feat(sshx): raise default bounded-pass cap from one to five - #1036
Merged
Conversation
The one-pass default in the sshx "Fix Or Done" contract was too tight for iterating a fix/review loop to convergence on high-stakes decisions; it forced re-asking the user every round. Raise the default to at most five passes — still bounded, still overridable by explicit user authorization, and "do not loop indefinitely" still holds. Update the contract test assertion to match, and bump the plugin version 1.0.0-beta.26 -> 1.0.0-beta.27 across all manifests. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01UJTXbfYjxCQs8AdRdPD3bJ
ElonSG
pushed a commit
that referenced
this pull request
Jul 10, 2026
Supersedes the concurrent beta.27 claimed by #1036 (bounded-pass cap 1->5); this sshx beauty/worth/nyxid upgrade takes the next prerelease coordinate. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012SDE383VM4UNp2KazNfZB4
ElonSG
added a commit
that referenced
this pull request
Jul 10, 2026
* feat(sshx): add 值不值 worth seat, strengthen 美不美, wire nyxid-oracle carrier Upgrade the sshx thinking stage with two philosophical judgments and a model-diverse reasoning channel. - worth (值不值): add as the sixth thinking-panel seat — an independent decision-value objective (counterfactual, opportunity cost, lifecycle burden) distinct from parsimony (how much mechanism) and proportional-containment (where/how far it binds). Kept as a seat, not a cross-cutting lens, so the panel is not homogenized into every seat re-deriving the same value verdict. The meta-judge blocks `implement` on a not-worth-it verdict; beauty<->worth is a conditional elegance-premium / accepted-debt challenge, not a forced dyad. - 美不美: strengthen the shared Aesthetic/adversarial lens to a symmetric verdict (permit `no material defect found`) instead of a presumed indictment, and define beauty as smaller — gold-plating past the goal is itself ugly. Kept as the single aesthetic source of truth; the shared needle with consensus-loop is preserved and no second beauty seat is added. - nyxid-oracle: add as a first-class model-diverse worker carrier (ChatGPT Pro via `nyxid oracle`) with its completion contract (structured status=completed + envelope in the response, single collection read, no busy-poll), per-seat isolation verification (a new conversation alone is not a sterile context), model-family truth-marking, and diversity-aware dispatch so it is not a decorative fallback. - Bump 1.0.0-beta.26 -> 1.0.0-beta.27 across all mapped manifests. The design was independently reviewed via nyxid GPT PRO (ChatGPT Pro) as an adversarial bypass channel; its critiques promoted worth from a lens to a seat, made the aesthetic verdict symmetric, replaced a forced beauty/worth dyad with a conditional challenge, and required diversity-aware dispatch. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012SDE383VM4UNp2KazNfZB4 * chore(release): bump to 1.0.0-beta.28 Supersedes the concurrent beta.27 claimed by #1036 (bounded-pass cap 1->5); this sshx beauty/worth/nyxid upgrade takes the next prerelease coordinate. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012SDE383VM4UNp2KazNfZB4 --------- Co-authored-by: Loning <aloning@gmail.com> Co-authored-by: Claude Opus 4.8 (1M context) <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
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.
Motivation
The sshx
## Fix Or Donecontract capped every bounded pass (meta-layer convergence, arepeated review pass, and fix passes) at one by default. In practice, iterating a
fix→review loop to convergence on a high-stakes decision needs more than one pass; the
one-pass default forced re-asking the user every round. Raise the default to at most five
passes — still bounded, still overridable by explicit user authorization, and "do not loop
indefinitely" still holds.
Change
skills/sshx/SKILL.md: default bound "at most one pass" → "at most five passes".skills/sshx/tests/test_sshx_contract.py: update the asserted phrase to match.1.0.0-beta.26→1.0.0-beta.27across all plugin manifests:package.json,.claude-plugin/plugin.json,.cursor-plugin/plugin.json,.codex-plugin/plugin.json,.claude-plugin/marketplace.json,gemini-extension.json,skills/consensus-loop/VERSION.json.Test
python3 -m pytest skills/sshx/tests/test_sshx_contract.py -q→ 35 passed.🤖 Generated with Claude Code
https://claude.ai/code/session_01UJTXbfYjxCQs8AdRdPD3bJ