The values fixed in Project in the Marketplace edition. Values that are the same as in the SaaS edition are not listed. For items not listed, see the SaaS edition's Settings Reference. Item names follow what the screen shows.
Project
| Item | Value |
|---|
| Default type | Workflow (Cycle in the SaaS edition) |
| How to create it | On the creation screen: “Build it myself”, “Create from a company template” and “Import from file”. Creating through consultation with the AI, by guided setup, or from an existing project is not on the creation screen (an existing project can be copied with “Duplicate” or “Copy to my projects”) |
| Number of projects | Up to 5 in one environment (deleted and archived ones are not counted). Creating or restoring from archive a project that would be the 6th is refused |
| Retries per step (failed / timed out) | The screen lets you choose 0 to 10, but on a registered agent's step up to 2 take effect. It retries automatically only when the contract's “Effect on external systems” is “Read only” or “Writes (safe to retry)”, and only for failures where what happened is known. Once the retries are used up, the case closes as failed (Trigger Guide) |
Steps
| Item | Value |
|---|
| Agents that can be placed on steps (workflows) | Only certified registered agents with a workflow contract registered in the admin console |
| Agents that can be placed on steps (cycles) | Registered agents enabled in the admin console (agents used in chat). No contract or certification is needed |
| Parallel lanes, Run at the same time, Destination options, destinations by result | Not available. Refused on saving or starting |
| Branch (workflows) | The same as the SaaS edition (SaaS edition Settings Reference), except that there is no “question” to check |
| Handling of results (when not allowed by rules, when the agent reports it could not finish) | There is no setting to choose it on the screen. It runs with a fixed handling (only “when not allowed by rules” can be set to “close as failed” through the external API. Trigger Guide) |
Step settings (workflows)
| Item | Value |
|---|
| Parameters | Of the contract's settings items, those of type string, integer, number, yes/no and choice can be changed on the screen. Only the items you enter a value for are saved; items left blank keep the agent's default values. Items of other types are settled by the administrator in the contract |
| Connectors | For each connection slot in the contract, choose one of this project's connectors. Only the connectors the administrator allowed for the slot can be chosen |
| Input bindings | Bind the output of an agent on a preceding step as this agent's input file. Files not bound are passed when the case starts |
| Execution limit | Settled by the contract. 900 seconds by default. 0 means no limit (up to the deployment's limit; 24 hours by default). It cannot be changed in the project |
Environment variables (cycles)
| Item | Value |
|---|
| Number | Up to 32 per placement |
| Size | Names and values totalling up to 16 KiB |
| Name | Starts with an uppercase letter and uses uppercase letters, digits and _, up to 64 characters |
| Names that cannot be used | Reserved names (names starting with ASTER_ or MP_, PATH and so on), names registered as secrets for the agent, names the administrator has blocked from change |
Review steps
| Item | Value |
|---|
| Review method | “Stop and wait for a decision” or “Wait for an external condition” |
| When the deadline passes | Not decided automatically. Even after the deadline it waits for the reviewers' decision, and notifies that it is overdue |
| Destination options | Not available |
| External condition | One kind: “GitLab merge request merged” |
| Interval for observing the external condition | 300 seconds. When a webhook from the connector is received, it is also read again at that point |
| External-wait webhook secret | 16 to 255 characters. The saved value is not shown |
Cases (workflows)
| Item | Value |
|---|
| Size of artifacts | Up to 20 MiB per file. Up to 100 MiB and 100 files in total per execution |
| Artifacts that can be edited on the screen | Text up to 1 MiB. YAML and JSON are checked for syntax and for the shape set by the contract before they are saved |
| When the settings and artifacts can be changed | Only while awaiting review. They cannot be changed while running or in a finished case |
| Starting a case | Project participants (owners, operators and members) and people with the organization operations role |
| Cancelling a case | Owners, operators and people with the organization operations role, and the person who started that case |
| Stopping a cycle execution | Owners, operators and people with the organization operations role |
