docs(vrs): add the 05-harness-state subsystem intent layer - #320
Conversation
schickling
changed the title
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: bf4735675f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
bf47356 to
256de7b
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 256de7b485
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
256de7b to
b0d7495
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: b0d7495a9a
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
b0d7495 to
64f9f4a
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 64f9f4ae84
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
64f9f4a to
a9c1518
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: a9c1518855
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
a9c1518 to
5d936fd
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5d936fd51b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
5d936fd to
0e18ec0
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 0e18ec0c42
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
0e18ec0 to
184656e
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 184656e053
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
184656e to
31df27d
Compare
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 31df27d685
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| reported activity), `Review`, `WaitingOnApproval`, and `WaitingOnUserInput` | ||
| report `active` with `blockedOn: human`, `Compaction` reports `active`, and |
There was a problem hiding this comment.
Fresh evidence in the reviewed tree is that normative OHS-R05 still groups Review with the two actual human waits and requires blockedOn: human, while the corrected projection in spec.md maps Held { Review } to blockedOn: none because review work continues inside the running model turn. A producer implemented from the requirements would therefore still show “blocked on you” during an autonomous review; remove Review from this grouping so the requirement and projection agree.
Useful? React with 👍 / 👎.
There was a problem hiding this comment.
Fixed in 789f91f on schickling/hs-02-vrs-intent: OHS-R05 no longer groups Review under blockedOn: human; the requirement now matches the corrected projection (Held { Review } -> active / blockedOn: none, nothing awaiting a human).
Posted on behalf of @schickling
| field | value |
|---|---|
agent_identity |
unknown |
agent_persona |
generalist |
agent_supervisor |
unavailable |
agent_tool |
OMP |
agent_tool_version |
18.0.3 |
agent_runtime |
OMP 18.0.3 |
tooling_profile |
dotfiles@f33cd9c-dirty |
There was a problem hiding this comment.
Fixed in 789f91f on schickling/hs-02-vrs-intent: OHS-R05 no longer groups Review under blockedOn: human; the requirement now matches the corrected projection (Held { Review } -> active / blockedOn: none, nothing awaiting a human).
Posted on behalf of @schickling
| field | value |
|---|---|
agent_identity |
unknown |
agent_persona |
generalist |
agent_supervisor |
unavailable |
agent_tool |
OMP |
agent_tool_version |
18.0.3 |
agent_runtime |
OMP 18.0.3 |
tooling_profile |
dotfiles@f33cd9c-dirty |
Decision 0006 (observed harness state is a driver-written catalog record), the 05-harness-state requirements/spec/open-questions set, ontology terms with collision rules pinning the new axis against presence, session state, declared activity status, and working state, root DQ2/DQ3 updates, and DELTA-005 fencing the spec landing ahead of its producers. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…d the real writer set The v1 contract states ask and incarnation beside the original tuple, the codex/claude producer sections name their ask kinds, and OHS-R01/OHS-A03 name the actual writers — the owning session's driver processes sharing one incarnation token — instead of the wrapper alone. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
incarnation and seq appear in the wire example with the directional ownership rule and the legacy missing-field derivation stated beside them. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…ss-check OHS-T04/OHS-R07 no longer overclaim: the probe narrows the ungraceful window only for provably dead sessions (pidfile present, process gone), pty kill's pidfile removal is named, and the relaunch-time written claim is what supersedes what the probe cannot prove. The ownership bullet carries the written-claim semantics. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
… on a human Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
…he ungraceful window is honest same-host too Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
… residual are on the record The spec now states the shipped algorithm — a claim mints one above max(on-disk sequence, .harness-state.seq floor) — and documents the sidecar file and the accepted both-files-damaged residual instead of asserting a plain on-disk-plus-one. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The normative requirement still grouped Review under blockedOn:human while
the corrected projection maps Held{Review} to active with no block — a
producer built from the requirement would report 'blocked on you' during
autonomous review. The grouping now matches the projection.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
agent-identity: unknown
agent-persona: generalist
agent-supervisor: unavailable
agent-tool: OMP
agent-tool-version: 18.0.3
agent-runtime: OMP 18.0.3
tooling-profile: dotfiles@f33cd9c-dirty
31df27d to
789f91f
Compare
2 participants
The VRS intent layer for observed harness state, following the 04-stream precedent.
0006-observed-harness-state-is-a-driver-written-catalog-record— the interview decisions of 2026-08-23: all-four-harness coverage (incl. OpenCode full parity), full tuple withblockedOnas a field, catalog record besidestatus, transition + slow-heartbeat freshness, roster join over a sibling command, history deferred (burst-coalescing argument), both teardown fixes, proceed without waiting on Specify native DING and Codex app-server delivery #242.docs/vrs/05-harness-state/: requirements (OHS-A/T/R series refining the R08/DQ3 observed slot, root-wins clause), Draft spec (schema, derivation table, Codex projection table, wire shape, verification plan), open questions DQ-H1..H6.Part of the #268 stack (#319→#326).
🤖 Generated with Claude Code