Repository navigation
Conversation
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Follow-up to #34, tightening the boundary between authoritative documents, research material, and now the control/state/observation split.
Root directory emptied of research material
masterstill carried four research/recommendation/report documents at the repository root. They move underdocs/:docs/research/task-feedback-v0-1-2026-10-05/ZCODE_APPSERVER_EVENT_CAPABILITY_PROBE.mdZCODE_APPSERVER_EVENT_CAPABILITY_PROBE.mddocs/research/task-feedback-v0-1-2026-10-05/TASK_FEEDBACK_SCHEMA_RECOMMENDATION.mdTASK_FEEDBACK_SCHEMA_RECOMMENDATION.mddocs/research/task-feedback-v0-1-2026-10-05/NATIVE_FEEDBACK_RENDERER_RECOMMENDATION.mdNATIVE_FEEDBACK_RENDERER_RECOMMENDATION.mddocs/research/task-feedback-v0-1-2026-10-05/evidence/run-001-summary.jsonprobe/evidence/run-001-summary.jsondocs/reports/task-feedback-v0-1-implementation-2026-10-06.mdTASK_FEEDBACK_V01_IMPLEMENTATION_REPORT.mdThe root now holds only
README.md,README.zh-CN.md,AGENTS.md,CHANGELOG.md,LICENSE,NOTICEand the build/config files.The three documents stay together in one dated dossier rather than the three separate folders the review suggested: they are a cross-linked set produced for one deliverable, and single-file folders add path depth without a benefit.
Authority headers on every authoritative document
ARCHITECTURE.md,INTERFACES.md,SHARED_CORE.md,ZCODE_RUNTIME.mdanddocs/README.mdnow carry the header that previously onlyPROJECT_STATE.mdhad.docs/README.mdno longer says status is assigned centrally; each file states its own status, and the index only summarises.Last updatedandLast verifiedare kept separate, and partial verification says so:ARCHITECTURE.md- verified items named explicitly (heartbeat cadence, 15-second grace, result checkpoint, event replay/degradation); the rest is not re-checked.INTERFACES.md- same treatment for the replay/degradation boundary, thesession/readgap and the default tool count.SHARED_CORE.mdandZCODE_RUNTIME.md-未逐条核对.Verified against: ed2d402.AGENTS.mdnow spells out the required header fields. TheOwnerfield the review proposed is deliberately not adopted: there is no ownership structure to record. Note that a document cannot name the commit containing itself, soVerified againstrecords the commit the verification was performed against.ARCHITECTURE.mdgains a system overviewThirty-five lines at the top: a component flow from the Codex host through the MCP server, TaskManager, adapter and ZCode app-server, plus a table of what each component is and is not responsible for. Existing detail is unchanged and now sits under a
运行细节heading.ADR-004 accepted with a three-way split
ADR-004 was proposed because the repository had no explicit record of the control/observation boundary. The review correctly pointed out that the original wording conflated two different things. It is now
Acceptedwith this decision:tasks-index.sqliteThe key correction: local artifacts may feed a supplemental observation plane for diagnostics and visibility - what is forbidden is their becoming the authority for Bridge lifecycle, state transitions or recovery. Recovery keeps relying on Bridge-owned evidence (TaskStore, heartbeat, attempt,
execution.claim, outcome checkpoint, cleanup verification). The Desktop index write is kept but registered as an integration exception with an explicit MUST NOT list: it may not gate execution, change task status, take part in recovery, decide completion, decide cleanup success, or affect the task lifecycle when it fails.The ADR also now clarifies that
src/observation/is not the ZCode-local observation plane: it derives bounded observations from Bridge-owned TaskStore evidence.Renamed to
ADR-004-separate-control-state-and-observation.mdto match the wider scope.PROJECT_STATE.mdandINTERFACES.mdeach gained a pointer to it.No code was added for reading local ZCode artifacts. The principle is frozen now; whether a
ZCodeObserveris worth building is left to the planned Local Observability experiment.Not done here
AGENTS.mdalready carries an explicit split trigger (a document over roughly 15 KB, or one topic read independently by three or more readers).ZCODE_RUNTIME.mdis not renamed or split; renaming now would break links and buy nothing until the extra ZCode subjects exist as documents.Validation
AGENTS.mdonly; no source, schema or bundle change