work
4 次代码提交
| 作者 | SHA1 | 备注 | 提交日期 | |
|---|---|---|---|---|
 Asim Aslam
|
844ab4a14b |
Add provider-conformance harness, contract test, and reframe docs to 'agent harness' (#3006)
* docs: reposition go micro as agent harness * ci: rename universe workflow to harness |
||
|
Asim Aslam
|
9fdcc24cce |
Implement durable execution and scoped state management for flows (#2972)
* docs: design note for flow steps + Checkpoint durable execution
* docs: fold in durable-execution decisions (State struct, single Step, run retention, retry)
* docs: rename State.Payload to State.Data
* feat(flow): ordered steps + Checkpoint durable execution
A flow can now be an ordered list of steps (a task with stages) instead
of a single LLM turn. State carries typed Data plus a Stage marker; each
step is checkpointed before and after via a pluggable Checkpoint
(store-backed by default), so a run survives a crash and resumes where it
stopped without re-running completed steps. Flow-level Retry with a
per-step override; runs retained for audit unless DeleteOnSuccess.
Step actions: Call (RPC), LLM (augmented turn), Dispatch (to an agent),
or any StepFunc. Single-step and agent-dispatch flows are unchanged.
* feat(flow): top-level re-exports + durable flow example
Expose the step/checkpoint API from the micro package (FlowSteps,
FlowStep, FlowState, FlowRetry, FlowWithCheckpoint, FlowCall/LLM/Dispatch,
Checkpoint, StoreCheckpoint) and add a runnable, key-free example
demonstrating crash + resume.
* docs: document durable flow steps (guide, README, CLI help)
* docs: blog post + changelog for durable workflows
* fix(flow): scope checkpoint keys by flow name (flow/{name}/runs/{id})
Run keys were flow/runs/{id} — a single global keyspace shared by every
flow on the default store. Namespace them by flow name so each flow's
state is kept apart. StoreCheckpoint now takes a scope argument (the flow
passes its name by default).
* feat(store): Scope handle; scope agent and flow state by name
Add store.Scope(s, database, table) — a store handle that confines every
operation to a database/table without mutating the shared store, so
co-located components don't clobber each other's table (the failure mode
of the global Init(Table(...)) approach).
Use it to keep each agent's memory and plan in its own table
(agent/{name}) and each flow's runs in its own (flow/{name}), instead of
one global table partitioned only by key prefix. Services already scope
by service name.
* feat: consistent state model — service store scoping, flow registry, list/history CLI
- service: scope store via store.Scope (database service / table name),
retiring the Init(store.Table(name)) global-mutation hack; bridge the
default store so handlers using store.DefaultStore stay isolated.
- flow: register in the registry as type=flow while running (with trigger
and step count), deregister on Stop. Live discovery, like agents.
- cli: micro flow list (registry), micro flow runs <name> (durable store),
micro agent history <name> (durable store). list = running, runs/history
= durable, mirroring the service model.
* test: mini-universe end-to-end harness + scheduled GitHub Action
internal/harness/universe boots a small but real go-micro world — four
services, a durable checkout flow that crashes at payment and resumes,
and a guardrailed agent with a tool wrapper reached over RPC — drives the
scenario, asserts the end state (10 checks), and shuts down. Everything
is real except the LLM (mocked), so it's deterministic and needs no key;
-provider anthropic runs it live. Exits non-zero on failure, so it's an
end-to-end test, not just a demo.
Adds .github/workflows/universe.yml (push/PR/daily/dispatch) running the
universe + existing harnesses on the mock provider, plus an opt-in job
that runs live when ANTHROPIC_API_KEY is set. 'make harness' runs them
locally.
* ci: run the live universe job against AtlasCloud (ATLASCLOUD_API_KEY)
* ci: run the live universe job only on schedule or manual dispatch
The deterministic mock job still runs on push/PR/daily; the live
(AtlasCloud) job runs daily and on manual workflow_dispatch only, so
changes don't burn API credits on every PR but can still be checked
against a real model on demand.
---------
Co-authored-by: Claude <noreply@anthropic.com>
|
||
|
Asim Aslam
|
d3be610367 |
Refactor README features and add end-to-end flow testing (#2954)
goreleaser / goreleaser (push) Has been cancelled
* docs: group README features by section (AI / Framework / DX) The features table repeated 'AI' down the Category column. Split into three grouped tables — AI, Framework, Developer experience & deployment — dropping the repetitive column. Adds a Guardrails row (MaxSteps, ApproveTool). * test: flow-to-agent end-to-end in the harness Proves 'Flow triggers, Agent reasons': a workflow with FlowAgent hands an event to the registered conductor agent over RPC, which plans, creates tasks, and delegates to comms — the whole chain over real RPC with only the LLM mocked. Deterministic (shared in-memory registry, no sleeps), passes under -race. * blog: 'The Evolution of Microservices' (#19) A technical history of distributed-systems eras — the monolith's coordination cost, the distributed-systems tax, containers and declarative orchestration, the service mesh, and the modular-monolith correction — establishing the durable unit (named, typed, discoverable, independently deployable) that every runtime wave required. Then the technical argument for agents: an LLM tool call needs exactly a service interface, so the caller shifts from deterministic code to a reasoner that composes typed capabilities from intent, with the honest caveats (non-determinism, cost, guardrails). Not a product pitch. --------- Co-authored-by: Claude <noreply@anthropic.com> |
||
|
Asim Aslam
|
e416ea4a75 |
Enhance agent workflows with guardrails and documentation updates (#2952)
* docs: map go-micro onto Anthropic's workflows-vs-agents taxonomy - new guide 'Agents and Workflows': adopts Anthropic's Building Effective Agents vocabulary — workflow (predefined path) = flow, agent (dynamic self-direction) = agent — maps the augmented-LLM building block and the five workflow patterns onto go-micro, and shows routing (chat router) and orchestrator-workers (conductor + plan/delegate) are already native. - flow package doc reframed as a workflow (predefined path) per the same taxonomy, with guidance on flow vs agent. - nav + README link the new guide. * feat: agent guardrails — step limit and tool approval hook Anthropic's Building Effective Agents stresses stopping conditions and human-in-the-loop checkpoints for autonomous agents. Add both as plain options enforced at the tool-handler choke point — no provider changes, no new abstraction: - MaxSteps(n): bound tool executions per Ask; beyond the limit, actions are refused and the model is told to stop and summarize. - ApproveTool(fn): gate each action before it runs; returning false blocks it and surfaces the reason to the model. The internal plan tool is never gated. Exposed at the micro package (AgentMaxSteps, AgentApproveTool, ApproveFunc). Tests cover the limit, blocking, and that plan is not gated. Guardrails section of the agents-and-workflows guide updated from 'active work' to documented options. * feat: flow can dispatch to an agent (flow triggers, agent reasons) Unify the engine without collapsing the workflow/agent distinction. A Flow with Agent set hands each event's rendered prompt to a named registered agent over RPC (Agent.Chat) instead of running its own LLM step — so the workflow stays the deterministic trigger and the agent is the reasoning engine, with its plan, delegate, memory, and guardrails. A plain flow is unchanged (single augmented-LLM step). - flow.Agent(name) / micro.FlowAgent(name); flow stores the client and skips model setup when dispatching. - test: dispatch routes to comms.Agent.Chat with the rendered prompt and records the reply. - guide: 'Flow triggers, Agent reasons' section. --------- Co-authored-by: Claude <noreply@anthropic.com> |