@ostinato/aionis
v0.4.0
Published
Complete Aionis Core SDK surface for runtime, bridge, handoff, replay, governance, and execution memory integration.
Maintainers
Readme
@ostinato/aionis
@ostinato/aionis is the public SDK for integrating Aionis Runtime into a coding agent or local agent runtime.
Current release line: 0.4.0 for the Aionis Lite Developer Preview.
Use it when you want an agent to:
- start repeated tasks with a better first action
- rehydrate archived execution memory back into the active working set
- record node reuse outcomes on Lite memory nodes
- store and recover structured handoffs
- turn successful runs into replayable playbooks
- inspect continuity state through typed runtime contracts
Install
npm install @ostinato/aionisDefault Product Path
| Path | What To Prove | Primary Surfaces |
| --- | --- | --- |
| Core | Continuity works at all | memory.write(...), memory.taskStart(...) or memory.planningContext(...), handoff.store(...), memory.replay.run.* |
| Enhanced | Continuity improves over time | memory.archive.rehydrate(...), memory.nodes.activate(...), memory.reviewPacks.*, memory.sessions.* |
| Advanced | The runtime exposes deeper learning and control | memory.experienceIntelligence(...), memory.executionIntrospect(...), memory.delegationRecords.*, memory.tools.*, memory.rules.*, memory.patterns.* |
Recommended order:
- prove the Core path first
- add the Enhanced path when reuse quality matters
- move into the Advanced path only when your host needs deeper substrate controls
Fastest repository proof:
npm run example:sdk:core-pathWhat The SDK Covers
- memory write / recall / planning / context assembly
- explicit action retrieval, uncertainty, and operator-facing gate surfaces
- kickoff recommendation, task-start, task-start-plan, execution outcome, and workflow-contract facades
- archive rehydrate and node activation lifecycle surfaces
- replay run lifecycle and playbook operations
- handoff storage and recovery
- automation definitions and runs
- sandbox session, execution, and run inspection
- sessions, packs, review packs, rule/tool operator surfaces
- agent-memory inspect, review, resume, and handoff packs
- host bridge integration
Usage
import { createAionisClient } from "@ostinato/aionis";
const aionis = createAionisClient({
baseUrl: "http://127.0.0.1:3001",
});
const taskStart = await aionis.memory.taskStart({
tenant_id: "default",
scope: "default",
query_text: "debug the failed replay run",
context: {
goal: "debug the failed replay run",
},
candidates: ["read", "bash", "edit", "test"],
});
console.log(taskStart.first_action);
const taskStartPlan = await aionis.memory.taskStartPlan({
tenant_id: "default",
scope: "default",
query_text: "debug the failed replay run",
context: {
goal: "debug the failed replay run",
},
candidates: ["read", "bash", "edit", "test"],
});
console.log(taskStartPlan.first_action);
console.log(taskStartPlan.gate_action);After execution, a host can store the validated outcome and optionally compile/simulate a replay playbook without manually stitching replay routes:
const outcome = await aionis.memory.storeExecutionOutcome({
tenant_id: "default",
scope: "default",
actor: "sdk-host",
goal: "debug the failed replay run",
status: "success",
steps: [{
tool_name: "bash",
tool_input: { argv: ["npm", "run", "-s", "test:lite"] },
status: "success",
output_signature: { summary: "lite validation passed" },
}],
compile_playbook: true,
simulate_playbook: true,
});
const workflow = await aionis.memory.retrieveWorkflowContract({
tenant_id: "default",
scope: "default",
file_path: "src/routes/replay.ts",
});
console.log(outcome.run_id);
console.log(workflow.authority_summary);retrieveWorkflowContract returns the stable workflow contract and authority summary by default. The full execution
introspection payload is debug-only; request it explicitly with include_introspection: true when building operator
diagnostics.
Explicit action-retrieval example:
const retrieval = await aionis.memory.actionRetrieval({
tenant_id: "default",
scope: "default",
query_text: "debug the failed replay run",
context: {
goal: "debug the failed replay run",
},
candidates: ["read", "bash", "edit", "test"],
});
console.log(retrieval.selected_tool);
console.log(retrieval.recommended_next_action);
console.log(retrieval.uncertainty);Lite runtime lifecycle example:
await aionis.memory.archive.rehydrate({
tenant_id: "default",
scope: "demo-sdk",
client_ids: ["billing-timeout-repair"],
target_tier: "warm",
reason: "bring the archived repair memory back into the active set",
});
await aionis.memory.nodes.activate({
tenant_id: "default",
scope: "demo-sdk",
client_ids: ["billing-timeout-repair"],
outcome: "positive",
activate: true,
reason: "the rehydrated node helped complete the repair",
});Local Workflow
Recommended standalone runtime path:
npx @ostinato/aionis-runtime startIf you are working from this repository and want the bundled examples:
cd /path/to/AionisRuntime
npm install
npm run sdk:build
npm run lite:startThen try one of the bundled examples:
npm run example:sdk:core-path
npm run example:sdk:recall
npm run example:sdk:replay
npm run example:sdk:sessions
npm run example:sdk:automation
npm run example:sdk:sandbox
npm run example:sdk:host-bridge
npm run example:sdk:agent-memory
npm run example:sdk:action-retrieval
npm run example:sdk:continuity-provenanceProof demos for the self-evolving claim:
npm run example:sdk:task-start-proof
npm run example:sdk:policy-memory
npm run example:sdk:policy-governance
WORKFLOW_GOVERNANCE_STATIC_PROMOTE_MEMORY_PROVIDER_ENABLED=true npm run lite:start
npm run example:sdk:continuity-provenance
npm run example:sdk:session-continuity
npm run example:sdk:semantic-forgetting
npm run example:sdk:dogfood-loop
npm run example:sdk:service-lifecycle-dogfoodAgent-memory façade example:
const inspect = await aionis.memory.agent.inspect({
tenant_id: "default",
scope: "demo-sdk",
query_text: "repair export mismatch in src/routes/export.ts",
context: {
goal: "repair export mismatch in src/routes/export.ts",
},
candidates: ["bash", "edit", "test"],
file_path: "src/routes/export.ts",
});
console.log(inspect.agent_memory_summary);Next Reads
Release Checks
npm run sdk:test
npm run sdk:pack:dry-run
npm run sdk:publish:dry-run
npm run sdk:release:check