when the step had tool calls or steering was injected, else `stop`; a `continue` `reason` is recorded as next-step steering in the same turn — the typed `/goal` pattern):
|
||||
`agent/turn-continuation` returns a `ContinuationDecision` (the loop's default is `continue` when the step had tool calls or steering was injected, else `stop`; a `continue` `reason` is recorded as next-step steering in the same turn and therefore carries no context envelope or metadata — the typed `/goal` pattern):
|
||||
|
||||
```ts type-equiv
|
||||
type ContinuationDecision =
|
||||
| { action: 'stop' }
|
||||
| { action: 'continue'; reason?: HookContext }
|
||||
| { action: 'continue'; reason?: { content: ContentBlock[]; source: MessageSource } }
|
||||
```
|
||||
|
||||
`agent/turn-stop` returns the stop-only `ContinuationStop` subset or `undefined`. The loop calls this serial checkpoint after folding the ordinary decision, its reason, and pending steering; a stop is terminal and discards pending steering.
|
||||
@@ -398,7 +406,7 @@ type ContinuationStop = Extract<ContinuationDecision, { action: 'stop' }>
|
||||
type SessionStartSource = 'startup' | 'resume' | 'clear' | 'compact'
|
||||
```
|
||||
|
||||
`agent/session-prefix` composes the session prefix — a plain `Message[]`, no dedicated payload type. Fired ONCE per loop instance, lazily on its first request: the composed list is deep-frozen, recorded as the header's `messagePrefix` ([the request envelope](#the-request-envelope-llmcallconfig-and-the-logged-header)), and placed in front of the ENTIRE derived history on every request the instance sends — the home for session-stable openers like a skills catalog or an AGENTS.md digest, never returned by `deriveMessages()`. Reuse is structural, so the prefix cannot drift mid-session (resume = a new instance = a recompose); content that changes mid-session goes through the append-only history channels instead (`agent.inject()`, `tools/post-execute` / prompt-submit `additionalContext`). Not a Decision union: the seam contributes content instead of vetoing, so the shape is the contribution itself.
|
||||
`agent/session-prefix` composes a `Message[]` once per loop instance. The deep-frozen result is recorded in the request header and prepended to every derived history, making it the home for session-stable openers. A resumed instance recomposes; mid-session changes use append-only context channels. The waterfall returns content directly because it contributes rather than decides.
|
||||
|
||||
## `ToolDefinition`
|
||||
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# Filesystem
|
||||
|
||||
The filesystem stack is split across four packages: a provider seam ([dsh-fs](../../packages/fs/fs), `ctx.fs`, text IO + atomic mutation primitives whose version guard is optional), a local implementation ([dsh-fs-local](../../packages/fs/fs-local), local disk), a policy plugin ([dsh-fs-policy](../../packages/fs/fs-policy), observed-state + read-before-edit + version-guarded write/edit, contributed through the `fs/*` event gate — NO service), and a consumer ([dsh-tool-fs](../../packages/fs/tool-fs), the model-facing `read`/`write`/`edit` tools, which is also the EXECUTOR — it reads/writes/edits through `ctx.fs` directly and owns read windowing). Filesystem access is an optional capability, not part of the agent-loop spine, so its vocabulary lives here rather than in [core.md](core.md). A sandboxed, remote, virtual, or project-scoped backend can implement the same `FileSystem` service without changing the policy plugin or the tool schemas.
|
||||
The optional filesystem capability has four parts: [dsh-fs](../../packages/fs/fs) owns `ctx.fs` and atomic text operations with optional version guards, [dsh-fs-local](../../packages/fs/fs-local) implements local disk, [dsh-fs-policy](../../packages/fs/fs-policy) adds observed-state and freshness rules through events rather than a service, and [dsh-tool-fs](../../packages/fs/tool-fs) directly executes model-facing read/write/edit calls and renders windows. It is outside the agent-loop spine; alternate backends do not change policy or tool schemas.
|
||||
|
||||
The model is **additive, not subtractive**: `ctx.fs` alone is a complete, unconstrained text-storage seam (`write` unconditionally creates-or-overwrites, `edit` unconditionally replaces literal text). `dsh-fs-policy` is a plugin that *adds* policy on top by deciding the `fs/*` waterfalls; removing it leaves the bare provider rather than breaking the tool, because the tool is not method-coupled to the policy. A deployment that loads `dsh-tool-fs` is expected to also load `dsh-fs-policy` so the default behavior is read-before-write/edit.
|
||||
|
||||
@@ -37,6 +37,16 @@ interface FsInfo {
|
||||
}
|
||||
```
|
||||
|
||||
`lstat` is the path-level no-follow metadata primitive. It takes a path instead of an `FsTarget` because `resolve` intentionally follows symlinks to produce stable identity; consumers that need trust-boundary checks can call `lstat` first and reject `symlink` before resolving.
