History and exports

What are history and exports in Serpel?

Almost everything Serpel does for you runs as a background job: a crawl, a rank check, a monitoring run, a sync. The History page is the log of those jobs, plus the report emails and keyword researches of the project. It answers three questions: did it run, did it work, and what do I do if it didn’t? Exports and tasks sit on the same page of the docs because they work with the same data: you export what the jobs produced and track what you fix.

What does the History page show?

History sits under Activity in the project sidebar. Two cards on top show Jobs in 7 days and Failed, with the types that failed and when. To the right, By type counts the entries per type. The main list groups entries by day, newest first, with 50 entries per page and Newer and Older buttons. Each entry shows its type, a status badge, the time, a one-line result and the duration. While a job is queued or running, the page refreshes every few seconds.

Click an entry to open its details: status, when it was queued, scheduled, started and finished, the duration, the attempts as “1 of 3” and, after failures, the error of every attempt. A finished monitoring run has a View report link. Viewers can read everything but can’t cancel or retry.

Which entries appear and what do the statuses mean?

The API and CLI use the values queued, running, succeeded, failed, cancelled for the same five statuses. The summary numbers on the page cover the last 7 days. In the API and CLI you can widen them up to 90 days with days or --days.

How do you cancel or retry a job?

1. 1

   Open the entry.

   Click it in the list. Only owners and editors see the buttons.

2. 2

   Press Cancel job or Retry.

   Cancel job appears for queued and running jobs. A queued job stops at once, a running job stops at its next checkpoint and shows “Cancelling …” until then. Retry appears for failed and cancelled jobs and queues the same job again.

3. 3

   Watch the entry.

   A retried job starts with a fresh set of attempts. The toast says “Cancellation requested” or that the job is queued again.

Serpel also retries on its own. Each job gets up to three attempts in total. After a temporary error, such as a provider timeout, the worker waits about half a minute before the second attempt and about two minutes before the third. Errors that trying again can’t fix, such as a deleted project, fail at once. A job whose worker disappears is released and runs again within the same limit. If the last attempt fails, the job is Failed and waits for you.

How long does Serpel keep history?

Finished jobs are removed after 180 days. Serpel keeps the newest finished job of each type and outcome per project, so the last result of every kind stays visible. Running and queued jobs are never removed. The saved results of a keyword research are emptied after 180 days as well.

What can you export?

Which tables have Export as CSV?

Open Options at the right end of the view bar and choose Export as CSV on Tracked keywords, Opportunities, the routes of the Codebase page, the audit Pages table and the list of URLs of one issue. The file is named after the table and the date, for example tracked-keywords-2026-11-05.csv. Two more lists have their own export: Export on search data and web analytics offers several files, and the command menu (Ctrl K) offers Download issues as CSV and Download issues as JSON on the Issues page, and Download keyword ideas as CSV and Download keyword ideas as JSON after a keyword research (see keyword research).

What does the project export contain?

The type is one of issues, keywords, rankings, tasks. The file is named serpel_<domain>_<type>_<date>.csv or .json. The CSV is comma-separated and starts with a byte order mark so Excel reads UTF-8 correctly. The JSON repeats the project, the type and the export time before the rows. Keyword lists also export through GET /keyword-lists/{id}/export, and the account data through GET /account/export.

How do tasks work?

A task is an optimisation to-do for a project. There is no Tasks page in the dashboard, so you manage tasks with the CLI or the REST API. A task is either free (just a title) or created from an audit issue. An issue task carries the affected URLs of the latest completed crawl, uses the issue’s explanation and fix as its description and takes its priority from the severity: errors are High, warnings Medium and notices Low. There is only one open task per issue, so creating it twice returns the existing task.

The status is Open, In progress or Done. After every completed crawl, Serpel re-checks each task that was created from an issue, if it isn’t done or was completed in the last 90 days, against the new crawl. The result is Resolved when none of the linked URLs still has the issue, Partially resolved when some do and Still present when all do. The re-check never changes the status of a task: you decide when it is done. serpel tasks show lists the URLs that are still affected.

How do you use history, exports and tasks from the CLI, API or MCP?

Reading jobs needs the scope of the job’s area, for example audit:read for crawls, projects:read for syncs and rankings:read for rank checks, AI checks and monitoring. Retrying needs the matching write scope. The activity list accepts any of the read scopes. Tasks need tasks:read and tasks:write.

serpel jobs activity --project "$PROJECT_ID" --days 30
serpel jobs list --project "$PROJECT_ID" --status failed
serpel jobs status "$JOB_ID" --wait
serpel jobs retry "$JOB_ID"
serpel jobs cancel "$JOB_ID"
serpel export --project "$PROJECT_ID" --type issues --format csv --output issues.csv
serpel export --project "$PROJECT_ID" --type rankings --format json --output rankings.json
serpel tasks create --project "$PROJECT_ID" --issue title_missing
serpel tasks update "$TASK_ID" --status done
serpel tasks show "$TASK_ID"

curl "https://serpel.app/api/v1/projects/$PROJECT_ID/activity?days=30&limit=50" \
  -H "Authorization: Bearer $SERPEL_TOKEN"

curl -X POST https://serpel.app/api/v1/jobs/$JOB_ID/retry \
  -H "Authorization: Bearer $SERPEL_TOKEN"

curl -o issues.csv "https://serpel.app/api/v1/projects/$PROJECT_ID/export?type=issues&format=csv" \
  -H "Authorization: Bearer $SERPEL_TOKEN"

curl -X POST https://serpel.app/api/v1/projects/$PROJECT_ID/tasks \
  -H "Authorization: Bearer $SERPEL_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"issueCode": "title_missing", "priority": "high"}'

With the CLI, serpel jobs status and the --wait flag of long commands poll a job for you and exit with a non-zero code when it fails. For AI assistants, the MCP tool get_job_status reports the status, progress, attempts and a summary of any job by ID and suggests polling every 10 to 30 seconds. There is no MCP tool for tasks or exports. get_project_overview includes the task counts. Setup is on the MCP page for Claude Code.

Where to go next