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 jongwony/epistemic-protocols --skill sketchgit clone --depth 1 https://github.com/jongwony/epistemic-protocolsWrote 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/jongwony/epistemic-protocols/sketch)<a href="https://agentmods.dev/skills/jongwony/epistemic-protocols/sketch"><img src="https://agentmods.dev/badge/skills/jongwony/epistemic-protocols/sketch/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/jongwony/epistemic-protocols/sketch"><img src="https://agentmods.dev/badge/skills/jongwony/epistemic-protocols/sketch.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 | $0.00081 | $0.10125 |
| Opus 5.5 | $0.00032 | $0.04050 |
| Sonnet 5.5 | $0.00016 | $0.02025 |
| Haiku 4.5 | $0.00008 | $0.01013 |
Grade A, and why
sketch 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 2d 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 — 582 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Hypotyposis Protocol
Discover what a form should be by encountering concrete versions and reacting to them, until the user recognizes one version they encountered as the form for their purpose. Type: (FitUnrecognized, Hybrid, SKETCH-RECOGNIZE-CYCLE, FormIntentSeed) → RecognizedForm.
Definition
Hypotyposis (ὑποτύπωσις): an outline or sketch — a first drawing of a position rather than its finished statement. A dialogical act for the moment when a form has to be made and the intent behind it cannot yet be settled from descriptions: the user can say what is wrong with a thing in front of them long before they can say what the thing should be, because good fit has no positive description of its own while each misfit can be pointed at (Alexander, 1964). So the user need not specify the form first. The AI relays what the next versions will try — what they revise, how many, the medium each is realized in and what that medium can and cannot show, and where each will be — each with its basis, and then makes them. Each round shows the new versions first, in a medium in which the qualities the user's judgment concerns can be encountered, and then the standing: what the user's words set and on which version, what of the AI's they took, what rests on a check, and what stays unestablished. The user reacts in their own words — a mark, a keep, a request, a value set — and the AI revises the versions those words point to into new versions, never regenerating one from a description of it, and may also make versions it would propose. The AI's reading of the purpose grows from those reactions rather than being settled up front: each part names the person's words or the gathered material it came from, or is marked as a default of its own. The run closes when the user takes one version they encountered as the form for their purpose, its home shown in the round or named in that same turn; they can stop at any time, and the run stays as it stood, with what they ask done to the versions as they stop carried out. Once it has run, the user can point at one version as they met it and say "this is the form, for this purpose", seeing where it lives, where the passed-over versions are, what they asked done to any of them that is not yet observed done, and which values rest on their own words or on the AI's proposals they took, and what rests on checked evidence. The recognition establishes their judgment of fit for that purpose on what the version showed — not correctness against a standard outside preference, and not readiness for use.
What ships with it
1 file 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.
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.
- 2d ago Changed 325bf7c9646d
- 10d ago Changed · +36 tokens per session 3008c597c363
- 20d ago Changed · -2 lines 05e061729d2d
- 24d ago Changed 3e74c4c94fd9
- 26d ago Changed · +10 lines 77365179950e
- 28d ago Changed · +8 lines 95694ecd4287
- 29d ago Changed · -10 lines 822233409167
- 1mo ago Changed · -170 lines · +9 tokens per session 306ade20cb1d
- 1mo ago Changed 6e527e408421
- 1mo ago First seen · 746 lines · 36 tokens per session scan A 94640dbc90c9
sketch is a skill published in the GitHub repository jongwony/epistemic-protocols (176 stars, last pushed yesterday), licensed MIT. It adds 81 tokens to every session and 10,125 once invoked, about $0.0003 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 2026-09-07.
sketch compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| sketchjongwony/epistemic-protocols | Skill | 176 | yesterday | MIT | A | 81 | Claude Code |
| taiyi-ui-designDong90/oh-my-taiyiforge | Skill | 855 | 1mo ago | MIT | A | 35 | Claude Code, Codex |
| taiyi-diagram-archDong90/oh-my-taiyiforge | Skill | 855 | 1mo ago | MIT | A | 48 | Claude Code, Codex |
| taiyi-restyleDong90/oh-my-taiyiforge | Skill | 855 | 1mo ago | MIT | A | 39 | Claude Code, Codex |
| taiyi-diagram-pipelineDong90/oh-my-taiyiforge | Skill | 855 | 1mo ago | MIT | A | 52 | Claude Code, Codex |
| taiyi-diagram-renderDong90/oh-my-taiyiforge | Skill | 855 | 1mo ago | MIT | A | 50 | Claude Code, Codex |
| review-core-correctnessexplyt/spring-plugin | Skill | 167 | 2d ago | Apache-2.0 | A | 67 | Claude Code, Codex |