garrytan / garrytan/gstack

/autoplan: single-voice mode (Codex unavailable) is mislabeled as CONFIRMED in consensus tables

Open
#1,956 2 comments 0 reactions 0 assignees View on GitHub
Dominant language
TypeScript
Stars
133k
Forks
19.9k
Avg merge
18h 46m
Merged PRs (30d)
26

Description

Filed downstream at aviraldua93/gstack-copilot#8. The bug exists in canonical upstream gstack — surfaced via the auto-generated `/autoplan` skill when `codex` CLI is unavailable.

## Summary

When Codex CLI is unavailable, every `/autoplan` phase degrades to single-voice mode: only the Claude subagent runs. The degradation matrix tags each phase as `[subagent-only]` (good), but the Final Approval Gate output template does NOT surface this degradation prominently. Consensus tables read:

| Dimension | Claude | Codex | Consensus |
|---|---|---|---|
| 1. Architecture sound? | ✗ | N/A | CONFIRMED |

A user reasonably interprets `CONFIRMED` as "both independent reviewers agreed." It actually means "one reviewer flagged this, no second reviewer was available, the orchestrator concurs." This is a silent quality regression.

## Suggested fix

1. Add top-of-report banner when ANY phase ran without dual voices:
```
⚠ SINGLE-VOICE MODE — Codex CLI unavailable for [N] of 4 phases.
Consensus columns below reflect one independent reviewer per phase, not two.
```

2. Change column label from `CONFIRMED` to `SUBAGENT-ONLY` (or `CONFIRMED-1V`) for any row where Codex contributed `N/A`.

Trivially small change in `autoplan/SKILL.md.tmpl`; large effect on trustworthiness.

---
*Filed via GitHub Copilot CLI by @aviraldua93 during stress-testing a downstream port.*

Contributor guide

Open the contributing guide

Research direction

Open autoplan/SKILL.md.tmpl and find the Final Approval Gate output template and its degradation matrix. Update the report to make single-voice phases prominent and distinguish rows where Codex is N/A. Verify the generated /autoplan output when Codex CLI is unavailable.

Written by the indexing model from the issue text.

Assessment

Tech stack
typescript
Domain
cli, tooling
Issue type
Bug
Difficulty
1/5
Estimated time
Under an hour
Activity status
Quiet
Clarity
Clearly specified
Newbie friendliness
55/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.