nickmeron/skills/bulk-delete

The user wants to create or manage bulk deletion jobs in Dynamics 365.

Argument provided: $ARGUMENTS

Process

"Delete all records of entity X" / "Create a bulk delete job"

1. Select environment — call list_environments, ask the user, call select_environment.

2. Confirm scope — ask the user to confirm the target entity and any filter conditions before creating the job. Bulk deletion is irreversible.

3. Resolve the entity logical name — if the user provides a display name (e.g. "Knowledge Article"), use list_entities to find the logical name (e.g. knowledgearticle).

4. Build filter conditions (optional) — if the user wants to delete a subset of records, map their criteria to QueryExpression conditions:

   • Each condition has: attribute (logical name), operator, and optionally values
   • Common operators: Equal, NotEqual, Null, NotNull, Like, In, GreaterThan, LessThan, OlderThanXDays
   • Example — delete only inactive records: { attribute: "statecode", operator: "Equal", values: ["1"] }
   • If no filter is provided, ALL records of the entity will be deleted.

5. Create the job — call create_bulk_delete_job:

   job_name: descriptive name (e.g. "Delete all Knowledge Articles – 2026-05-04")
   entity: logical name (e.g. "knowledgearticle")
   filter_conditions: [...] or omit for all records
   

   Returns a JobId (asyncoperationid) — save this for monitoring.

6. Confirm to the user — report the JobId and explain they can monitor it with get_bulk_delete_job.

---

"Check status of bulk delete job" / "Is the bulk delete done?"

Call get_bulk_delete_job with the job_id.

Status interpretation:

| statecode | statuscode | Meaning | |-----------|------------|---------| | 0 | 0 | Waiting for resources | | 1 | 10 | Waiting | | 2 | 20 | In progress | | 2 | 21 | Pausing | | 2 | 22 | Canceling | | 3 | 30 | Succeeded | | 3 | 32 | Cancelled | | 3 | 33 | Failed |

• If in progress — report current state; tell user to check again shortly.
• If succeeded — report completion time; offer to check for any failures with get_bulk_delete_failures.
• If failed — call get_bulk_delete_failures to show which records could not be deleted and why.

---

"List all bulk delete jobs" / "Show recent bulk deletes"

Call list_bulk_delete_jobs. Optional status filter: waiting, running, completed, failed, cancelled.

Present results as a table: Job Name | Status | Created | Completed | JobId.

---

"Cancel the bulk delete job"

1. Call get_bulk_delete_job to confirm the job is still active (statecode 0, 1, or 2).
2. Warn the user that cancellation stops the job — already-deleted records are NOT restored.
3. Call cancel_bulk_delete_job with the job_id.

---

"Why did some records fail to delete?"

Call get_bulk_delete_failures with the job_id. The response includes:

• objectid — the GUID of the record that failed
• errordescription — the reason for failure (e.g. cascading relationships, active child records)

Common failure reasons:

• Relationship constraints — child records still exist and the relationship is set to Restrict
• Locked records — record in use by another process
• Security — the service principal lacks delete privilege on the record

---

Tips

• Irreversible — bulk delete permanently removes records. Always confirm with the user before creating a job.
• Recurrence — set recurrence_pattern to run the job on a schedule (e.g. daily cleanup). Leave empty for a one-time run.
• Large datasets — jobs run asynchronously; large tables may take minutes to hours. Poll with get_bulk_delete_job.
• Failures are normal — a job can succeed overall even if some records failed. Check get_bulk_delete_failures after completion.