What

Adds four FAQ entries collected from community support (Sep 7–20):

faq/troubleshooting.md

1. Code node intermittently fails with SyntaxError: unterminated string literal / downstream node gets no input. Cause: multi-line string inputs (typically JSON from an LLM node) were corrupted before reaching the Pyodide executor. Points to the fix in fix(workflow): preserve code node source escapes #1671 / fix(workflow): preserve code node source escapes #1672, notes it landed after v1.1.2 (pull the latest core-workflow image), and gives the interim workaround (ask the model for single-line JSON).
2. Trace for the last turn / long reply is missing, log shows KafkaError{code=MSG_SIZE_TOO_LARGE}. Explains that raising the broker's message.max.bytes alone does not help, and points to the fix in fix(workflow): reduce trace payload size for long conversations #1673 / fix(workflow): reduce trace payload size for long conversations #1675.

faq/features.md

3. Where are the workflow service logs, and why is docker logs empty? LOG_STDOUT_ENABLE=0 by default; container path /opt/core/logs/app_*.log, host path config/workflow/logs/app_*.log; a keyword table for trace-reporting failures; timestamps are UTC.
4. Blank suggestion buttons under a reply. These are next-question suggestions and need AI Ability Chat configured under Platform Account Management (with a screenshot).

Why

Both workflow bugs were reported in the community groups and fixed by the maintainers within two days, but users still running v1.1.2 images keep hitting them. Documenting the symptom, cause and upgrade path lets them self-serve, and the log-location entry shortens future troubleshooting.

Verification

Docs-only change. Checked against the current main: LOG_STDOUT_ENABLE=0 in docker/astronAgent/config/workflow/config.env, the ./config/workflow/logs/:/opt/core/logs volume and ghcr.io/iflytek/astron-agent/core-workflow image in docker/astronAgent/docker-compose.yaml, and the fixes merged in #1671–#1675.