|
||||
|
||||
```ts type-equiv
|
||||
interface FsPathInfo {
|
||||
version: FsVersion
|
||||
type: 'file' | 'directory' | 'symlink' | 'other'
|
||||
size?: number
|
||||
}
|
||||
```
|
||||
|
||||
`listDir` returns direct child entries in stable name order. Each entry carries the child basename, type, resolved target, and cheap metadata when the backend can report it. It must not read file contents, so `size` is only for regular files and `version` is metadata-derived. Broken or disappeared children may be returned as `other` without metadata; permission or backend I/O failures while listing or resolving child metadata fail the whole listing with `FS_PERMISSION_DENIED` or `FS_IO_ERROR`.
|
||||
|
||||
```ts type-equiv
|
||||
@@ -144,4 +154,4 @@ type FsErrorCode =
|
||||
|
||||
## The service and the plugin
|
||||
|
||||
`FileSystem` (`ctx.fs`, abstract) owns the provider primitives: `resolve`, `stat`, `readText`, `streamText`, `listDir`, `writeText`, and `editText`. `dsh-fs-policy` registers **no service** — it is a plugin that adds policy through the `fs/*` event gate: it decides the write/edit intent waterfalls (supplying `createIfAbsent`/`replaceIfVersion`/`{ version }` or throwing `FS_NOT_OBSERVED`) and records on `fs/observed`. The executor is `dsh-tool-fs`: it reads/writes/edits through `ctx.fs`, dispatches the waterfalls, and emits the recording event. The generated wiring catalog shows the exact `ctx.fs` signatures on [services.md](../cordis-catalog/services.md#ctxfs--filesystem-abstract-seam).
|
||||
`FileSystem` (`ctx.fs`, abstract) owns the provider primitives: `resolve`, `stat`, `lstat`, `readText`, `streamText`, `listDir`, `writeText`, and `editText`. `dsh-fs-policy` registers **no service** — it is a plugin that adds policy through the `fs/*` event gate: it decides the write/edit intent waterfalls (supplying `createIfAbsent`/`replaceIfVersion`/`{ version }` or throwing `FS_NOT_OBSERVED`) and records on `fs/observed`. The executor is `dsh-tool-fs`: it reads/writes/edits through `ctx.fs`, dispatches the waterfalls, and emits the recording event. The generated wiring catalog shows the exact `ctx.fs` signatures on [services.md](../cordis-catalog/services.md#ctxfs--filesystem-abstract-seam).
|
||||
|
||||
@@ -4,6 +4,14 @@ The in-memory, event-sourced model of [dsh-session](../../packages/core/session)
|
||||
|
||||
Source: [`packages/core/session/src/types.ts`](../../packages/core/session/src/types.ts)
|
||||
|
||||
## Context framing
|
||||
|
||||
`ContextEnvelope` selects the standard tagged projection or preserves a producer-owned complete frame. The latter changes framing only; the event remains a user-role `context/message` in chronological history.
|
||||
|
||||
```ts type-equiv
|
||||
type ContextEnvelope = 'context' | 'raw'
|
||||
```
|
||||
|
||||
## `SessionEventMap` — the event vocabulary
|
||||
|
||||
The append-only event types. Merge-extensible: a plugin declares extra event types via declaration merging — e.g. the [compaction seam](compaction.md) adds `compact/start` / `compact/summary` / `compact/end`, and `@deepseek-ai/dsh-hook-protocol` adds log-only `hook/invoked` / `hook/result` provenance for a hook bridge. Like `compact/*`, these are NOT `SurfaceEventType`s (no `surfaceOp`). The generated [persistence log event catalog](../persistence-catalog.md) enumerates every member — core and merged — with its payload, surface badge, and declaration site.
|
||||
@@ -30,9 +38,16 @@ interface SessionEventMap {
|
||||
/**
|
||||
* In-session context injection (file-change notices, subdir AGENTS.md,
|
||||
* skill content, cron notifications, …). Rendered into the derived history
|
||||
* as tagged synthetic context — NOT a user prompt.
|
||||
* as synthetic context — NOT a user prompt. `envelope: 'raw'` lets a caller
|
||||
* supply its own complete framing; `meta` is persisted JSON hidden from the
|
||||
* model.
|
||||
*/
|
||||
'context/message': { content: ContentBlock[]; source: MessageSource }
|
||||
'context/message': {
|
||||
content: ContentBlock[]
|
||||
source: MessageSource
|
||||
envelope?: ContextEnvelope
|
||||
meta?: JsonValue
|
||||
}
|
||||
/** Raw stream chunk — token-level replay fidelity. */
|
||||
'assistant/chunk': { turn: number; step: number; chunk: StreamChunk }
|
||||
/**
|
||||
@@ -223,7 +238,8 @@ export interface SurfaceFoldResult {
|
||||
- `user/message` → a user message.
|
||||
- `assistant/message` → an assistant message. Raw `assistant/chunk` events are replay/UI data and are **skipped** in derivation (the assembled message is authoritative). An **empty-content** `assistant/message` is also skipped — a max-tokens step cut off with no content still records an `assistant/message` to host its `usage`, but a content-less assistant turn must not enter the provider transcript.
