Repost-with-agent

Run Repost-with-agent for one pair, all enabled pairs, or an explicit subset — scrape source, dedupe (local + destination), expand URLs, publish when allowed, append history, and notify the user. Use when the user asks to "run pair <id>", "post the latest from <pair>", "tick the <pair-id> pair", "run all pairs", or invokes /repost-run. Also invoked by scheduled agents for listen-for-future pairs.

Process one selected source item across every enabled destination pair for that source, especially scheduled/backfill slots. Use when the user or scheduler asks to backfill/catch up a source platform such as LinkedIn to all configured destinations, or when a partial fanout needs resume.

User-configured allow/skip rules for Repost-with-agent candidates. Apply before dedupe/publish so Ethan can block categories such as X video/livestream promos without polluting publish ledgers.

Walk back through historical source posts and repost the missing ones (newest-first). Use when the user asks to "backfill <pair>", "catch up the destination", "repost the last N posts", or invokes /repost-backfill. Multi-post operation with rate-limiting between publishes.

Install current-harness scheduler entries for Repost-with-agent. Default listen mode is one fresh agent/subagent running /repost-run all across enabled listen-for-future pairs; source backfills use source-item fanout; custom per-pair, subset, dry/preview, and arbitrary cadence jobs are supported when the user asks.

Create or save a new Repost-with-agent source → destination pair. Use when the user wants to set up cross-posting between websites the agent can safely operate through a logged-in browser (with documented examples for LinkedIn, X, Bluesky, Threads, and Facebook), or asks to "configure a repost pair", "set up reposting", "wire up cross-posting", or "add a new pair to repost-with-agent / repost_with_agent".

Cross-pair/global dedupe for Repost-with-agent. Use before any publish-capable run so pairs with different sources/destinations share one content ledger and avoid double-posting the same item through alternate hops such as LinkedIn→X→Bluesky and X→Bluesky.

Reference for the Repost-with-agent dedupe algorithm — how to check a candidate against per-pair history, the global cross-pair ledger, and recent destination posts to avoid double-posting. Used by repost-run, repost-backfill, and any other publish path.

Expand shortened URLs (lnkd.in, t.co, bit.ly, buff.ly, etc.) in a draft post body to their canonical form before publishing. Use as part of the repost-run / repost-backfill flow, or when the user asks "expand this URL", "follow this redirect", or invokes /urls expand.

Send the configured user-facing confirmation after a successful repost publish or aggregate source-item fanout, OR test that notification delivery is wired up. Use as part of the repost-run / repost-backfill / repost-source-fanout flow, or standalone when the user asks "test the repost telegram", "send a test ping", or "verify notify is configured".

List all configured Repost-with-agent pairs with their mode, runMode, schedule, and last-run summary. Use when the user asks "what pairs are configured?", "list my repost pairs", or "show me the repost-with-agent state".

Show full details for one Repost-with-agent pair, including config, recent posts, recent audit events, and learnings. Use when the user asks "show me <pair-id>", "what's the state of <pair-id>", or "audit <pair-id>".

Layer 2 semantic-similarity dedupe for Repost-with-agent. After Layer 1 (exact + fuzzy-string match) returns "no duplicate", you (the agent) read the candidate draft alongside the destination's most recent posts and use your OWN reasoning to decide whether the candidate is "saying the same thing in different words" as anything already on the destination. If yes, skip the publish — Ethan would rather miss a post than ship an embarrassing paraphrased duplicate.

Read + write the per-pair learnings.md file so the running agent builds up institutional knowledge over time and doesn't re-figure the same platform quirks every run. Use as a reference when running, backfilling, or showing a pair — and whenever you discover a quirk worth saving.

Read the per-pair posted.jsonl history file and show recent entries. Use when the user asks "what did I repost from <pair>", "show <pair> history", or "tail repost history".