Custom agent frontmatter: accept array for `model:` field
- Langage dominant
- Shell
- Étoiles
- 11.2k
- Forks
- 1.9k
- Merge moyen
- 14 h 16 min
- PR mergées (30 j)
- 6
Description
### Feature request
Allow custom agent markdown frontmatter to accept an **array** of strings for the `model:` field, in addition to a single string.
### Motivation
VS Code chat modes already support `model:` as a list — it's used to present a model picker for the chat mode. This means the same agent definition file can't currently be shared cleanly between VS Code and Copilot CLI: the CLI rejects the file with:
`custom agent markdown frontmatter is malformed: model: Expected string, received array`
This forces plugin authors to either:
- Maintain two copies of every agent file (one for VS Code, one for CLI), or
- Pick one tool and accept that the agent won't load in the other.
### Proposed behavior
When `model:` is an array, the CLI uses the **first** entry as the default model for that agent (and ideally treats the rest as acceptable alternates the user can switch to with `/model` while invoking the agent). When `model:` is a string, behavior is unchanged.
### Alternatives considered
- Authors flatten `model:` to a single string — works in both, but loses the picker UX in VS Code.
- A local post-install patch script — fragile, breaks after every `plugin update`.
Accepting an array in the CLI parser is the only change that lets the same plugin work unmodified in both surfaces.
Guide de contribution
Ouvrir le guide de contribution
Piste de recherche
Commencez par localiser l’analyseur de frontmatter Markdown de custom-agent et la validation de son champ model. Vérifiez que le comportement des chaînes reste inchangé, que les tableaux utilisent la première entrée comme valeur par défaut et que les valeurs malformées échouent toujours ; ajoutez ou mettez à jour les tests de l’analyseur s’ils existent.
Rédigé par le modèle d'indexation à partir du texte de l'issue.
Évaluation
- Stack technique
- markdown, shell
- Domaine
- cli, tooling
- Type d'issue
- Fonctionnalité
- Difficulté
- 3/5
- Temps estimé
- 1-2 jours
- Activité
- Calme
- Clarté
- Plutôt claire
- Accessibilité débutants
- 68/100