An Agentic App plugin for Packfiles Warp migration workflows. It helps teams plan migrations, diagnose failed/stuck migrations, and validate Migration HQ configuration for Azure DevOps and Bitbucket Server to GitHub migrations.
.
├── plugin.json
├── agents/
│ └── main.agent.md
└── skills/
├── diagnose-migration/
│ ├── SKILL.md
│ └── failure-catalog.md
├── plan-migration/
│ ├── SKILL.md
│ ├── output-template.md
│ └── scoring.md
└── validate-config/
├── SKILL.md
└── warp-yml-schema.md
Agents live in agents/ and skills in skills/<name>/ at the repository
root, as required by the Copilot CLI plugin spec — the harness discovers the
main agent (agents/main.agent.md) from these paths declared in
plugin.json.
agents/main.agent.md defines the behavior and orchestration for:
- Diagnosing migration failures on backlog issues
- Planning wave-based migrations
- Validating
config/warp.ymlupdates
Folder: skills/diagnose-migration
- Main instructions:
SKILL.md - Classification reference:
failure-catalog.md - Purpose: classify failed/stuck migration issues from issue body, labels, and comments
- Output behavior: post a diagnosis comment to the issue with root cause, remediation, and retry command (typically
/migrate)
Folder: skills/plan-migration
- Main instructions:
SKILL.md - Scoring rules:
scoring.md - Document format:
output-template.md - Purpose: score repositories, assign migration waves, apply
planning:*labels, and generate a customer-ready migration strategy document - Output behavior: generates
migration-strategy.md(default location: repository root unless the user specifies otherwise)
Folder: skills/validate-config
- Main instructions:
SKILL.md - Schema reference:
warp-yml-schema.md - Purpose: validate
config/warp.ymlfor schema correctness, policy conflicts, and security implications - Output behavior:
- Review mode: post a structured findings comment
- Authoring mode: return corrected YAML snippets in chat
This repository is public because the contents are synced and run as a deployed
Agentic App. Every file here defines production agent behavior, and the agent
processes untrusted issue/PR/comment content. The plugin stores no secrets —
it authenticates to the Warp API via OIDC. See SECURITY.md for
the security model and how to report a vulnerability, and
.github/CODEOWNERS for required review on agent changes.