Repository navigation
fix: one completeness verdict; reply context carries the whole conversation (v1.6.13) - #93
Merged
Merged
Conversation
) The unattended `complete` field came from the strict assessment, which is always clean on the ok path, so it said true for capsules whose Status line said INCOMPLETE. The receipt modal had its own rule that ignored truncated text and missing quoted posts and flagged poster-only videos. Both now use captureVerdict, the same captureGaps verdict the capsule prints. Fixes #91 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A syndication payload names only its own parent, so the reply-context pass stopped after one card and a deep reply lost the post it was about. It now walks up one fetch per ancestor, prepends them oldest first, and labels only the direct parent "In reply to". A cap (8) leaves a note linking onward to X; a 404 or a failed fetch ends the walk with an honest tombstone. None of these is a capture gap. A tombstone from a failed fetch also stopped rendering "already gone on X" in the HTML, the .llm.md and the receipt notes (#25). Fixes #92 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…lace ancestors above the exported post A tombstone from a non-404 failure still fed quoteTombstones, so the receipt and the CLI warning called the post deleted (#25). Inside a thread export the ancestors were unshifted above earlier thread posts they reply to; they now go directly above the exported post's own marker (#92). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This was referenced Oct 18, 2026
Closed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Two more cases of the #85 pattern: a surface that stops after one level, or reports COMPLETE when the capsule doesn't.
#91: three completeness verdicts disagreed. The unattended
completefield came from the strict assessment, which is always clean on theok: truepath, so it saidtruefor capsules whoseStatus:line said INCOMPLETE. The receipt modal had its own rule: it ignored truncated text and missing quoted posts, and it flagged poster-only videos that the capsule accepts. Both now usecaptureVerdict, the samecaptureGapsverdict the capsule prints. The CLI warnings carry the capsule's gap summary.#92: reply context kept only the direct parent. A payload names only its own parent, so exporting a reply deep in a conversation kept one ancestor. Live example: hibtc37/status/2111300625627783421 kept @Xinbtc20's question and dropped the root post it asks about. The pass now walks up one fetch per ancestor, prepends ancestors oldest first, and labels only the direct parent "In reply to"; older ones read "Earlier in the conversation". Past 8 posts a note links onward to X. A 404 or a failed fetch ends the walk with an honest tombstone. The cap note and the tombstones are not capture gaps; truncated text or a missing image inside a context card still is, as in any embedded post. Inside a thread export, ancestors go directly above the exported post, never above an earlier thread post.
#25 follow-through. A reply-context tombstone from a failed (non-404) fetch kept an honest
notice, but the HTML card, the.llm.mdline and the receipt note still said "already gone on X". They now say the fetch failed, and a failed fetch no longer feeds the "gone on X" counter behind the receipt and the CLI's "deleted on X" warning.v1.6.13 bump included.
Verification:
test/capture-verdict.test.mjsdrives a real unattended capture-share into a stub Worker and assertsok: true, complete: falseagainst an uploaded.mdthat says INCOMPLETE, plus the modal-agrees cases. Five DOM tests cover the ancestor walk, the cap, a mid-chain failure, placement inside a thread, and the Reply-context card claims the parent is gone on any fetch failure #25 copy and counters. All failed before the fix.npm test, lint and format pass locally.Fixes #91
Fixes #92
🤖 Generated with Claude Code