Deploy agents for feature development, code review, and more. Each agent gets an isolated sandbox, scoped permissions, and hard budget caps. Full observability included. Get started with a $100 credit.
Claude
Codex
GeminiTrusted by 400+ engineering teams
Each agent runs in its own sandbox with scoped credentials and a hard budget. Developers get parallelism, the platform team gets the audit trail.
$ agent session start "backfill the rest-period audit table"
using config "Feature builder" (repo default)
session_4Kp9dQ2w · sandbox ready in 8s
$ agent session list
session_4Kp9dQ2wrunningFeature builder$0.62
session_8Rn2tY6crunningFeature builder$1.18
session_2Wm7xB4hrunningMigration runner$0.34
session_9Jf5cV3kcompletedIssue triage$0.87
$ agent session search "tip-pool rounding"
session_9Jf5cV3k · opened PR #514 · 3 days ago
Choose the models and prompts, add multiple reviewers, and gate what lands with a filter you write. Too noisy? Write a stricter prompt.
Combinator
ellipsis: version: v1 kind: code_reviewreview: - name: bugs claude: model: claude-opus-5 system: | Flag correctness bugs and unhandled edge cases, on the exact line. - name: migrations pull_requests: paths: ["migrations/**"] claude: system: | Check every schema change for lock risk and backfill safety.filter: name: gatekeeper claude: system: | Post only findings a senior engineer would raise.budget: run: 10.00An agent is a YAML file in your repo: the prompt, the model, the sandbox, the permissions, the budget. Like infrastructure-as-code, for your agents.
ellipsis: version: v1 name: Dependency triageclaude: model: claude-opus-5 system: | Update dependencies with failing audits and open a pull request.trigger: type: cron schedule: "0 6 * * 1-5"sandbox: repositories: - name: splitshift-api github: permissions: contents: read pull_requests: writebudget: session: 5.00Ellipsis is the infrastructure under whichever agent wins: sandboxes, credentials, budgets, and logs stay the same while the harness and model are config fields.
claude: model: claude-opus-5 system: | Reproduce the Sentry alert and open a fix.codex: model: bedrock-gpt-5.6-terra system: | Reproduce the Sentry alert and open a fix.Same sandbox, credentials, budget, and logs either way.
Each agent spawns the full stack and runs the tests in its own sandbox, so ten agents on one repo never clobber each other.
sandbox: repositories: - name: splitshift-web - name: splitshift-api image: setup: | poetry install && npm ci compute: cpu: 4 memory: 8GB● sbox_kF82mQ4r · image cache hit · ready in 8s
● 10 sessions on splitshift-web · isolated sandboxes
Agents stop running on developers’ own credentials. Permissions are minted per session, spend is capped and attributed, and the transcript outlives the sandbox.
Self-host in your AWS account and nothing leaves your VPC. Model calls can route through your own gateway for compliance and logging.
Your VPC
● Ellipsis platform · sandboxes · session logs
● model calls → your Bedrock account
● or routed through your own LLM gateway
outbound data to ellipsis.dev: none
Put an LLM decision-maker anywhere in your SDLC with no infra to manage. Spawn a session from any workflow and get typed, structured output back.
Learn more →curl https://api.ellipsis.dev/v1/sessions \
-H "Authorization: Bearer $ELLIPSIS_API_KEY" \
-d '{
"config_id": "agent_x9Kd3Fq2",
"prompt": "Classify this incident and
draft the postmortem outline"
}'{
"id": "session_4Kp9dQ2w…",
"status": "scheduled",
"resolved_config_name": "Incident triage"
}The platform
From the YAML that defines an agent to the audit trail it leaves behind.
Why do agents need a cloud? →Latest
Ellipsis started in 2023 reviewing code and fixing bugs. Today we run managed coding agents: defined in your repo, deployed with git push, governed with permissions, transcripts, and hard spend caps.
Read post →The future of software is agents defined in code: versioned behavior, isolated cloud sessions, scoped permissions, and human judgment at the checkpoints that matter.
Read post →What we learned shipping autonomous coding agents to hundreds of engineering teams. Sandboxing, context management, cost control, and the things that actually matter.
Read post →Write the config, push to GitHub, and Ellipsis handles the rest.