● l1vestack · IDP11 stages · 1 loop

An SDLC whereteam and agentsshare context

(translation pending)

Community OSS · Enterprise self-hosted · Cloud managed on h3llo
l1vestack · payments-api · stage 04
live
01Ideation
02Planning
03Architecture
04Development● in progress
05Code Review·
06Test & QA·
07Build·
08Release·
09Operate·
10Incident·
11Feedback·
h3 l1ve sync
$

It's not AI agents that share context. Team and agents — together.

When AI agents show up on every SDLC stage, time-to-market drops several fold and quality climbs. But only if humans and agents share the same context, rules, and practices. Not across 7 repos and Notion pages — in a single product.

alternative A

A stack of 7 products

  • Each product — best in its category
  • Confluence · Jira · GitHub · Argo · Grafana · Notion · Slack
  • Context scattered across 7 systems
  • AI agent only knows what's in one repo
  • Artifacts aren't connected: spec ↔ PR ↔ release ↔ incident — by hand
h3llo · l1vestack

L1veStack

  • One product, a closed 11-stage SDLC
  • Artifacts are connected: spec → PR → release → SLO → incident → action item
  • Shared context for humans AND an army of AI agents
  • Catalog, golden paths, GitOps, observability — out of the box
  • AI agents with MCP access and sandboxed runtime
  • DORA metrics automatically from git/CI/incident
alternative B

Backstage / Port

  • Rich service catalog
  • Portal-of-portals: 30 plugins, each with its own UX
  • There's a catalog — but it's not wired to the workflow
  • No agents — add them via your own plugin
  • app-config.yaml at 800 lines of YAML
● 11 stagesclosed loop

Full SDLC, with no gaps between stages

Each stage's artifact automatically becomes the next stage's input: issue → spec → ADR → PR → image digest → release → SLO → incident → action item → next issue.

01 / Ideation

Discovery

Ideas from chats, tickets, and calls are aggregated into a single backlog with semantic embeddings. RFCs are machine-readable.

RFC · brief · backlog
02 / Planning

Specification

Spec-driven: spec before code. Epic → story decomposition by an analyst agent. AC in Gherkin or structured YAML.

PRD · stories · DoD
03 / Architecture

Design

Architecture-as-code. ADR is mandatory. C4 from text + code analysis. SLOs are defined here, not after prod.

ADR · C4 · OpenAPI · SLO
04 / Development

Code

Trunk-based. Golden paths: a new microservice in 30 seconds. Agent runs in firecracker/gVisor with MCP access.

code · AGENTS.md · devcontainer
05 / Code Review

PR

AI reviewer as the first line: style, security smells, missing tests. Human review — for intent and architecture.

PR · AI review · CODEOWNERS
06 / Test & QA

Quality

Contract tests are mandatory. Self-healing flaky tests. Ephemeral envs per PR. Mutation testing on critical paths.

tests · SBOM · coverage
07 / Build

Package

Hermetic builds, distroless/Wolfi bases, SLSA L3+, Cosign-signed images, SBOM per artifact.

OCI · SBOM · provenance
08 / Release

Deploy

GitOps everywhere. Progressive delivery with auto-analysis against SLO. DB migrations expand/contract. Auto-rollback.

GitOps PR · canary · runbook
09 / Operate

Observe

OTel is the only standard. SLO-based alerting. AI agent summarizes alerts (logs + traces + recent deploys).

metrics · traces · dashboards
10 / Incident

Response

Blameless postmortems. An incident commander agent spins up a war room, ChatOps commands → actions, action items land in the tracker.

incident · postmortem · AIs
11 / Feedback

Loop

DORA from git/CI/incidents. DevEx surveys. AI clusters feedback into roadmap themes. Every action item has a trace.

DORA · DevEx · cost
the loop closes back into Ideation
● shared context

One file — for humans and agents

No separate AGENTS.md in one place, RFCs in another, runbooks in a third. Service context and rules live in .l1ve/ right next to the code, readable by humans and agents alike.

.l1ve/context.md · shared contextlive
# .l1ve/context.md — shared context for humans and agents
service:    payments-api
team:       billing
tier:       prod
on-call:    @alice (primary), @ivan (backup)

rules:
  - PR > 400 LOC requires a split (the splitter agent will suggest one)
  - migrations via Atlas, separate from deploy (expand/contract)
  - protected paths: infra/prod/, billing/secrets/

agents:
  - codegen:   sandboxed · MCP[git, fs, lsp, run-tests, docs]
  - reviewer:  read-only · MCP[git, semgrep, sast]
  - rca:       on incident · MCP[loki, prom, tempo, runbooks]

slo:
  - availability: 99.95%/30d  → burn alert >2%/h
  - p99-latency: 250ms        → page if >500ms 5min
.l1ve/agent.yaml · agent-as-catalog-entryv1
# .l1ve/agent.yaml — agent-as-catalog-entry
apiVersion: l1vestack.dev/v1
kind: Agent
metadata:
  name: codegen-payments
spec:
  base: openhands  # or aider, cline
  model:
    provider: h3llo-ai          # or anthropic, openai, ollama
    model: llama-3.1-70b
  context:
    rag:                         # corp knowledge layer
      - service:payments-api
      - adr:billing/*
      - runbook:billing/*
    spec_path: docs/specs/
  sandbox:
    runtime: firecracker
    network:
      egress: [github.com, registry.h3llo.cloud]
  mcp:
    - git           # scoped: read-only
    - filesystem    # scoped: ./
    - test-runner
    - docs-search   # ./docs/**
  budget:
    monthly_tokens: 10_000_000
    monthly_rub:    30_000
● use cases