|
||||
- `tool/result` → a user message carrying a `tool-result` block.
|
||||
- `context/message`, `steering/message` → user-role messages wrapped in a tagged envelope (`<context source="…">…</context>`) at their chronological position — the "system-reminder" pattern; the model distinguishes them from real prompts by the envelope.
|
||||
- `context/message` → a user-role message at its chronological position. The default `envelope` is `context`, which wraps content as `<context source="…">…</context>`; `envelope: 'raw'` uses caller-owned framing verbatim. Optional JSON `meta` remains in the event log and is never rendered.
|
||||
- `steering/message` → a user-role message wrapped in `<steering source="…">…</steering>` at its chronological position.
|
||||
|
||||
Everything else (`turn/*`, `step/*`) is structural and does not project into a message. Token usage is observed on `assistant/message.usage` (the step that produced it); an operational error's step number is on `turn/end.reason` for `kind: 'error'`.
|
||||
|
||||
|
||||
@@ -6,7 +6,9 @@ Source: [`packages/skill/skill/src/index.ts`](../../packages/skill/skill/src/ind
|
||||
|
||||
## Provider registry
|
||||
|
||||
`ctx.skills` is a multi-provider registry. Providers can represent local directories, embedded plugin data, HTTP catalogs, or another source. Provider plugins register synchronously during `apply()`; remote initialization, authentication, and discovery are awaited by `list()`. Provider objects, lookup options, and candidates are readonly same-process contracts, so the registry borrows them instead of manufacturing defensive snapshots. The registry still validates semantic fields, resolves duplicate skill names first-wins by rank/provider order/local order, and sorts the final summaries by `name` for deterministic consumers. A provider `list()` rejection is logged and skipped without caching the degraded catalog; malformed candidates still fail fast because they violate the provider contract.
|
||||
`ctx.skills` combines local, embedded, remote, or other providers. Registration is synchronous; remote initialization and discovery belong in awaited `list()`. Provider objects, options, and candidates are borrowed readonly, while semantic fields are validated.
|
||||
|
||||
Duplicate names resolve by rank, provider order, then local order; summaries sort by name. A rejected `list()` is logged and skipped without caching the degraded catalog, while malformed candidates fail fast.
|
||||
|
||||
```ts type-equiv
|
||||
interface SkillProvider {
|
||||
|
||||
@@ -21,7 +21,7 @@ interface SubagentCapabilities {
|
||||
|
||||
## The start request
|
||||
|
||||
What a caller asks for when starting a subagent. The tool layer builds this from the model's `{ description, prompt }` plus its own config; the service validates the start-time capabilities against the named provider, then passes it to `provider.start`. `parent` is REQUIRED — in-process backends read `parent.session.header` for the working directory, the `parentSession` lineage, and the delegation depth. The four optional fields (`outputSchema`, `maxDepth`, `toolFilter`, `persona`) each gate on the matching `SubagentCapabilities` flag — in-process backends realize `toolFilter` as a scoped `tools.restrict()` and `persona` as a scoped shadowing `deployment:persona` section, both composed in the child's creation window. `outputSchema` is an object-rooted JSON Schema within the subset `assertSupportedOutputSchema` (dsh-tools) enforces — a schema outside it is rejected loud at start; the in-process backends realize it with a forced `structured_output` capture tool (see the [driver README](../../packages/subagent/subagent-inprocess/README.md)).
|
||||
The tool layer builds this request from the model input and its own config; the service validates it against the named provider before `start`. Required `parent` supplies the session cwd, lineage, and delegation depth. Optional output schema, depth, tool filter, and persona require matching capability flags. Unsupported schemas fail at start; in-process backends scope filters and personas to child creation and implement the supported object-rooted schema with a forced capture tool.
|
||||
|
||||
```ts type-equiv
|
||||
interface SubagentStartRequest {
|
||||
@@ -64,7 +64,7 @@ interface SubagentStopReasonMap {
|
||||
|
||||
## A live run: `SubagentRun`
|
||||
|
||||
The handle the consumer holds after a provider has established a ready child. The consumer awaits `result` and MUST `dispose` on every path to cancel remaining work and reach child quiescence. `result` does NOT reject on a child-level failure — a model/transport failure resolves with `stopReason: 'error'` — so the consumer maps a non-`completed` reason to an `isError` result; it rejects only on an infrastructure fault the seam cannot represent. `sendMessage` and `resume` are OPTIONAL: a provider that supports the runtime capability defines the method; one that doesn't omits it.
|
||||
`SubagentRun` is the consumer-owned handle for a ready child. Consumers await `result` and always dispose the run to reach quiescence. Child failures resolve with a non-completed stop reason; only unrepresentable infrastructure faults reject. Optional `sendMessage` and `resume` methods advertise their runtime capabilities by presence.
