cross-ai: fix stale agy/cursor model ids and cursor trust gate#15
Merged
Conversation
The embedded cross-ai skill shipped model ids that no longer resolve, causing failures when the skill was used: - agy flagship gemini-3.1-pro -> gemini-3.1-pro-high (no bare form exists; only Low/High tiers). Updated in the dispatch table and both pty truncation-guard commands. - cursor flagship gpt-5.5 -> gpt-5.5-high (no bare form exists). - cursor also blocked on a Workspace Trust prompt in any untrusted directory; added --trust (grants directory trust only, not --force/--yolo, so the read-only posture holds) and documented it. - Refreshed both override-table 'other selectable' lists with ids verified against agy/cursor-agent directly; dropped ones that no longer resolve (agy claude-sonnet, cursor gemini/grok-4.3). - Documented agy's display-name-vs-id quirk in the stale-id rule. codex row left unchanged per current preference. All updated ids were confirmed with live one-line test runs against each CLI.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Why
The embedded
cross-aiskill carried model ids that no longer resolve, so the skill failed at runtime — the errors were a mix of rejected model ids and cursor's workspace-trust gate (not an auth problem). Verified every replacement id with a live one-line run against each CLI.What changed (
internal/embedded/default-skills/cross-ai/SKILL.md)agy — flagship
gemini-3.1-pro→gemini-3.1-pro-highgemini-3.1-pro;agy modelsonly exposes(Low)/(High)tiers. Fixed in the dispatch table and both pty truncation-guard commands.gemini-3.5-flash-low,claude-opus-4-6-thinking,gpt-oss-120b-medium(droppedgemini-3.5-flash,claude-opus,claude-sonnet— all rejected by the CLI).cursor — flagship
gpt-5.5→gpt-5.5-highgpt-5.5in cursor either. Also added--trust: cursor-agent blocks on a "Workspace Trust Required" prompt in any untrusted directory.--trustgrants directory trust only — it is not--force/--yoloand does not auto-approve edits/commands, so the read-only posture holds. Documented alongside the agy truncation guard.cursor-agent --list-models:gpt-5.5-high-fast,claude-opus-4-8-thinking-high,composer-2.5,cursor-grok-4.5-high,gpt-5.3-codex(droppedgemini-3.1-proandgrok-4.3— cursor has no gemini and grok is nowcursor-grok-4.5-high).codex — left unchanged (
gpt-5.5) per current preference.Maintenance note added — documented that
agy modelsprints display names, not ids, and the display→id mapping is inconsistent (not every shown tier resolves; dotted vs dashed version separators vary), so a specific agy id should be confirmed with a one-line run before trusting it.Verification
All updated ids confirmed live:
agy --model gemini-3.1-pro-high→OK(and each listed agy alt individually).cursor-agent --model gpt-5.5-high --trust --output-format json | jq -r '.result'→OK(full dispatch pipeline).go build ./...clean (embed intact).🤖 Generated with Claude Code