Running a real-phone agent that posts to TikTok: the hard part is verifying your action actually happened
Orfeo_Lyrakis · reddit · 2026-11-02
The author built an Android-phone agent that publishes to TikTok via the app for a meat farm's large account (the official API won't allow public posting without an audit). Tapping was easy; the bugs were all about knowing whether the action actually happened:
- Silent hashtag trimming: TikTok keeps 5 hashtags and silently deletes the rest; their checker saw missing tags and refused to post for a day — the "bug" was an undocumented platform rule.
- Warning dialog misfire: after tapping Publish, a "situation evolving in your country" dialog appeared; the failure detector matched "violation" and reported rejection, while the post was just waiting for one more tap.
- Caption quotes broke verification: confirming by matching the caption on the profile failed when quotes broke the parser, so the agent searched for 5 minutes and reported "not confirmed" on a live post.
- Timeout retries nearly double-posted: the caller retried the same video on timeout; the second run luckily died before Publish. Now an idempotency key returns "in progress" and every error states whether Publish was already pressed.
Key lesson: agents need a cheap, reliable way to answer "did my action happen?", and callers must treat timeouts as "unknown", never "failed".
More from coding & agent
- Dev builds Claude-powered pipeline that auto-generates quantum computing explainer videos — AIandDesign · 2026-11-02
- Patched Strata cuts DeepSeek V4.1 agent session startup 20× on a gaming GPU — HearMeOut-13 · 2026-11-02
- Anthropic design lead says agentic AI freed him from his laptop for most daily work — davidhoang · 2026-11-02
- Sketchy: open-source tool lets Claude draw hand-sketched interactive diagrams in chat — Nkgsporty · 2026-11-02
- Ex-Unity senior engineer builds Kaigen 3D engine with stunning browser performance — gdechichi · 2026-11-02
- GSM1k Rebuild Exposes Benchmark Contamination: Up to 8-Point Drops, Agent Eval Advice — krishnan · 2026-11-02