|
||||
|
||||
```ts type-equiv
|
||||
interface SubagentRun {
|
||||
@@ -78,7 +78,7 @@ interface SubagentRun {
|
||||
|
||||
## The provider seam: `SubagentProvider`
|
||||
|
||||
One transport for running a child agent. Implementations register under a unique name via `SubagentService.registerProvider`; multiple coexist in one context. The service validates every requested start-time capability before calling `start`, so an implementation may assume e.g. `request.maxDepth` is honorable when present. `inheritsParentContext` is a DESCRIPTIVE fact beside the capabilities (nothing validates against it): whether a child sees the parent conversation (`fork`: true, `spawn`/`acp`: false) — the model-facing consumer derives truthful tool wording from it. It describes conversation history only, not tool registrations, injected services, or authority inheritance.
|
||||
Each provider is a named child-agent transport, and multiple providers may coexist. The service validates requested start-time capabilities before `start()`. `inheritsParentContext` describes only conversation seeding (`fork`: true; `spawn` and `acp`: false), allowing consumers to generate accurate model-facing wording without implying inherited tools, services, or authority.
|
||||
|
||||
```ts type-equiv
|
||||
interface SubagentProvider {
|
||||
@@ -89,11 +89,11 @@ interface SubagentProvider {
|
||||
}
|
||||
```
|
||||
|
||||
`SubagentProvider.start()` and `ctx.subagents.start()` are the publication boundary: their promises fulfill only with a ready run. The service attaches result observation, emits `subagent/start`, and returns the same holder-owned run; a rejected start has already cleaned provider-owned partial resources and emits neither lifecycle event. For an in-process provider, a start listener can resolve the live child with `ctx.agents.get(info.id)`; a remote provider need not publish into the local registry. `subagent/end` carries `lastAssistantMessage` (the child's final `output`) on the settle path and reports `error` on infrastructure rejection. Both lifecycle events are observe-only emits with per-listener exception containment.
|
||||
`start()` fulfills only with a ready run. The service observes its result, emits `subagent/start`, and returns the same run; rejection implies provider cleanup and emits no lifecycle pair. In-process children are discoverable through `ctx.agents`, while remote children need not be. `subagent/end` reports final output or infrastructure failure. Both events are observe-only and contain listener exceptions.
|
||||
|
||||
## In-process backends: depth and seed
|
||||
|
||||
The two in-process backends ([dsh-subagent-spawn](../../packages/subagent/subagent-spawn) fresh, [dsh-subagent-fork](../../packages/subagent/subagent-fork) seeded) run the child as an ordinary `Agent` in the same application. The provider creates it directly through `parent.ctx`, passes the required signal into the core creation transaction, and delegates quiescent disposal to the returned `AgentHandle`. Provider removal prevents new starts but does not revoke an accepted run. The child receives a flat new scope rather than inheriting the parent's registrations. Two pieces of vocabulary ride on the existing agent/session types rather than new core types:
|
||||
The spawn and fork backends create an ordinary agent through `parent.ctx`, pass cancellation into core creation, and dispose through `AgentHandle`. Provider removal blocks new starts without revoking accepted runs. Each child gets a new flat scope rather than inheriting parent registrations. Depth and fork seeding reuse existing agent and session vocabulary:
|
||||
|
||||
- **Delegation depth** is a merge-extensible `AgentOptions.subagentDepth` field (`0` for a top-level agent, parent + 1 for a child). Only `undefined` means top level; every stored present value must be a non-negative safe integer. The seam owns it — the loop neither sets nor reads it — so a nested spawn validates its parent's stored depth, rejects a derived child depth outside the safe-integer domain, and applies a defined absolute `request.maxDepth` cap to that child.
|
||||
- **Fork seeding** uses `CreateAgentOptions.seed` (a `SessionEvent[]` prefix threaded through `AgentLoop.createAgent` → `ctx.sessions.prepare({ seed })`, the same primitive `resume` uses). The fork backend passes a *balanced completed-turn prefix* of the parent's log — the parent's events up to and including its last `turn/end` — so the seed is contiguous-from-0 and the [invariants](../../packages/support/invariants) replay accepts it (the in-flight, unbalanced turn is excluded).
|
||||
|
||||
129
docs/core-data-structures/tasks.md
Normal file
129
docs/core-data-structures/tasks.md
Normal file
@@ -0,0 +1,129 @@
|
||||
# Background Task Runtime
|
||||
|
||||
Types shared by long-running producers, `ctx.tasks`, and task control surfaces. The [runtime RFC](../rfc/implemented/architecture/2026-06-20-generic-long-running-tool-runtime.md) owns the design; this page records the literal shapes from [`packages/tasks/tasks/src/types.ts`](../../packages/tasks/tasks/src/types.ts).
|
||||
|
||||
## Ids and status
|
||||
|
||||
`TaskId` is a [branded id](core.md#branded-ids) generated as `<kind>-N`. Access control relies on owner authorization, not id secrecy. `TaskKind` derives from a merge-extensible map; the registry treats kinds as opaque id namespaces.
