Agent skill · software engineering · jeremylongshore
ideogram-migration-deep-dive
Plan and execute a staged migration across Ideogram legacy, V3, V4, P-Image, and tool endpoints with semantic comparison and rollback. Use when modernizing a production image workflow. Trigger with "migrate Ideogram legacy API", "move Ideogram V3 to V4", or "redesign an Ideogram image pipeline".
Why this skill is useful
Provides detailed migration scripts and API mapping for transitioning to Ideogram from various image generation APIs, which the AI wouldn't reliably generate on its own.
What it needs
Requires npm installed locally. Requires ideogram account access. About 3k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Ideogram Production Migration Overview Modernize a complete Ideogram image workflow without assuming endpoint names imply equivalent semantics. Inventory legacy and V3 behavior, select V4, P-Image, or outcome-focused tools by requirement, compare safe durable results, and preserve in-flight state and rollback. Prerequisites Source routes, payloads, traffic, consumers, stored assets, async state, quality rubric, and owners. Target use cases, model-control needs, latency, safety, rights, cost, and retention constraints. Feature flags, shadow destination, reconciliation plan, canary budget, and rollback window. Current Contract First-party navigation labels /generate, /edit, /remix, /reframe, and /describe as Legacy Endpoints. Current surfaces include V3, V4 sync/async/transparent, P-Image, edit, remix, structured describe, magic prompt, outcome-focused tools, and custom training. Payload and rendering semantics differ. Authentication Preserve the server-side Api-Key boundary and environment isolation throughout migration. Dual-run fixtures and receipts must exclude keys, prompts, uploaded images, generated assets, and temporary URLs. Instructions 1. Inventory each source route, field, enum, output, error, retry, safety, storage, consumer, and in-flight state. 2. Classify every use case by required model control, transparency, edit semantics, structured prompting, specialized outcome, latency, and cost. 3. Select a documented target route for each use case; record non-equivalence and retire unsupported assumptions. 4. Build owned translation at the application boundary instead of leaking mixed legacy and target schemas to consumers. 5. Compare sanitized fixtures, then run approved shadow or dual generation into isolated storage without double publication. 6. Evaluate transport, safety, useful quality, latency, cost, URL persistence, async reconciliation, and deletion. 7. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/ideogram-migration-deep-dive