rustakka

Use when turning a diarized STT transcript into structured meeting artifacts in atomr-agents — building a `MeetingsHarness`, accumulating a `MeetingAnalysis` (attendees, notes, actions-with-owners, tiered summaries), picking batch vs live mode, persisting via the configured `Checkpointer` under the same `conversation_id` as the source transcript, or serving the meetings review UI (`meetings-harness-web`). Triggers on `MeetingsHarness`, `MeetingAnalysis`, "extract action items", "meeting summary", "attendees from transcript", "live meeting analysis", `atomr_agents.meetings_harness`.

Use when driving speech-to-text as an agentic streaming pipeline in atomr-agents — building an `SttHarness`, accumulating an `SttConversation`, feeding it to an agent, choosing a `DiarizationPolicy`, persisting conversations via a `ConversationStore`, or serving the diarized-transcript review UI (`stt-harness-web`). Triggers on `SttHarness`, `SttConversation`, `AudioSource`, `DiarizationPolicy`, "diarize a recording", "editable speaker labels", "transcript review UI", `atomr_agents.stt_harness`.

Use when wrapping the agent's per-turn pipeline with cross-cutting policies — logging, retry, rate-limit, redaction, tool-error recovery, dynamic prompt override, before/after hooks. Triggers on `impl AgentMiddleware for`, `MiddlewareStack::new().push(...)`, or porting a LangChain `create_agent` middleware (`@wrap_model_call` / `@wrap_tool_call` / `@dynamic_prompt`).

Use when wiring tracing into an agent / workflow / harness — `EventBus` subscribers, `RunTreeBuilder` for parent-child run trees, `StdoutTracer` / `JsonlTracer` / `LangSmithTracer` exporters, or propagating `RunId` / `parent_run_id` through events. Triggers on `EventBus::new`, `RunTreeBuilder::new`, `LangSmithTracer::new`, `bus.subscribe(...)`, or `bus.emit_run(...)`.

Use when composing `Callable`s with the `Pipeline` builder or wrapping any handle in retry / fallback / config / timeout / branch decorators. Triggers on writing `Pipeline::from(...).then(...)`, calling `with_retry` / `with_fallbacks` / `with_config` / `with_timeout` / `Branch::new`, fan-out into a JSON object, or migrating a LangChain LCEL chain.

Use when standing up the first atomr-agents project, picking feature flags for the `atomr-agents` umbrella, building a first `Agent`, or running an end-to-end agent turn against `MockRunner`. Triggers on adding `atomr-agents = ...` to Cargo.toml, writing the first `Agent { ... }` literal, wiring an `InferenceClient`, or asking "how do I get atomr-agents running".

Use when building a retriever pipeline (BM25 / dense / hybrid / contextual-compression / parent-document / self-query / time-weighted), ingesting documents (loaders + splitters), or wiring `LongStore` into an agent. Triggers on `Bm25Retriever::new`, `VectorRetriever::new`, `EnsembleRetriever::with_rrf`, `RecursiveCharacterSplitter`, `LongStore::put`, or porting a LangChain retriever zoo.

Use when designing or modifying channelled state — declaring a `StateSchema`, picking a reducer, persisting via `Checkpointer`, or forking a divergent run. Triggers on writing `StateSchema::builder()`, `RunState::new()`, `Checkpointer::save / latest / fork`, or porting a LangGraph `StateGraph(MyState)` definition.

Use when deploying atomr-infer to a cluster — choosing feature flags, the `remote-only` invariant, `atomr-infer serve --config` project files, hot-swap and credential rotation, container image sizes. Triggers on writing a `inference.toml`, a Dockerfile for an inference service, configuring `kubectl apply` / Helm, or asking "how do I deploy atomr-infer to production".

Use when adding a new backend to atomr-infer — implementing `ModelRunner`, plugging into the rollup, slotting a new crate into the publish dep-order. Triggers on writing `impl ModelRunner for ...`, asking "how do I add Bedrock / Cohere / a custom CUDA kernel package", or considering a fork.

Use when composing multi-runtime pipelines in atomr-infer — hybrid local→remote escalation, fallback on `RateLimitExceeded` / `CircuitOpen`, dynamic batching, cascade routing, replica pools, hot-swap, speculative decoding, MoE. Triggers on writing an actor that calls multiple `Deployment`s, using `inference::accel_patterns::*`, or asking "how do I escalate from local to OpenAI when confidence is low".

Use when standing up the first atomr-infer deployment in a consumer project, choosing feature flags for the `inference` rollup, or writing a `Deployment` value object. Triggers on adding `inference = ...` to Cargo.toml, writing a `Deployment {...}` literal, configuring `atomr-infer serve`, or asking "how do I get atomr-infer running".

Use when choosing a backend for a model deployment in atomr-infer, configuring `RuntimeConfig`, deciding between local Rust-native (Candle / cudarc / mistralrs) vs Python (vLLM) vs FFI (TensorRT / ORT) vs remote (OpenAI / Anthropic / Gemini / LiteLLM). Triggers on writing a `Deployment.runtime = ...` field, choosing a feature flag, asking "what's the right backend for X model".

Use when debugging atomr-infer symptoms — `InferenceError::RateLimited` storms, `CircuitOpen`, `ContentFiltered`, `CudaContextPoisoned`, mailbox backpressure, gateway 429s, missing fallbacks, the `remote-only` invariant violation. Triggers on a stack trace mentioning `inference-*`, an unexplained typed error, or asking "why is my deployment returning X".

Use when standing up zero-config local Gemma 4 on a workstation with a working GPU + Python + vLLM, picking a Gemma 4 variant (E2B / E4B, base vs `-it`), or wiring the `gemma-default` feature. Triggers on enabling `--features gemma-default`, setting `ATOMR_INFER_GEMMA_*` env vars, asking "how do I run Gemma locally", or seeing a `gemma-local` deployment auto-appear.

Use when wiring a remote inference provider (OpenAI / Anthropic / Gemini / LiteLLM) in a atomr-infer project — credentials, rate limits, retries, circuit breakers, cost estimation, fallback chains. Triggers on configuring `OpenAiConfig` / `AnthropicConfig` / `GeminiConfig` / `LiteLlmConfig`, handling 429 / `InferenceError::RateLimited` / `CircuitOpen`, or asking "how do I add OpenAI to my deployment".