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskKindMap {
|
||||
bash: 'bash'
|
||||
subagent: 'subagent'
|
||||
}
|
||||
```
|
||||
|
||||
`TaskStatus` is `'running' | 'stopping' | 'completed' | 'killed' | 'failed'`; producer-specific facts belong in `TaskSnapshot.detail`.
|
||||
|
||||
## Producer contract
|
||||
|
||||
`TaskStart` declares identity and a starter. The runtime finishes preflight before calling `run()` and commits without a later failable step. Producers own execution resources; the runtime owns identity, access, and lifecycle state.
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskStart {
|
||||
/** Producer kind — also the id prefix (`bash`, `subagent`, …). */
|
||||
kind: TaskKind
|
||||
/** One-line model-facing label (the command; the delegation description). */
|
||||
label: string
|
||||
/**
|
||||
* Owning live agent. Access is fenced by its session id, and agent disposal
|
||||
* cancels and awaits the task. The instance must be the one currently
|
||||
* registered under its agent id. Omitting the owner creates an unowned task,
|
||||
* open to any caller until service disposal.
|
||||
*/
|
||||
owner?: Agent
|
||||
/**
|
||||
* Start the work after preflight and synchronously return its hooks. Called
|
||||
* once; a throw leaves nothing registered, and the producer must clean up any
|
||||
* partially started resources.
|
||||
*/
|
||||
run(): TaskHooks
|
||||
}
|
||||
```
|
||||
|
||||
`TaskHooks.done` is the quiescence boundary. Optional `readOutput` distinguishes consuming stream tasks from final-output-only tasks.
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskHooks {
|
||||
/**
|
||||
* Request termination. Must be synchronous, idempotent, and eventually settle
|
||||
* {@link done}; throws propagate. The optional reason is forwarded verbatim.
|
||||
*/
|
||||
cancel(reason?: string): void
|
||||
/**
|
||||
* Resolves after the producer releases its resources, not merely when work
|
||||
* finishes. Must not reject; the runtime converts a rejection to `failed`.
|
||||
* If teardown cancellation throws, the runtime may force-fail only the
|
||||
* registry record without claiming that the work stopped.
|
||||
*/
|
||||
done: Promise<TaskOutcome>
|
||||
/**
|
||||
* Consume output produced since the previous call. The producer formats
|
||||
* truncation and spill notices. Absence marks a final-output-only task; each
|
||||
* task has one consuming cursor.
|
||||
*/
|
||||
readOutput?(): string
|
||||
}
|
||||
```
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskOutcome {
|
||||
/** How the task ended: finished (`completed`), cancelled (`killed`), or broke (`failed`). */
|
||||
status: 'completed' | 'killed' | 'failed'
|
||||
/** Kind-specific detail rendered into status lines ('exit code: 3', 'max-tokens'). */
|
||||
detail?: string
|
||||
/** Final output for tasks without `readOutput`; stream tasks leave it unset. */
|
||||
output?: string
|
||||
}
|
||||
```
|
||||
|
||||
## Consumer views
|
||||
|
||||
Snapshots are fresh read-only projections. `ownerSession` carries the shared `SessionId` used for authorization; completion listeners separately receive the exact owner object used for lifecycle cleanup. `reported` suppresses a completion notice after another surface has delivered or committed to deliver the terminal state.
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskSnapshot {
|
||||
/** The registry-issued id (`<kind>-N`). */
|
||||
id: TaskId
|
||||
/** The producer kind the task was registered with. */
|
||||
kind: TaskKind
|
||||
/** The producer-supplied one-line label. */
|
||||
label: string
|
||||
/**
|
||||
* Owner session id used for authorization and correlation; absent for
|
||||
* unowned tasks. Completion listeners receive the exact {@link Agent}
|
||||
* separately through {@link TaskDoneListener}.
|
||||
*/
|
||||
ownerSession?: SessionId
|
||||
/** Current lifecycle state. */
|
||||
status: TaskStatus
|
||||
/** Kind-specific status detail, present once the producer supplied one (usually terminal). */
|
||||
detail?: string
|
||||
/** Epoch ms when the task was registered. */
|
||||
startedAt: number
|
||||
/** Epoch ms when the task settled; absent while `running`/`stopping`. */
|
||||
finishedAt?: number
|
||||
/**
|
||||
* True when a kill, read, or wait has reported or committed to report the
|
||||
* terminal state. Completion surfaces suppress redundant notices when set.
|
||||
*/
|
||||
reported: boolean
|
||||
}
|
||||
```
|
||||
|
||||
```ts type-equiv
|
||||
interface TaskRead {
|
||||
/**
|
||||
* Stream kinds: the consuming delta since the previous read. Final-output
|
||||
* kinds: empty while live, the terminal {@link TaskOutcome.output} (or
|
||||
* empty) once settled — idempotent, never consumed.
|
||||
*/
|
||||
text: string
|
||||
/** The task's state at read time. */
|
||||
snapshot: TaskSnapshot
|
||||
}
|
||||
```
|
||||
|
||||
## Service behavior
|
||||
|
||||
[`TaskService`](../../packages/tasks/tasks/src/index.ts) provides atomic `start`, caller-scoped `get` and `list`, `read`, `kill`, bounded `wait`, contained `onTaskDone` listeners, and the `attachSurface` availability fence. Authorization compares owner sessions; owner cleanup selects the exact registered `Agent` instance. See [`dsh-tasks`](../../packages/tasks/tasks/README.md) for the package contract and [`dsh-tool-tasks`](../../packages/tasks/tool-tasks/README.md) for the model-facing surface.
