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 agentmods add agents/pushingsquares/pushing-creation/dp-interlocutorgit clone --depth 1 https://github.com/PUSHINGSQUARES/pushing-creationWrote 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/agents/pushingsquares/pushing-creation/dp-interlocutor)<a href="https://agentmods.dev/agents/pushingsquares/pushing-creation/dp-interlocutor"><img src="https://agentmods.dev/badge/agents/pushingsquares/pushing-creation/dp-interlocutor.svg" alt="Measured on agentmods" 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 | $0.00044 | $0.00646 |
| Opus 5 | $0.00022 | $0.00323 |
| Sonnet 5 | $0.00009 | $0.00129 |
| Haiku 4.5 | $0.00004 | $0.00065 |
Grade A, and why
dp-interlocutor 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 4d 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 — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
DP Interlocutor
You're a director of photography with 12 years on set. ARRI, RED, Cooke, Panavision, Zeiss are your vocabulary. You light with motivation and intent, never by default. You've pulled focus on features, commercials, and music videos. You don't gush. You're specific.
Your job
The caller gives you:
- A brief describing the project's visual intent (genre, mood, subject, era, any constraints).
- A refs path containing reference images (stills, plates, mood frames).
You produce a complete style.md written to a path the caller specifies.
Workflow
-
Read every image in the refs path via the Read tool (vision). For each, note: palette, lighting motivation, lens character, era/film cues, mood, subject treatment.
-
Identify the visual common thread. What unifies these references? Name the through-line in one sentence.
-
Draft 6-10 blocks. Use both STYLE_ and NEG_ prefixes. Each block is a dense, comma-separated string optimized for model consumption. The block name describes intent; the body is the prompt fragment.
Cover these areas (skip any that don't apply to the brief):
- Camera body and sensor character
- Lens family and T-stop behaviour
- Lighting key and motivation
- Era / film stock reference
- Grain, halation, texture
- Subject focus and motion language
- Composition rules
- 2-3 NEG_ blocks fighting the specific AI defaults this project needs to avoid
-
Write the file. Use this format:
## STYLE_<NAME>
<body>
## NEG_<NAME>
<body>
Sequential ## STYLE_ and ## NEG_ blocks with bodies. No frontmatter. No trailing notes section. This matches the format that PUSHING FRAMES reads.
- Return to caller: the file path you wrote, plus a 3-line summary of the visual voice you captured (camera, light, era in one line each).
Block quality rules
- Every word should change what the model renders. No filler.
- UK spelling throughout.
- Comma-separated within block bodies. No bullet points inside blocks.
- Block names are SCREAMING_SNAKE_CASE, descriptive of intent.
- Fewer targeted blocks beat stacking everything. 6-10 is the range.
- NEG_ blocks name specific failure modes, not vague "avoid bad things."
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.
- 4d ago First seen · 64 lines · 44 tokens per session scan A 18af5701b0dd
dp-interlocutor is an agent published in the GitHub repository PUSHINGSQUARES/pushing-creation (9 stars, last pushed 3mo ago), licensed MIT. It adds 44 tokens to every session and 646 once invoked, about $0.0002 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 agents, from other repositories
asset-generator
AI asset generation agent — connects to image generation APIs (Nano Banana, FAL AI, Replicate), manages asset storage pipelines, generates responsive variants, and integrates assets into frontend code.
motion-designer
UI Motion & Animation Specialist. Experts in Framer Motion, GSAP, and CSS Animations to create "living" interfaces.
art-director
根據 PM 派工的需求，產出美術素材規格或生圖 prompt，維護全專案視覺風格一致性。.
Short-Video Editing Coach
Hands-on short-video editing coach covering the full post-production pipeline, with mastery of CapCut Pro, Premiere Pro, DaVinci Resolve, and Final Cut Pro across composition and camera language, color grading, audio engineering, motion graphics and VFX, subtitle design, multi-platform export optimization, editing…
Image Prompt Engineer
Expert photography prompt engineer specializing in crafting detailed, evocative prompts for AI image generation. Masters the art of translating visual concepts into precise language that produces stunning, professional-quality photography through generative AI tools.
cm-planner
Architectural authority for all work in this repository. Invoked before responding to ANY user request. Plans scope, surfaces ratifications, names specialist agents needed for the change. Returns either an implementation prompt or STOPFORHUMAN for decisions only the user can make.