Inspect prompt classifications and fix duplicates, thin topics, and tags that no longer distinguish the data.
When to use this workflow
“Our report filters are cluttered; clean up the topics and tags.”
Before fetching
Identify the reports people use and the distinctions they need. Do not delete a topic solely because it is small; check whether it represents a distinct intent.
Complete the CLI quickstart and confirm the workspace with orc workspace current --json. Replace example IDs, URLs, dates, and settings. Apply only approved mutations.
1. Map prompts to their classifications
Read the complete lists and find unassigned prompts, near-duplicate topics, and tags applied to nearly every prompt. Separately record missing brand context.
orc topics list --json
orc tags list --json
orc prompts list --json
orc prompts tags get YOUR_PROMPT_ID --json
orc brands get YOUR_BRAND_ID --json2. Prioritize by impact on report interpretation
Have the agent prepare a before-and-after mapping. Prioritize distinctions that currently distort comparisons, then settle prompt IDs and retained tag sets.
3. Apply and verify each approved change
Replace the approved tag set using the current config_revision and read it back. If a concurrent edit conflicts, refetch and review the difference.
orc prompts tags update YOUR_PROMPT_ID --tag-ids YOUR_TAG_IDS --config-revision YOUR_CONFIG_REVISION --json
orc prompts tags get YOUR_PROMPT_ID --jsonDeliverable
Deliver reasons, before-and-after mappings, verified IDs, and remaining unclassified prompts. Record the change time for future comparisons.
On fetch failure, retain the error and request scope. Do not count empty data, incomplete pagination, or missing permissions as a completed investigation.
Next steps
If a failure remains, report it and verify the fix, including this workflow and the failed step.