|
||||
@@ -10,7 +10,7 @@ A `ToolSchema` (the model-facing fields) plus the `execute` function and optiona
|
||||
|
||||
```ts type-equiv
|
||||
interface ToolDefinition extends ToolSchema {
|
||||
execute(args: unknown, exec: ToolExecution): Promise<ToolExecuteReturn>
|
||||
execute(args: unknown, exec: ToolRunContext): Promise<ToolExecuteReturn>
|
||||
/**
|
||||
* Cooperative tool-call timeout budget in milliseconds. Omit for no deadline.
|
||||
* Enforced by `@deepseek-ai/dsh-timeout-policy` (a `tools/execute` wrapper); it
|
||||
@@ -120,6 +120,19 @@ interface ToolExecutionInput {
|
||||
}
|
||||
```
|
||||
|
||||
A tool body receives the runtime extension. `deferContext()` is the composite-tool channel: it records nested-dispatch context without injecting inside the still-open outer call.
|
||||
|
||||
```ts type-equiv
|
||||
interface ToolRunContext extends ToolExecution {
|
||||
/**
|
||||
* Defer one nested-dispatch context until this tool's final result reaches
|
||||
* the agent loop. Contexts retain their individual source, envelope, and
|
||||
* metadata and are emitted in call order.
|
||||
*/
|
||||
deferContext(context: HookContext): void
|
||||
}
|
||||
```
|
||||
|
||||
```ts type-equiv
|
||||
interface ToolExecution extends ToolExecutionInput {
|
||||
/** Registry-assigned identity shared with nested calls only as their opaque `parent` token. */
|
||||
@@ -127,7 +140,7 @@ interface ToolExecution extends ToolExecutionInput {
|
||||
}
|
||||
```
|
||||
|
||||
`ToolExecutionToken` is a compile-time opaque fresh `Symbol` at runtime; identity comparison is its only operation. Before policy runs, `ctx.tools.execute()` materializes `arguments` as detached lossless JSON, assigns the token, and deep-freezes the accepted arguments. A non-JSON value is normalized to an error before policy. `token`, `callId`, `name`, `arguments`, `agent`, and the optional `parent` token are readonly throughout the waterfalls, while an around-dispatch wrapper may add, replace, or remove only optional `signal`. After the complete pipeline the registry freezes the execution and exposes its stable identity to `tools/result` observers.
|
||||
`ToolExecutionToken` is an opaque runtime `Symbol` used only for identity comparison. Before policy, `execute()` materializes and freezes arguments, rejects non-JSON input, and assigns the token. Identity fields and the optional parent token remain readonly; only `signal` may change around dispatch. Final observers receive the frozen execution identity.
|
||||
|
||||
A `ToolGuard` is scope-aware final pre-dispatch policy. Its shape deliberately has no allow result: `undefined` preserves the waterfall decision, while a returned reason can only reduce permission, so a later listener cannot undo it.
|
||||
|
||||
@@ -137,7 +150,6 @@ type ToolGuard = (execution: Readonly<ToolExecution>) => string | undefined
|
||||
|
||||
```ts type-equiv
|
||||
interface ToolExecutionResult {
|
||||
callId: CallId
|
||||
content: ContentBlock[]
|
||||
isError: boolean
|
||||
/**
|
||||
@@ -147,16 +159,14 @@ interface ToolExecutionResult {
|
||||
*/
|
||||
error?: ToolErrorInfo
|
||||
/**
|
||||
* Extra model-facing context a `tools/post-execute` listener attached for the
|
||||
* NEXT request (Claude Code's PostToolUse `additionalContext`). It is NOT part
|
||||
* of this call's `content` — `content`/`feedback` shape the tool RESULT, but
|
||||
* `additionalContext` is a SEPARATE `context/message`. A step can carry
|
||||
* multiple tool calls, so the loop BUFFERS every call's `additionalContext`
|
||||
* and appends them only AFTER all `tool/result`s for the step, keeping
|
||||
* tool-call/result adjacency intact. Carried on the result purely to ferry it
|
||||
* from `execute()` up to the loop's per-step buffer.
|
||||
* Extra model-facing contexts deferred by a composite tool or attached by
|
||||
* `tools/post-execute` listeners for the NEXT request. They are NOT part of
|
||||
* this call's `content`: the loop buffers every context and appends them only
|
||||
* AFTER all `tool/result`s for the step, preserving tool-call/result
|
||||
* adjacency. The array preserves each context's source, envelope, metadata,
|
||||
* and production order instead of flattening mixed plugin provenance.
