mode: auto Is a Claim About Reality, Not an Intention

Description and use

Automations declared as "auto" can diverge from reality, causing silent failures. Runtime verification of actual publication and removal of duplicate source‑of‑truth references uncovered two hidden defects: an automated channel that never published and a hard‑coded platform list that kept writing drafts for a retired platform.

Problem

Scheduled automations declared as "auto" can diverge from reality, leading to silent failures in content distribution.

Outcome

Adding runtime verification of automation claims and eliminating duplicate source‑of‑truth references uncovered two silent failures: an automated channel that never published and a hard‑coded platform list that caused drafts for a retired platform to be written.

Workflow

• 1. Run scheduled automation against declared config.
• 2. Write drafts for each platform per config.
• 3. Publish drafts via adapters.
• 4. Record published items in a ledger.
• 5. Verify automation claim by comparing drafts written vs. items published per channel over a trailing window.
• 6. Ensure instructions reference a single source of truth and do not restate values.

Evidence

The check this needed: don’t trust mode: auto as a fact. Compute it — drafts written vs. items actually published, per channel, over a trailing window.

Source

The instruction said "read the target list from config," and then — a few lines later — named a fixed list of platforms as if for clarity. The fixed list won.

Source

We eventually pulled this whole pattern — the config contract, the adapters, the daily plays that write drafts, and the verifier that checks claims against the publish ledger instead of trusting the config — into a standalone kit (Distribution Autopilot).

Source