Live execution
See the run move from queued to starting, running, and a final outcome, with runtime and counts of AI calls, tool calls and issues updating live.
Sulcus Cloud
See the whole agent run while it happens.
Workflow steps, model and tool calls, token usage and failures, live as your agent runs in Sulcus Cloud. Stop the run at any time, and keep an exact record of what executed.
Start freeAvailable now
Sulcus Cloud Run Details for a running LangGraph workflow: 14 AI calls, 9 tool calls, 1 minute 12 seconds of runtime, 38,420 of a 50,000 token limit used, no issues, a workflow tree of graph steps, a live activity feed, and a Stop run action.
Visibility
Follow the run as it happens.
Events stream into Run Details while the agent works. Reload the page or come back later and the history is still there.
Workflow and agent steps
A workflow tree shows the structure your framework exposes: graph nodes in LangGraph, crew tasks and agents in CrewAI, agents and handoffs in the OpenAI Agents SDK.
Depth varies by framework; CrewAI’s tree is shallower than LangGraph’s.
Model and tool activity
Every model call and tool call appears in the Activity feed as it starts, completes or fails, filterable by agents, LLM, tools, errors and system events.
Token usage
Sulcus totals the input, output and total tokens your framework reports, per call and for the whole run.
Unreported usage stays unknown. Sulcus shows tokens, not dollar costs.
Control
Stop the run. Cap its tokens.
Sulcus Cloud starts the container your agent runs in, so it can also end it.
- Stop run ends the run and its container.
- Token limit is optional and set per run. Once reported usage reaches it, Sulcus blocks the next model call and ends the run.
- Platform limits give every run fixed CPU, memory and execution time. They aren’t configurable today.
The limit acts on reported usage, so a request already in flight can finish and go slightly over.
AI usage21,380 / 50,00028,620 remaining
AI usage43,910 / 50,000Approaching limit
AI usage51,204 / 50,000Limit exceeded by 1,204 · run ended
Failure analysis
When a run fails, see why.
Run Details groups errors into a named cause, explains it in plain language, and keeps the technical details, logs and the full activity leading up to the failure.
FAILURE CATEGORIES
Provider usage limits are reported as the provider’s, not as a Sulcus intervention.
Exact run history
The project moves on.
The run record doesn’t.
Every run keeps the exact Git commit Sulcus resolved and a snapshot of the configuration it started with: repository, reference, entrypoint, framework and dependency mode.
Change the project tomorrow and the next run uses the new settings. Yesterday’s run still shows what actually executed.
Privacy
Behavior, not content.
No prompts, outputs or tool payloads
Framework instrumentation records which model and tool calls ran, their timing and reported tokens. It deliberately leaves out prompts, model outputs and tool inputs and outputs.
Secrets per run, not stored
API keys and other runtime secrets are supplied when you start a run. They aren’t saved with the project or the run, and known values are redacted from events and logs on a best-effort basis.
Developer Preview
Local Codex and Claude Code sessions.
Launch Codex or Claude Code locally through the Sulcus CLI and follow the session in Sulcus Cloud, with privacy-reduced telemetry. This is observation only: Cloud can’t stop the local agent or approve or deny its actions.
Ask about the Developer PreviewBuilt for today
Current product boundaries
Public HTTPS repositories
Private repositories, SSH URLs and GitHub App access aren’t supported yet.
Python and three frameworks
Cloud runs support Python apps built with LangGraph, CrewAI or the OpenAI Agents SDK.
Trusted code
The container limits resources and filesystem access, but it isn’t a sandbox for hostile code. Run code you trust.
No approvals or policies in Cloud
Sulcus Cloud stops runs and enforces token limits. It doesn’t offer approval workflows or configurable tool policies.