docs(spec04): align §3.2/§5.4 with the merged harness-tool impl#29
Merged
Conversation
Slice 2 shipped (PR #26) using a terminal harness tool (emit_talk_beats / parse_talk_beats in talk_tools.py) for structured batch output, but the spec still described the pre-refactor prompt-JSON + _parse_talk_batch approach that no longer exists in code. Rewrite §3.2 and acceptance §5.4 to the tool-seam design (same seam as music discovery, spec 03-01), and fix the stale "slice 2 (next build)" label now that both slices are merged. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Aligns
specs/spec04/04-no-dead-air.mdwith the slice-2 code already merged tomain.What & why
Slice 2 (talk look-ahead) shipped in PR #26 using a terminal harness tool for structured batch output —
emit_talk_beats/parse_talk_beatsinsrc/murmur/talk_tools.py, the same seam as music discovery (spec 03-01). But the spec still described the pre-refactor design: a JSON array requested in the prompt, parsed by a_parse_talk_batchthat no longer exists in code. That's a spec↔code drift (CLAUDE.md alignment red line).parse_talk_beatsis the single trust point; empty/malformed/missing tool result degrades to a singlenext_talkbeat.parse_talk_beatsinstead of the removed_parse_talk_batch.slice 2 (next build)label — both slices are now merged.Docs-only; no code change. The graphify graph was refreshed for the edited spec.
AI coding brief
talk-lookaheadbranch was a redundant duplicate of feat(director): talk look-ahead buffer [spec 04] #26 (also missing chore(voice): report real wav duration, ignoring a streaming placeholder header #27's wav fix) and was deleted.main+ pull, then check whether spec 04 had updates, then execute the fix. Pulling first surfaced that feat(director): talk look-ahead buffer [spec 04] #26 had already landed the code, which reframed the task from "finish slice 2" to "fix the stale spec."mainand check merged PRs first — a local WIP branch can already be redundant. Checking the merge state up front would have skipped the initial "slice 2 still needs a PR" read.🤖 Generated with Claude Code