docs(design): Ward Gate 3 — add Non-goals section + reframe open question #4#421
Merged
Conversation
#4 Adds a stated Non-goals section immediately after the Summary of the Gate 3 coherence-review design, and rewrites open question #4 to match. Non-goals codify four properties as Ward-contract-level, not Gate 3 implementation choices: - Model-scored probes as gate inputs (advisory only, always) - Auto-approval on any probe score at any threshold (never) - Probe evidence as a substitute for the principal's read - Ward-side coherence policy (what 'coherent' means stays with the familiar, not the Ward) Open question #4 is reframed from 'does auto-approval ever happen' (now settled: no) to 'how are model-scored probes presented as advisory evidence when they arrive' — a presentation decision, not a policy one. Rationale: preserves Gate 3's design property that the principal is the sole approver across future probe-mechanism changes; any relaxation becomes an RFC-0001 amendment, visible at spec-review time.
Contributor
There was a problem hiding this comment.
Pull request overview
This PR refines the Gate 3 coherence-review design doc by making key “principal remains the sole approver” constraints explicit at the contract level, and by reframing an open question away from auto-approval policy toward future probe presentation/UX.
Changes:
- Adds a Non-goals section to codify Ward-contract-level invariants (e.g., probes remain advisory; no auto-approval).
- Rewrites Open question #4 to focus on how future model-scored probes should be presented (separate block vs per-entry marker).
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+31
to
+34
| - **Model-scored probes as gate inputs.** Probes at Gate 3 may become | ||
| model-scored in a later version, but their role remains advisory: they | ||
| inform the principal's decision, they never gate the transition from | ||
| *held* to *applied* on their own. |
Comment on lines
+40
to
+43
| - **Probe evidence as a substitute for the principal's read.** Probes surface | ||
| what a deterministic check can see (parse, size delta, protected-region | ||
| touch, pattern lint). They do not summarize intent, drift risk, or | ||
| identity coherence — those remain the principal's judgment. |
This was referenced Jul 18, 2026
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.
Follow-up to #420 (Ward Gate 3 coherence-review design).
What
Two edits to
docs/design/ward-gate3-coherence.md:New Non-goals section immediately after the Summary. Codifies four properties as Ward-contract-level, not Gate 3 implementation choices:
Open question chore(deps): bump portable-pty from 0.8.1 to 0.9.0 #4 rewritten — the old framing ("does auto-approval ever happen?") is now settled by Non-goals. The remaining live question is the presentation of model-scored probes when they arrive (separate
advisory_probesblock vs. per-entrykind: "advisory"marker) — a UX decision, not a policy one.Why
The Gate 3 doc's strongest single design property is that the principal remains the sole approver, with probes advisory forever. That property was implicit — spread across §G3.2, §Compatibility, and open question #4. Making it a stated non-goal moves the constraint from implementation intent to contract: any future PR proposing auto-approval on a probe score is visibly out of scope at design-doc level, and relaxing it becomes an RFC-0001 amendment against
familiar-contractrather than a Gate 3 tuning decision.Same shape as WARD-C1–C6 in the two-compaction contract: name the invariants so drift is visible at review time.
Non-changes
familiar-contractcan land once Gate 3 v1 ships, so the RFC amendment has a reference implementation behind it.Verification
git diffclean, 33 insertions / 4 deletions, one file.🌿