| Cases that can run at once | Up to 3 (running plus awaiting review). Beyond that, manual starts and the external API are refused, and runs due from a time trigger or a step time are skipped (Trigger Guide) |
| Default run deadline | Hours from the start. 1 to 2,160 hours are accepted. The default is “None”. The case does not stop when it passes; it appears in “Action needed” |
| Hours before a run counts as stalled | The default is 24 hours. 1 to 720 hours are accepted. A running case (not one awaiting review, one for which cancellation was requested, or one past its deadline; a case past its deadline appears as “Run deadline exceeded”) that has not progressed for this long since the last step finished (or since the start, if none has) appears in “Action needed” |
| Editing a case's settings and artifacts | Owners only |
Triggers
| Item | Value |
|---|
| Start by email | Not available |
| External API | Can be called with a service account token (System Integration API Reference) and with the token of a user who signed in from an external tool (User API Reference (Projects)). When starting a run through the external API, parameters, connectors and input files can be passed at the start (config). They cannot be passed to a case started from a time trigger or a webhook trigger call (payload of the external API is passed to the agent of the first step as a request) |
| Overlapping sweeps (cycles) | A round that comes while the same placement is running is skipped (not piled up) |
| Late sweeps (cycles) | A round more than 5 minutes past its schedule is recorded once as skipped, and it moves on to the next scheduled time |
| Before the same placement's next start (cycles) | 60 seconds after the previous execution ends. Sweeps during that time are skipped; webhook calls wait and then start |
| Webhooks (cycles) | Calls received are not merged; each is passed to the same placement one at a time, in order. Up to 20 can wait per placement (beyond that, 429 queue_full). A call more than 24 hours past being accepted is not started |
Execution rules
| Item | Value |
|---|
| Whole-environment basic rule | Decided on “Execution rules” in the admin console. When it is not set, runs proceed under Open. A new version takes effect from the runs that start after it, and the version applied to past runs does not change |
| Basic rule | Open / Guarded / Strict. The tenant-wide basic rule and the project's basic rule are combined, and the stricter one becomes that run's basic rule |
| Where the basic rule applies | Workflows only. Before launch, it checks whether the rule is among the “supported run rules” in the agent's certification. It has no effect in a cycle |
| Prohibited operations added in a project | Not passed to the agent (combined with the tenant-wide rules before launch and fixed to that run) |
| Outbound traffic | A setting for the whole environment. Narrowed with “Banned URL” in the admin console and with a NetworkPolicy (Execution Rules and Network Access Guide) |
Action needed
| Item | Value |
|---|
| Kinds | Eight: awaiting approval, overdue, waiting on external, failed, awaiting rework, sweep failed, past deadline (run), stalled |
| Priority | P1 to P4. Decided mechanically and cannot be changed by hand (Record Guide) |
| Rows listed at once | Up to 100 |
| Bulk actions | Approve, reject, send back, extend deadline, reassign approvers. There is no “Rerun sweep”. Approving, rejecting, and sending back review requests for cases cannot be done in bulk |
How to write a custom (cron) schedule
The ways to choose a time are Interval, Hourly, Daily, Weekdays, Weekly, and Custom (cron). For an interval, you can specify only a number of minutes that divides 60 when it is under 60 minutes, and a number of hours that divides 24 when it is 60 minutes or more. You also specify the time zone. A time trigger, a step's time, and a sweep schedule are all decided in the same way.
A custom (cron) schedule is written as five fields (minute, hour, day, month, day of week) separated by half-width spaces. It is treated as a time in the time zone you chose.
| Field | Value |
|---|
| Month | 1 to 12, or JAN to DEC |
| Day of week | 0 to 7, or SUN to SAT (7 is Sunday) |
Every field accepts * (all), ? (same as *), a number, n,m (a list), n-m (a range), */n (every n), n/k (every k, starting at n), and n-m/k (every k within a range). A value out of range (including the start of a step, such as the 70 in 70/5), a reversed range, a step of 0 or less, or an empty element makes the whole expression invalid. An expression whose next run is more than 366 days away (for example, February 31) can be saved, but no next time is shown and it does not run.