Who needs this

01 / scaleups
Scaleups of 50–500 engineers
When 7 products start cracking at the seams and context falls apart. One loop — pulls the SDLC back together into a system.
02 / regulated
Regulated markets
Banks, public sector, healthcare. Tamper-evident audit log, air-gapped install, ready-to-use FZ-152/FSTEC/PCI policy bundles.
03 / agent-first
Agent-first teams
If AI agents are first-class citizens in your process. Sandboxed runtime, RAG, MCP gateway, evals — non-negotiable.
04 / platform
Platform teams
Internal teams as customers. Self-service over tickets, golden paths, measurable DevEx and DORA out of the box.
● 3 tiershealthy upgrade path

Community · Enterprise · Cloud

No crippleware. Community — a full SDLC for teams up to 20 people. Enterprise — when you need multi-tenancy and compliance. Cloud — network effects and premium agents that can't be reproduced locally.

Community
OSS release. Full 11-stage closed loop. A team of up to 20 should be able to close the whole SDLC without hitting limits.
0 ₽ · Apache-2.0
  • All 11 SDLC stages
  • Catalog · golden paths · scaffolder
  • CI (Tekton) · GitOps (Argo)
  • Observability glue · SLO-as-code
  • Basic AI agents + any LLM
  • MCP gateway self-hosted
self-hosted · single-tenant · pluggable OIDC
popular
Enterprise
Everything from Community + multi-tenancy, SSO/SCIM, tamper-evident audit log, air-gapped install, policy engine, agent fine-tuning.
from 89,000 ₽ / mo · 50 engineers
  • Multi-tenancy · org → team → project
  • RBAC/ABAC · SSO + SCIM
  • Signed audit log (append-only)
  • Air-gapped install · offline LLM bundle
  • OPA/Kyverno policy bundles
  • Bring-your-own-model + fine-tuning
  • 24×7 support · DR runbooks
SOC2 · ISO · PCI ready-made controls and reports
Cloud
Managed on h3llo cloud. Everything that can't be reproduced locally — cross-customer benchmarks, vuln intel feed, premium frontier agents on our GPU fleet.
from 2,400 ₽ / engineer / mo
  • Anonymized DORA percentile benchmarks
  • Vulnerability intel feed
  • Cross-customer flaky test patterns
  • Marketplace · agents · MCP servers
  • Premium frontier agents on B300
  • Hosted MCP gateway · curated catalog
  • Backups · DR · upgrades — on us
Hybrid mode available · control plane with us, data plane with you
● materialsfree

Guides and case studies on L1veStack

Real practices from the h3llo platform team: how we write specs for agents, how AGENTS.md is structured, what we measure once the SDLC is closed.

All materials →
● quickstart

Close the SDLC in a day

Helm install → ingest git → wire in agents → artifacts are connected automatically. No portal-builder and no 800 lines of YAML.

Download Community →
1

Download Community

helm install l1vestack h3llo/l1vestack · one command, up in 4 minutes.
2

Connect git

l1ve catalog ingest --from git@github.com/your/repos · services show up automatically from catalog-info.yaml.
3

Wire in agents

Drop .l1ve/agent.yaml into the repo · the agent registers itself in the catalog with its runtime and MCP access.
4

Close the loop

Spec → code → review → release → operate → feedback → next spec. Artifacts are linked automatically.
● faq

What people usually ask

Is this Backstage or not?
Under the hood we use three Backstage backend plugins (Catalog, Scaffolder, TechDocs) — and nothing else. No app-config.yaml, no Material-UI. On top — our own GraphQL gateway and UI. A L1veStack user never sees the word "Backstage". This is a connected system, not a portal-of-portals.
How does Community differ from Enterprise and Cloud?
Community (OSS) — a full closed 11-stage loop, basic AI agents, any LLM provider. Enough for a team up to 20 people. Enterprise adds multi-tenancy, RBAC/ABAC, SSO+SCIM, tamper-evident audit log, air-gapped install, policy engine. Cloud — managed on h3llo + network effects: cross-customer benchmarks, vuln intel feed, marketplace, premium frontier agents on our GPU fleet.
Which AI agents ship out of the box?
Codegen (code generation from spec), Reviewer (first-line code review), RCA (incident analysis), Triage (classifying ideas and issues), Diagram (C4 from text and code), Spec-helper (epic decomposition), Threat-model (STRIDE), Incident commander. All in a sandboxed runtime (firecracker/gVisor), with restricted network egress via the MCP gateway.
Can we plug in our own LLMs or do we have to use h3llo AI?
Any of them. In Community — OpenAI, Anthropic, Bedrock, local Ollama / vLLM. In Enterprise — a fine-tuning pipeline for agents on your corporate data. In Cloud — premium frontier models (Claude Opus, GPT-5 class) on our GPUs plus the option to plug in your own.
What about air-gapped installation?
Enterprise includes an offline registry and an offline LLM bundle. No outbound traffic — no LLM calls, no telemetry, no updates. Suitable for regulated perimeters and the public sector.
Hybrid mode — how does that work?
The control plane lives in our Cloud (UI, catalog, agent registry, knowledge layer), while the data plane (workloads, secrets, sources, build runners) lives with you. You get Cloud DevEx, prod data never leaves your perimeter. Available from Enterprise and Cloud.
● one context · one loop

An SDLC where
nobody loses context

An army of AI agents on every stage · DORA through the roof · TTM through the floor. Not across 7 SaaS tools, but in a single product — bring your own model and your own perimeter.

Download Community →Talk to an architect