|
||||
*/
|
||||
additionalContext?: HookContext
|
||||
additionalContexts?: HookContext[]
|
||||
/**
|
||||
* The tool-private presentation payload from a successful `execute` (the object
|
||||
* return form). Threaded onto the `tool/result` session event and back into
|
||||
@@ -167,6 +177,8 @@ interface ToolExecutionResult {
|
||||
}
|
||||
```
|
||||
|
||||
The result carries only the outcome. Call identity remains on the immutable `ToolExecution` that accompanies it through every hook and on the durable `tool/call` / `tool/result` session events, so wrappers cannot create a second, disagreeing identity.
|
||||
|
||||
The registry materializes and freezes the final accepted result immediately before `tools/result`. Its content, structured error, additional context, and presentation metadata must round-trip losslessly through JSON; an invalid outcome becomes a JSON-safe `isError` result, so the observed live outcome is safe for the later durable `tool/result` append.
|
||||
|
||||
Each interception waterfall returns a typed **Decision** (the idiom shared with the `agent/*` seams). `tools/pre-execute` listeners receive `(exec, next)` and return a `PreToolDecision`; `tools/execute` wrappers return a `ToolExecutionResult`; `tools/post-execute` listeners receive `(exec, result, next)` and return a `PostToolDecision`:
|
||||
@@ -180,11 +192,13 @@ type PreToolDecision =
|
||||
|
||||
```ts type-equiv
|
||||
type PostToolDecision =
|
||||
| { kind: 'accept'; content?: ContentBlock[]; additionalContext?: HookContext }
|
||||
| { kind: 'block'; feedback: ContentBlock[]; additionalContext?: HookContext }
|
||||
| { kind: 'accept'; content?: ContentBlock[]; additionalContexts?: HookContext[] }
|
||||
| { kind: 'block'; feedback: ContentBlock[]; additionalContexts?: HookContext[] }
|
||||
```
|
||||
|
||||
Call `next()` to delegate to the default (allow / dispatch / accept-unchanged), or return a decision/result to short-circuit. A `pre-execute` `deny` skips dispatch and yields an `isError` result. An `ask` resolves through the optional approval seam: only `allowed-once` proceeds, while every non-grant, missing channel/service, or agent-less request becomes a normalized denial. A registered `ToolGuard` then runs and can still impose a final denial. Input rewrite is deliberately NOT offered on `PreToolDecision` because it would desync the pre-execution audit/history/UI from what ran. A `post-execute` `accept` may replace the model-facing `content`; a `block` turns the call into an `isError` whose content is the corrective `feedback`. The synchronous `tools/result` notification then receives the frozen execution identity and a deep-frozen result snapshot after every wrapper, post decision, and outer error catch; observers cannot transform the outcome or race each other through payload mutation, and one observer failure neither changes the result nor starves peers. An unregistered tool routes through the same catch as a tool-thrown error, so both failure classes get a structured `{ name, code }` (`ToolNotFoundError` → `UNKNOWN_TOOL`) — the loop records a failed tool call instead of failing the whole turn.
|
||||
Call `next()` for the default or return a decision to short-circuit. Pre-policy may deny or ask; only `allowed-once` proceeds, while a non-grant, missing approval channel or service, or agent-less request becomes a denial. Guards may still impose a final denial. Arguments cannot be rewritten because history, audit, UI, and execution must agree.
|
||||
|
||||
Post-policy may replace content; a block becomes an `isError` result containing its corrective feedback. `tools/result` receives the frozen execution and result after normalization; observers cannot transform them, and observer failures are contained. Unknown and throwing tools both become structured errors (`ToolNotFoundError` maps to `UNKNOWN_TOOL`), so the call fails without ending the turn.
|
||||
|
||||
## The structured-output schema subset
|
||||
|
||||
@@ -229,4 +243,4 @@ How a tool wants its call shown in a UI (an editor tool-call card, a CLI log lin
|
||||
|
||||
`ToolCallKind` (`'read' | 'edit' | 'delete' | 'move' | 'search' | 'execute' | 'fetch' | 'other'`) picks an icon on a generic card. `FileLocation` (`{ path, line? }`) and `FileDiff` (`{ path, oldText, newText }`) are the shared file-card vocabulary. The design is pinned in [the render-intent-union RFC](../rfc/implemented/architecture/2026-07-02-tool-render-intent-union.md); the ACP bridge maps a `diff` card to a `{ type: 'diff' }` content block, a `terminal` card to the `_meta` terminal convention, and relativizes a file card's title against the session cwd.
|
||||
|
||||
The full presentation field docs live in [`packages/core/tools/src/presentation.ts`](../../packages/core/tools/src/presentation.ts). The bash tool's own schemas (`bash`/`bash_output`/`bash_kill`) and the executor they drive are on [bash.md](bash.md).
|
||||
The full presentation field docs live in [`packages/core/tools/src/presentation.ts`](../../packages/core/tools/src/presentation.ts). The `bash` schema and executor are on [bash.md](bash.md); generic background controls are on [tasks.md](tasks.md).
