Agent skill · Audited from source
female-portrait-director
skill · liyue-aigc/female-portrait-directorMIT licenceAudited 20 Oct 2026 · source 56379b2b2063
Is it safe? Yes — SAFE, grade B (89/100). We read this skill's source at commit 56379b2b2063 and found no filesystem, network or shell access and no hard-coded secrets. What it does: Generate, visually expand, optimize, diagnose, and route structured AI image prompts for adult female portraits. Use for lifestyle, curve-focused, fashion, oriental, fantasy, realism, beauty, CCD, low-key cinematic,…
Cite
Cite this page
ModelsAtlas. “female-portrait-director (liyue-aigc/female-portrait-director)”. Retrieved 20 Oct 2026. https://modelsatlas.com/skills/liyue-aigc/female-portrait-director
Trust audit
Audit 0.4.1 · source read 20 Oct 2026
SAFE means nothing in the source contradicts what the skill says it does. Grade B is shared by 13,985 of 14,816 audited skills.
What the source does
- FilesystemNone observedNo reads or writes in the source
- NetworkNone observedNo outbound calls in the source
- ShellNone observedNo commands run in the source
- DependenciesPinnedNothing floats
- SecretsNone foundNo keys in source
Five audit layers
| Layer | What we check | Result | Findings |
|---|---|---|---|
L0 Provenance | Repo owner, licence, archive status, activity | PASS | 0 |
L1 Static source | Secrets, path traversal, supply chain, hidden code | NA | 0 |
L2 Instructions | Prompt injection, authority framing, hidden directives | PASS | 0 |
L3 Inventory | The files and declarations match the source | PASS | 0 |
L4 Sandbox | Run in isolation and watch behaviour | SKIPPED why | 0 |
Gates applied: no behavioural pass.
Findings (0)
No findings at any severity. We re-audit when the source changes.
Audit 0.4.1 · source sha 56379b2b2063 · 20 Oct 2026Report an issue or request a re-scan
Provenance
Who published it, and what the verdict is pinned to
- Publisherliyue-aigc GitHub userOwner of the repository this skill is read from
- Source folderliyue-aigc/female-portrait-director/skills/female-portrait-directorThe exact folder we audited
- LicenceMITAs the repository states it
- Commit pinned for this verdict56379b2b2063A new audit runs when the source changes
- Repository statusNot archivedLast push 97 days ago
- Sandboxed runNot runSkills are read, not executed; layer L4 is skipped
What it tells the agent
Full source on GitHub ↗Its description: Generate, visually expand, optimize, diagnose, and route structured AI image prompts for adult female portraits. Use for lifestyle, curve-focused, fashion, oriental, fantasy, realism, beauty, CCD, low-key cinematic,…
--- name: female-portrait-director description: Generate, visually expand, optimize, diagnose, and route structured AI image prompts for adult female portraits. Use for lifestyle, curve-focused, fashion, oriental, fantasy, realism, beauty, CCD, low-key cinematic, studio, sport, travel, and e-commerce portrait requests; onboarding or usage help; parameter recommendations; prompt rewrites; image-to-prompt preparation; direct image generation; or identity- and product-preserving edits from authorized reference images. Preserve explicit user parameters and locked reference subjects while creating a coherent photography-directed scene. --- # Female Portrait Director Turn a small set of portrait parameters into a stable, visually directed, copy-ready image prompt for a fictional adult woman. Expand through visual reasoning, not summary writing or mechanical field filling. ## Required loading order 1. If the user asks how to use the Skill, says it was just installed or this is their first call, invokes only `$female-portrait-director`, or provides no actionable portrait parameters, read [skill/help.md](skill/help.md) and return its concise onboarding tutorial. Do not load a Route for a help-only request. 2. Read [skill/skill.md](skill/skill.md) for the canonical workflow. 3. For optimization, diagnosis, parameter recommendation, safety rewrite, image-to-prompt, or reference-image direct-generation tasks, read [skill/tool-registry.md](skill/tool-registry.md) and only the selected file under `skill/tools/`. 4. Read [skill/style-registry.md](skill/style-registry.md). Select exactly one implemented primary route when the task needs a visual direction. Do not invent a placeholder extension route. 5. Read only the selected file under `skill/routes/`. 6. If the request contains a compatible temperament direction, read [skill/overlay-registry.md](skill/overlay-registry.md) and only the selected file under `skill/overlays/`. 7. If uploaded images must remain recognizable, read [skill/core/reference-image-lock.md](skill/core/reference-image-lock.md) and form an image-role lock table before visual expansion. 8. Read [skill/core/director-gate.md](skill/core/director-gate.md). Complete its internal director-design phase before writing or generating the final image prompt. 9. Read only the relevant sections of the linked `core/` and `references/` files described by [skill/skill.md](skill/skill.md). ## Operating rules - Lock explicit user parameters, including palette direction and visual richness when provided. Expand them without silently replacing them. - Treat onboarding as a separate help mode. Show the V1.6 ready message, all 20 implemented styles, the minimal and advanced input templates, Route-plus-Overlay guidance, one parameter combination followed by its complete five-paragraph copy-ready prompt and separate negative constraints, and direct-image instructions from [skill/help.md](skill/help.md). Do not summarize, shorten, or replace the detailed example with a promise of what the Skill would generate. If the same request already contains actionable portrait parameters, perform the requested task and add only the short ready hint instead of repeating the full tutorial. - In standard detailed output, the