Simplify private spam deletion confirmations #35

Merged

backmeupplz merged 1 commit from fix/71-concise-confirmations into main

2026-10-10 23:17:31 +00:00

backmeupplz commented

2026-10-10 23:11:16 +00:00

(Migrated from github.com)

Summary

Implements agentboard #71: one concise private confirmation containing the linked actual sender name, original deleted text/caption, and direct reply text/caption only when present.

Escaped inert message bodies, safe existing sender links, factual textless-media markers and lossless bounded continuations.
No classifier/policy/recipient/logging/schema/retention changes; successful-deletion gating and linked-message snapshots remain unchanged.
Existing queued HTML parts finish byte-for-byte unchanged, including the old verbose format. No purge, migration or re-render.

Verification

Typecheck passed.
Full Bun suite with dedicated local PostgreSQL audit DB: 494 passed, 257 opt-in tests skipped, zero failures.
Exact output, malicious markup/URLs, long Unicode, direct-source attribution, cached/edited/historical deletion, failure statistics, restart/dedupe and old queued-format fixtures.
Independent Astra/high exact-head review PASS, no actionable findings (run 30ae5242-2575-4bef-9c3b-fd78eb024389), head 85a247a7a9. Reviewer independently passed typecheck and 474 tests (277 skipped without DB).
GitHub CI passed: https://github.com/backmeupplz/jev_antispam_bot/actions/runs/38094143949
No live spam repost or destructive Telegram test. Optional live UI evidence not performed.

Release

Based on merged #34. This PR only changes confirmation rendering; coordinate managed bot rollout with the existing #69 release owner. Implementation phase does not merge or deploy.

# Summary Implements [agentboard #71](https://agentboard.icefish-betta.ts.net/#71): one concise private confirmation containing the linked actual sender name, original deleted text/caption, and direct reply text/caption only when present. - Escaped inert message bodies, safe existing sender links, factual textless-media markers and lossless bounded continuations. - No classifier/policy/recipient/logging/schema/retention changes; successful-deletion gating and linked-message snapshots remain unchanged. - Existing queued HTML parts finish byte-for-byte unchanged, including the old verbose format. No purge, migration or re-render. # Verification - Typecheck passed. - Full Bun suite with dedicated local PostgreSQL audit DB: 494 passed, 257 opt-in tests skipped, zero failures. - Exact output, malicious markup/URLs, long Unicode, direct-source attribution, cached/edited/historical deletion, failure statistics, restart/dedupe and old queued-format fixtures. - Independent Astra/high exact-head review PASS, no actionable findings (run 30ae5242-2575-4bef-9c3b-fd78eb024389), head 85a247a7a90d7f846f61df7d8886b8c6e14db132. Reviewer independently passed typecheck and 474 tests (277 skipped without DB). - GitHub CI passed: https://github.com/backmeupplz/jev_antispam_bot/actions/runs/38094143949 - No live spam repost or destructive Telegram test. Optional live UI evidence not performed. # Release Based on merged #34. This PR only changes confirmation rendering; coordinate managed bot rollout with the existing #69 release owner. Implementation phase does not merge or deploy.