One principle: every fact in the assembled prompt has exactly one owner.
- dsh-system-prompt: merge-extensible AssembleContext on assemble();
a variable(name, provider) registry; {{name}} interpolation in
renderPrompt, strict (unknown/valueless/malformed references throw);
duplicate section and variable names rejected; assembly carries
resolved section text + variables through the assemble waterfall.
- dsh-agent declares AssembleContext.agent; dsh-agent-loop registers
the agent:persona section (order 0 - identity renders before tool
guidance) and the model/cwd variables, and drops its string join:
renderPrompt(assembly) IS the full prompt.
- Tool guidance moves to its owners: descriptions carry per-tool
semantics; sections only cross-call habits (tool:bash exit-code
habit at order 105; read's not-shell nudge). todo/subagent need no
section - their descriptions already carry the contract.
- SubagentProvider.inheritsParentContext (spawn/acp false, fork true);
dsh-tool-subagent derives truthful per-provider wording and resolves
the provider at load (backend must be listed first).
- Example personas shrink to identity + behavior with {{model}} (and
{{cwd}} in the ACP tree); the welcome banner stops enumerating tools.
RFC: docs/rfc/implemented/architecture/2026-07-05-prompt-variables-and-tool-guidance-ownership.md
dsh-system-prompt
System prompt assembly registry. Plugins contribute ordered text sections, tool-schema providers, and named prompt variables; the agent loop calls assemble(context) once per step, and renderPrompt(assembly) is the full system prompt the model sees.
Service: SystemPrompt (ctx key: systemPrompt)
Public API
ctx.systemPrompt.section(section: PromptSection): () => voidContribute a section. Duplicate names throw. Disposed with the calling fiber.ctx.systemPrompt.tools(provider: () => ToolSchema[]): () => voidContribute tool schemas (evaluated at each assembly). Disposed with the calling fiber.ctx.systemPrompt.variable(name: string, provider: (context) => string | undefined): () => voidContribute a prompt variable, referenced from section text as{{name}}. Duplicate or unreferenceable names throw;undefinedmeans "no value for this assembly". Disposed with the calling fiber.ctx.systemPrompt.assemble(context?: AssembleContext): Promise<PromptAssembly>Assemble the prompt for one caller. Runs through thesystem-prompt/assemblewaterfall.
Events
| Event | Mode | Purpose |
|---|---|---|
system-prompt/assemble |
waterfall | Mutate/extend the assembly (with the caller's context) before it reaches the model |
system-prompt/change |
emit | A section, tool provider, or variable was registered or unregistered |
Key types
AssembleContext— what oneassemble()call is FOR. Declared empty here and merge-extensible;dsh-agentdeclaresagent?: Agent, so providers project per-agent facts. Providers must tolerate absent fields (a bareassemble()carries an empty context).PromptSection—{ name, order, text: string | ((context) => string) }. Sections are concatenated in ascendingorder. Order bands:0is the per-agent persona (registered by the agent loop), tool guidance uses100–199; negative orders render before the persona.PromptAssembly—{ sections: AssembledSection[], tools: ToolSchema[], variables: Record<string, string | undefined> }. Section texts arrive resolved but not yet interpolated;variablesholds every registered variable resolved against the context. Tool schemas are part of the assembly by design: "what the model is told it can do" is one coherent thing, even though adapters transmit schemas as a separate wire field.renderPrompt(assembly)— interpolates{{variable}}references in each section, drops empty sections, joins with blank lines. STRICT: an unknown reference, a registered-but-valueless reference, or a malformed complete{{…}}group throws (fail loud beats shipping a malformed prompt). Only complete double-brace groups are interpreted; a lone{{passes through verbatim.
Merge-extensible: plugins can declare extra fields on PromptAssembly and AssembleContext via declaration merging.
Extension points
- Section providers: tool packages own their cross-call guidance (
tool:bash,tool:read, …); the agent loop ownsagent:persona. - Variable providers: the agent loop registers
modelandcwd; any plugin can register the facts it owns (a futuredate, git state, …). - Tool schema providers:
ToolRegistryregisters itself as a tool provider automatically. - The
system-prompt/assemblewaterfall: mutate or replace the assembly per caller (dynamic tool filtering, extra variables).
What is NOT here
- Any hardcoded prompt text — every section comes from plugins, every deployment-authored word from config.
- Prompt compaction (belongs on the
agent/pre-stepseam indsh-agent).
Design rationale: the prompt-variables RFC.