parameter lock result is a complete field-by-field record of the user's explicit input. Do not merge, paraphrase away, omit, or replace explicit fields with route or overlay notes. Add inferred defaults only as clearly labeled supplements. The prohibition on mechanical field filling applies to the director expansion and final fused prompt, not to the parameter lock result. - For identity- or product-preserving reference-image tasks, lock image roles and protected subject features before applying the selected route. - Use the registries as the only routing entry points. Load one primary route, one task tool when needed, and optional compatible overlays only when their files exist. - Match the V1.5 / V1.6 Routes by compound fingerprints in [skill/style-registry.md](skill/style-registry.md). Never choose among the three CCD curve Routes from the words `CCD` or `curve` alone. Select `low-key-cinematic-photography` only from the combined low-light, local continuous-light, readable-shadow, restrained-color, and cinematic-still fingerprint; `dark` or `cinematic` alone is insufficient. Never load two primary Routes as a style blend. - Complete the internal director-design phase before composing the final prompt. Then expand age cues, facial features, expression, body direction, pose, clothing, palette direction, visual richness, scene, camera, lighting, filter, and platform adaptation into one photographed moment. - Build one coherent photographed moment with a time slice, one small event, an action chain, a gaze target, and two or three selective environment details. - Use route templates as visual direction, not as sentence banks. Do not return a summary or mechanically fill fields. - Unless the user explicitly requests a concise output, use the standard detailed output: parameter lock result, public-facing director module expansion, final fused prompt, and negative constraints. The director module expansion must explain the visible design choices for age and facial features, body and pose, clothing, palette and accessories, visible richness, scene, camera, and lighting and filter in readable paragraphs. Do not reduce it to a short field recap. - Standard detailed output is the default whenever the user does not explicitly request a concise mode. Its final fused prompt must contain exactly five substantial paragraphs: (1) person, age, facial features, makeup, and temperament; (2) time slice, small event, pose, action chain, and gaze; (3) body direction, line priorities, clothing structure, palette, materials, accessories, and visible richness; (4) scene, selective environment details, depth layers, camera, composition, and depth of field; (5) lighting direction, light placement, highlights, shadows,
Excerpt, first 6,000 characters of SKILL.md, verbatim from the audited commit. It is text from someone else’s repository, quoted so you can judge it before your agent reads it.
Install
Pinned to the audited commit 56379b2
git clone https://github.com/liyue-aigc/female-portrait-director git -C female-portrait-director checkout 56379b2b2063
The author documents no install command, so this uses the repository directly. Our verdict covers commit 56379b2b2063 only.
Works in
The author has not said which agents this skill supports, and we have not tested it in any.
Audit history
Every audit, every source change
- ✓SAFE B 89 / 100
first audit · source
56379b2b2063
README badge
For maintainers · shows the latest audit
[](https://modelsatlas.com/skills/liyue-aigc/female-portrait-director/)
The badge shows our verdict and grade for this skill, and links back to this page. It changes when the audit does.
Similar skills
juliusbrussee/caveman
Find and label every LLM workflow in the repository so Caveman Cloud groups spend by workflow instead of one bucket. Use for "discover workflows" or breaking LLM spend down by workflow.
juliusbrussee/caveman
Turn a Caveman optimization observation into an operator-chosen candidate with a paired baseline evaluation. Use when asked to inspect or evaluate a Caveman optimization report. Needs explicit approval.
juliusbrussee/caveman
Act on a Caveman learn report - review the ranked token sinks, apply cost-lowering fixes with per-edit consent, and report what those fixes returned. Use when asked to lower an agent's token cost, what caveman has…
juliusbrussee/caveman
Fix bugs and small behavior changes at the narrowest responsible layer. Use when regression proof, preserved surrounding behavior, and task-relevant tests matter.
juliusbrussee/caveman
Prove existing work meets acceptance conditions without expanding scope. Use for validation-only tasks, completion checks, focused gate runs, and last-mile proof.
juliusbrussee/caveman
Diagnose ambiguous failures before editing. Use for unknown causes, intermittent behavior, performance regressions, or investigations needing evidence-ranked hypotheses.
Add this badge to your README
[](https://modelsatlas.com/skills/liyue-aigc/female-portrait-director/)
<a href="https://modelsatlas.com/skills/liyue-aigc/female-portrait-director/"><img src="https://modelsatlas.com/skills/liyue-aigc/female-portrait-director/badge.svg" alt="ModelsAtlas: SAFE B"></a>
Questions about female-portrait-director
What does the female-portrait-director skill do?
Is female-portrait-director safe to install?
What can female-portrait-director access on my machine?
How current is this page?
Sources
- github.com/liyue-aigc/female-portrait-director/tree/main/skills/ — the audited source at commit 56379b2b2063 and every finding aboveObserved · 20 Oct 2026
Spotted something wrong? Report a correction · Methodology
Keep going in your AI
Opens with this page as the source. We add nothing else to the prompt.
- What does the female-portrait-director skill do?
- Is female-portrait-director safe to install?
- What can female-portrait-director access on my machine?