Skip to content

fix(heartbeat): keep checkpoint capture independent of input display - #6659

Closed
loopx-agent wants to merge 1 commit into
codex/loopx-research-integration-20261016-03from
codex/checkpoint-input-mode-isolation-20261016
Closed

loopx-agent wants to merge 1 commit into
codex/loopx-research-integration-20261016-03from
codex/checkpoint-input-mode-isolation-20261016

Conversation

@loopx-agent

@loopx-agent loopx-agent commented Oct 16, 2026 •

Copy link
Copy Markdown
Collaborator

Checkpoint handoff silently selected TurnEnvelope even when the benchmark display option was off: the worker requested a capture root, and the heartbeat renderer appended --turn-envelope and switched its dispatcher. Capture now preserves the ordinary dispatcher/quota projection. Only an explicit heartbeat-prompt --turn-envelope (with a capture root), or the existing explicit worker option, selects the short view. Generated selection and regeneration commands retain the chosen transport without changing admission or settlement owners.

This changes the previous capture-root-only renderer behavior; the protocol/runtime guides document the migration. Full private captures and checkpoint handoff remain available in both display modes. Receipt projection now follows the explicit display option, and historical receipts are preserved. The existing helper is simplified rather than retaining a checkpoint-to-display compatibility mapping; no new strategy or permission owner is introduced.

Validation: a regression fails on the previous implicit activation; 247 focused tests pass on the candidate, including real worker → renderer → quota/selection/replay → single settlement with display on/off and handoff on/off. Another 247 optional Harbor/native tests are skipped in that local environment, not counted as passes. Semantic vocabulary smoke, public boundary check and changed-runtime Ruff pass; an unchanged test file has five pre-existing Ruff E702 findings. The clean exact head also passes the complete isolated non-root Linux qualification on the current pinned provider: six settled work wakes, pending pre-replan capture, ordinary and stride receivers, automatic full recapture, another receiver’s actual work, and the generic Core Goal journey without benchmark imports. The six underlying tests pass with no skips, using a real native binary, HTTP/SQLite/Docker and deterministic local inference; this proves mechanics, not model research gain. Ordinary-package validation is still being completed. No paid model or benchmark task run is launched.

Base: the latest owner-selected research integration, codex/loopx-research-integration-20261016-03 at 57e909e694b66e34784241839a350bad6d763531. This PR contains only the input/capture separation and its focused validation/docs. Independent exact-head review and merge authorization remain separate from test success.

Signed-off-by: LoopX Agent <337587101+loopx-agent@users.noreply.github.com>
@loopx-agent

Copy link
Copy Markdown
Collaborator Author

Superseded by standalone main-targeted #6661 at 9f1fc4c. The capture/input isolation patch is independent of the planning-horizon proposal; the research integration will compose the exact PR heads. Earlier integration-head validation remains bound to its original revision and is not presented as new-head validation. No branch was deleted or force-pushed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant