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 Glad-Labs/poindexter --skill storytellinggit clone --depth 1 https://github.com/Glad-Labs/poindexterWrote 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/glad-labs/poindexter/storytelling)<a href="https://agentmods.dev/skills/glad-labs/poindexter/storytelling"><img src="https://agentmods.dev/badge/skills/glad-labs/poindexter/storytelling/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/glad-labs/poindexter/storytelling"><img src="https://agentmods.dev/badge/skills/glad-labs/poindexter/storytelling.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 | $0.00092 | $0.10968 |
| Opus 5.5 | $0.00037 | $0.04387 |
| Sonnet 5.5 | $0.00018 | $0.02194 |
| Haiku 4.5 | $0.00009 | $0.01097 |
Grade A, and why
storytelling 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 today.
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 — 641 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Storytelling skill
Screenplays for the creative studio (poindexter#1134), written in two
steps: story.outline plans the scenes, then story.scene writes each one
against its share of the word budget. The cast is personas; code formats each
one as slug: name; personality; looks in {cast_block}. The word budget comes
from story_speech_words_per_second × the requested length, and the scene count
from story_seconds_per_scene. Code validates every reply and sends problems
back once through story.screenplay_fix.
story.outline
You are the head writer at {site_name}, outlining a short animated screenplay.
ASSIGNMENT
Topic: {topic}
Brief: {brief}
Format: {format_name}
Tone: {tone}
Audience: {audience}
Length: about {target_seconds} seconds of audio across about {scene_count} scene(s).
{source_block}
CAST — the only characters who may appear. Use the id before the colon exactly.
{cast_block}
Narrator available: {narrator_available}.
FORMATS — follow the one named above.
- explainer: one idea taught clearly. Open with a hook question, build the idea in small steps with a concrete example, end on the one thing to remember.
- story: a character wants something, meets an obstacle, and changes or learns by the end. The topic is the world the story happens in.
- dialogue: two or more cast members talk it through. One is curious and asks; the other knows and explains; both have personality.
- sketch: a short comedic scene built on one funny premise drawn from the topic, escalating to a punchline.
- visual: a story told in what the audience sees, like a silent short. The situation is clear from the first shots without anyone explaining it, something funny or surprising happens every ten to fifteen seconds, and it ends on a button: a last gag or image that pays off the opening. Characters speak only a few short lines, where a word lands harder than a picture.
FIND THE STORY FIRST
Every format, an explainer included, is a story about a character. Decide the spine before the scenes:
- "want": which cast member wants what, in their own words and as something concrete.
- "obstacle": what stands in the way, often a belief the character holds that is wrong.
- "turn": the moment the character sees it differently, and what makes it happen.
- "payoff": how the last scene answers the first: the opening problem solved, a line called back, the character changed.
The topic is what the character learns or runs into; the audience learns it by watching them.
When the topic is a well-known tale, tell that tale: keep its best-known moments, lines and ending, retold in fresh words, and let its plot be the spine.
- "main_event": the one event the whole piece is about, the thing its title and its turn depend on, in one sentence. It happens on screen, in an action the audience watches, never reported after it happened.
TRANSLATE AN ADAPTED TALE
When there is a SOURCE above, write "translation" first: one entry for each of the tale's key elements (its hero, what the hero guards or wants, the threat, the people around them, the key objects and the key acts), each with "tale" (the element in the tale) and "here" (its equivalent in this piece's world). For each key act, "why" gives its cause in the tale (what makes the character do it, or not do it); the adaptation keeps that cause, so the act happens here for the same reason. Choose each equivalent so a viewer of this world reads it at a glance and it fits the world's look, technology and daily life. Then write the outline from the translation: everything on screen is this world's, and nothing from the tale's own setting appears unless it already belongs here.
PLAN THE PLACES
- List the story's locations once under "locations": each has a short "id" (lowercase words joined by hyphens) and a "description" of the place itself that a background artist could paint: its architecture, objects, scale and look. Leave the time of day and mood to the scenes.
- A place the story returns to is the same location, so it is drawn the same way every time.
PLAN THE COLOUR
- Paint a colour script: each scene gets a "palette", the colours and light it is painted in (the main hues, the time of day, where the light comes from and how warm or cold it is), in one short phrase.
- A palette describes only colours and light. It is painted over every picture of the scene, the characters included, so the characters and what happens stay in the summary.
- Plan the palettes together, as one strip across the whole story. Colour carries the feeling: let it shift when the story turns, contrast the low point with the payoff, and come back to a palette when the story comes back to a feeling.
- Keep every palette inside the piece's art style; the style is the same throughout and only the colours and light change.
PLAN THE SCENES
- Give the piece about {scene_count} scene(s). Each scene is one step of the spine, and something is different at its end: a new understanding, a decision, a feeling.
- Keep the characters in one main place unless the story has a reason to move. When it moves, the summary says why (they go to try something, or a demonstration takes them there).
- "location": the id of the location the scene happens in.
- "setting": that place at this moment, with its time of day and mood, in one sentence a background artist could paint. It describes only the place; who is there belongs in "characters".
- "palette": the scene's entry in the colour script.
- "summary": what happens in the scene and what the audience learns or feels, in one or two sentences.
- "turn": what is different by the end of this scene, in one sentence.
- "characters": the cast ids visible in the scene.
- The first scene sets up the want and hooks the audience; the last one delivers the payoff.
OUTPUT
Reply with one JSON object and nothing else, shaped like this:
{{"title": "a short title", "logline": "the whole piece in one sentence", "spine": {{"want": "...", "obstacle": "...", "turn": "...", "payoff": "..."}}, "main_event": "...", "translation": [{{"tale": "...", "here": "...", "why": "..."}}], "locations": [{{"id": "workshop", "description": "a cluttered workshop with a long workbench, hanging tools and a tall rain-streaked window"}}], "scenes": [{{"scene_id": "s1", "location": "workshop", "setting": "the workshop at night, warm lamps, rain on the window", "palette": "deep blue night with warm amber lamplight from the bench", "summary": "what happens here", "turn": "what is different by the end", "characters": ["id"]}}]}}
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.
- today Changed · +1 lines 873cef5f7ac0
- yesterday Changed · +202 lines 57f0498d829c
- 3d ago Changed · +289 lines a39d002b224c
- 4d ago First seen · 149 lines · 92 tokens per session scan A 872c0a5e0956
storytelling is a skill published in the GitHub repository Glad-Labs/poindexter (6 stars, last pushed today), licensed Apache-2.0. It adds 92 tokens to every session and 10,968 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-10-13.
Other skills, from other repositories
datamodel-code-generator
Use this skill when the user wants Python data models, Pydantic models, dataclasses, TypedDicts, msgspec structs, or type-safe Python classes generated from OpenAPI, AsyncAPI, JSON Schema, GraphQL, JSON/YAML/CSV sample data, MCP tool schemas, Protocol Buffers, XML Schema, Apache Avro, or existing Python model objects.…
linkedin-employee-advocacy
Stand up and run a LinkedIn employee advocacy program for a marketing or sales team. Covers 14-day launch playbook, brand-guideline governance, per-post time budget, cadence benchmarks, and team ROI (reach, engagement, pipeline). Triggers on "employee advocacy", "get the team posting", "scale LinkedIn across team"…
company-brain
How an agent should operate as one mind inside a shared Company Brain built on Caura — recall before acting, obey fleet keystones, reuse and publish skills, and report outcomes so every task compounds across the whole organization. Use this whenever you work as part of a Caura-connected team or fleet and your work…
secops-investigate
Expert guidance for deep security incident and entity investigations in Google SecOps. Use when investigating cases, analyzing entities (hosts, IPs, domains, hashes, users), extracting and searching UDM events, performing asset and user timeline analysis, and detecting lateral movement across enterprise networks.…
aws-incident-response
Investigate AWS credential compromise, STS session abuse, and API breaches; produce an evidence-backed timeline, containment plan, and incident handoff.
refactor-planner
Plan safe refactoring with dependency analysis, impact assessment, and rollback strategies.