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# User Interaction
|
||||
|
||||
The user-interaction seam of [dsh-user-interaction](../../packages/ui/user-interaction). It is the provider-neutral vocabulary a tool or permission plugin uses when it needs the human to answer before the agent can continue. UI surfaces provide the active `UserInteractionProvider`: `dsh-stdio-agent` renders questions in readline, and `dsh-acp` maps them to ACP form elicitations.
|
||||
The user-interaction seam of [dsh-user-interaction](../../packages/ui/user-interaction). It is the provider-neutral vocabulary a tool or permission plugin uses when it needs the human to answer before the agent can continue. UI surfaces provide the active `UserInteractionProvider`: `dsh-stdio-demo` renders questions in readline, and `dsh-acp` maps them to ACP form elicitations.
|
||||
|
||||
Source: [`packages/ui/user-interaction/src/index.ts`](../../packages/ui/user-interaction/src/index.ts)
|
||||
|
||||
|
||||
@@ -25,8 +25,6 @@ interface WebSearchRequest {
|
||||
|
||||
```ts type-equiv
|
||||
interface WebSearchResult {
|
||||
readonly providerId: string
|
||||
readonly query: string
|
||||
readonly content?: string
|
||||
readonly sources: readonly WebSearchSource[]
|
||||
readonly truncated: boolean
|
||||
@@ -49,7 +47,6 @@ interface WebSearchSource {
|
||||
```ts type-equiv
|
||||
interface WebFetchRequest {
|
||||
readonly url: string
|
||||
readonly timeoutMs?: number
|
||||
}
|
||||
```
|
||||
|
||||
@@ -57,7 +54,6 @@ HTTP status is part of the fetched resource state, not automatically a failure:
|
||||
|
||||
```ts type-equiv
|
||||
interface WebFetchResult {
|
||||
readonly providerId: string
|
||||
readonly url: string
|
||||
readonly statusCode: number
|
||||
readonly body: WebFetchBody
|
||||
@@ -73,15 +69,9 @@ type WebFetchBody =
|
||||
| { readonly kind: 'text'; readonly content: string }
|
||||
```
|
||||
|
||||
## Provider status
|
||||
## Provider availability
|
||||
|
||||
A provider's `status()` is a cheap LOCAL check (credential presence, parseable config) and **must not make network calls**. It is an input to execution-time selection, not a health system: `search()`/`fetch()` read it to pick a usable provider, and a selection failure surfaces as the structured `WebError` the caller routes on — which carries the branchable detail (the missing id, the ambiguous candidate set) in its code and message.
|
||||
|
||||
```ts type-equiv
|
||||
type WebProviderStatus =
|
||||
| { readonly available: true }
|
||||
| { readonly available: false; readonly reason: 'missing-credential' | 'misconfigured' }
|
||||
```
|
||||
A provider's `available(): boolean` is a cheap LOCAL check (credential presence, parseable config) and **must not make network calls**. It is an input to execution-time selection, not a health system: `search()`/`fetch()` read it to pick a usable provider, and a selection failure surfaces as the structured `WebError` the caller routes on — which carries the branchable detail (the missing id or ambiguous candidate set) in its code and message.
|
||||
|
||||
Selection never depends on registration, config, or HMR order: a capability has an explicit provider id (config `searchProvider`/`fetchProvider`, or the matching env var feeding the same field), or auto-selects when exactly one usable provider is registered; multiple usable providers with no configured id is `WEB_PROVIDER_AMBIGUOUS`, not first-wins.
|
||||
|
||||
@@ -91,4 +81,4 @@ Selection never depends on registration, config, or HMR order: a capability has
|
||||
|
||||
## The service
|
||||
|
||||
`WebService` (`ctx.web`, defined in [`packages/web/web/src/index.ts`](../../packages/web/web/src/index.ts)) is a provider registry plus a provider-selecting execution surface, close to `LlmService`'s shape: `registerSearchProvider`/`registerFetchProvider` (duplicate ids throw `WEB_DUPLICATE_PROVIDER`, return disposers) and `search`/`fetch` (resolve the provider at call time, throw a structured `WebError` when the capability cannot run). Providers issue requests with platform-native `fetch` at the repo's Node floor, mirroring `dsh-llm-deepseek`; the `dsh-web-fetch-local` provider owns safe retrieval (http/https-only, credential rejection, byte/char/timeout/redirect caps, same-origin-only redirects with per-hop re-validation, charset decoding) while `dsh-tool-web` owns presentation (HTML→markdown). SSRF / private-network blocking is deferred (see the RFC) — until it lands, `web_fetch` must not be enabled where it can reach sensitive internal targets.
|
||||
`WebService` registers search and fetch providers, rejects duplicate ids with `WEB_DUPLICATE_PROVIDER`, and resolves providers at execution time with structured selection errors. The local fetch backend accepts only HTTP(S), rejects credentials, caps redirects, bytes, characters, and time, revalidates every same-origin redirect hop, and decodes the body; the tool owns presentation. Private-network blocking is deferred, so do not enable `web_fetch` where it can reach sensitive internal targets.
|
||||
|
||||
Reference in New Issue
Block a user