Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add 0xhughs/director-skills --skill model-adaptationgit clone --depth 1 https://github.com/0xhughs/director-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/0xhughs/director-skills/model-adaptation)<a href="https://agentmods.dev/skills/0xhughs/director-skills/model-adaptation"><img src="https://agentmods.dev/badge/skills/0xhughs/director-skills/model-adaptation/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/0xhughs/director-skills/model-adaptation"><img src="https://agentmods.dev/badge/skills/0xhughs/director-skills/model-adaptation.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00088 | $0.00854 |
| Opus 5 | $0.00044 | $0.00427 |
| Sonnet 5 | $0.00018 | $0.00171 |
| Haiku 4.5 | $0.00009 | $0.00085 |
Grade A, and why
model-adaptation scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.
model-adaptation
When to use
- The user names a specific image or video model/tool.
- The user asks to convert a prompt from one model to another.
- The user asks why a prompt works in one model but fails in another.
- The user wants best settings, a model-agnostic package, or exports for multiple tools.
- The prompt uses model-dependent features: negative prompts, reference images, seeds, duration, aspect ratio, resolution, camera control, audio, or text rendering.
When not to use
- The user only wants story ideation with no target tool.
- The user asks for unsafe policy bypasses.
- The user wants live, up-to-date official specs and source verification; in that case browse or check current docs before asserting changed capabilities.
Required inputs
- Universal prompt brief or existing prompt
- Target model(s)
- Modality: image, T2V, I2V, editing, extension, upscaling, etc.
- Desired aspect ratio/duration/resolution if relevant
Optional inputs
- Reference assets and their roles
- Negative constraints
- Seed/settings
- Known failure from previous output
- Current docs supplied by user
Workflow
- Parse the universal brief into creative intent, cinematic execution, and model-dependent controls.
- Look up the relevant model profile and capability matrix. If a capability is not in the references, mark it unknown rather than guessing.
- Choose prompt structure: natural prose, slot template, time-blocked prompt, shot list, or parameterized export.
- Translate negatives correctly: separate field where supported, positive restatement where unsupported, capped constraints for video.
- Map reference assets to supported syntax and roles. If unsupported, describe a workaround or mark unknown.
- Map settings: aspect ratio, duration, resolution, seed/reference controls, audio, and text/logo constraints.
- Produce a translation report noting preserved intent, changed syntax, unsupported fields, unknowns, and likely failure modes.
- For multi-model exports, keep one universal brief plus per-model prompts/settings.
What ships with it
22 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- checklists/model_export_checklist.md 333 B
- examples/failed_model_transfer_diagnosis.md 418 B
- examples/image_prompt_translation.md 646 B
- examples/same_scene_across_models.md 559 B
- examples/universal_prompt_to_model_specific_exports.md 1.1 KB
- examples/video_prompt_translation.md 627 B
- references/failure_modes_by_model.md 1.2 KB
- references/image_model_profiles.md 14 KB
- references/model_capability_matrix.md 4.0 KB
- references/model_update_protocol.md 697 B
- references/parameter_compatibility.md 553 B
- references/prompt_translation_rules.md 1004 B
- references/video_model_profiles.md 15 KB
- templates/model_comparison_table.md 152 B
- templates/model_specific_prompt_export.md 321 B
- templates/multi_model_export_package.md 208 B
- templates/prompt_translation_report.md 281 B
- templates/universal_prompt_brief.md 446 B
- tests/model_routing_tests.md 423 B
- tests/prompt_translation_tests.md 322 B
- tests/research_provenance_tests.md 355 B
- tests/unknown_capability_tests.md 220 B
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 11d ago First seen · 83 lines · 88 tokens per session scan A 98fba9bef326
model-adaptation is a skill published in the GitHub repository 0xhughs/director-skills (8 stars, last pushed 3mo ago), licensed MIT. It adds 88 tokens to every session and 854 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
higgsfield-seedance
Rewrites scene descriptions using professional cinematography language, structures prompts with a six-slot formula (camera + subject + action + setting + style + lighting), and diagnoses content filter rejections via a preflight linter. Use whenever the user asks for a Seedance 2.0 / Seedance Pro prompt, describes a…
higgsfield-prompt
Use when building, writing, refining, or structuring a Higgsfield AI prompt. Covers the MCSLA formula, prompt structure, narrative vs. timestamped formats, and how to write for both text-to-video and image-to-video workflows.
higgsfield-seedance-2-5
Seedance 2.5 prompt director — the omni-reference dialect. Routes the four generation modes (t2v / omnireference / videoedit / videoextension), writes explicit @Image/@Video/@Audio reference roles with exclusions, stages 30-second videos into end-state beats, and covers video editing, forward/backward extension…
higgsfield
Use this skill whenever the user asks anything about Higgsfield AI — writing or refining video/image prompts, choosing a model (Kling, Sora 2, Veo, Wan, Seedance, Minimax Hailuo, DoP, Soul, Nano Banana, Seedream, Flux, GPT Image, etc.), camera controls, named motion presets, Soul ID character consistency, Cinema…
higgsfield-acting
Writes the character-performance layer of a video prompt as behavior under pressure, not displayed emotion — objective, obstacle, tactics, beats, subtext, listening, body/status/proxemics, and mandatory eye life. Produces a reusable 150–220-word acting master profile per character plus a per-scene rewrite of it, and a…
higgsfield-assist
Use when the user asks about Higgsfield Assist (the built-in GPT-5 copilot), how to use the platform's native AI assistant, credit optimization strategies, plan selection, how to get more from fewer credits, or platform efficiency tips.