.github/workflows/algolia-reindex.yml (2)

57-62: Consider edge case: multiple crawlers matching the name.

If multiple crawlers share the same name, jq will output multiple lines, causing the variable assignment to capture only the first or behave unexpectedly. While unlikely for this use case, adding | first or limit(1; ...) would make the extraction defensive.

♻️ Optional defensive fix

           crawler_id="$(
             jq -er \
               --arg crawler_name "${ALGOLIA_CRAWLER_NAME}" \
-              '.data[] | select(.name == $crawler_name) | .id' \
+              '[.data[] | select(.name == $crawler_name)] | first | .id' \
               <<<"${response}"
           )"

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In @.github/workflows/algolia-reindex.yml around lines 57 - 62, The jq
extraction for crawler_id can return multiple matches if more than one crawler
shares ALGOLIA_CRAWLER_NAME; update the jq filter used in the pipeline where
crawler_id is assigned to defensively take only the first match (e.g. use
limit(1; .data[] | select(.name == $crawler_name) | .id) or pipe the stream to
first) so the assignment to crawler_id always captures a single id even when
multiple crawlers match.

---

100-137: Consider using action_id for precise verification.

The step captures action_id in the previous step but doesn't use it here. While checking .reindexing == true works, using the action-specific endpoint could provide more precise status tracking. However, the current approach is functional and avoids additional API complexity.

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In @.github/workflows/algolia-reindex.yml around lines 100 - 137, The step polls
the crawler by matching .name and checking .reindexing, but you captured an
action_id earlier and should use it for precise verification; update the curl
request or jq filter in this step to query the action-specific endpoint or
include the action_id parameter (use the captured variable name ACTION_ID or
whatever the previous step outputs) and change the jq selector from '.data[] |
select(.name == $crawler_name) | .reindexing' to select the entry by action_id
(e.g., select(.action_id == $action_id) | .reindexing) and similarly for
.status, so the loop directly inspects the specific reindex action rather than
matching by crawler name. Ensure the env var is passed into the step and keep
the existing retry/sleep logic.