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 instructions/hiyeshu/codeck/agents-mdgit clone --depth 1 https://github.com/hiyeshu/codeckWhat 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 | $0.01088 | $0.01088 |
| Opus 5 | $0.00544 | $0.00544 |
| Sonnet 5 | $0.00218 | $0.00218 |
| Haiku 4.5 | $0.00109 | $0.00109 |
Grade A, and why
codeck AGENTS.md 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- codeck CLAUDE.md — 100% identical, 2 lines differ
How it starts
The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.
codeck — AI presentation skill
Architecture
codeck outputs a single HTML file, built by build-html.sh wrapping assemble.sh:
| Author | File | Role |
|---|---|---|
| Human (fixed) | engine.js + engine.css + build-html.sh |
Navigation, fragments, overview, speaker mode, progress bar, final HTML validation |
| AI (per deck) | custom.css |
:root variables + layout primitives + per-page styles + mobile |
| AI (per deck) | slides.html |
<section class="slide" data-notes="..."> free HTML |
Engine code is fixed. AI handles content and visuals only.
Deck room
/codeck opens ~/.codeck/projects/{slug}/
↓
MEMORY.md + channel/ + tasks/ + threads/ + roles/
↓
Decision Ask records live in threads/ before any runtime question
↓
@outline → deck.md
↓
@design → DESIGN.md + custom.css + slides.html → build-html.sh → single HTML
↓
@review → export (PDF/PPTX) / speech
Core idea: fixed role lanes own artifacts; dynamic people from diagnosis.md shape the judgment inside those lanes. The room is the durable scope; necessary asks are decision records first and UI questions second.
Room documents have rank:
- Current truth: MEMORY.md, deck.md, DESIGN.md, custom.css, slides.html, latest assembled HTML, speech.md when present.
- Work state: diagnosis.md, active tasks, open threads plus the Decision Ask ledger, roles/*.md, latest valid review.md.
- Audit trail: channel/YYYY-MM-DD.md, legacy PROJECT.md, legacy design-notes.md, superseded reviews, and old previews. Audit never overrides current truth.
Three diagnostic signals
- Domain — determines outline role
- Expression challenge — determines design role
- Audience starting point — determines review role (inverse: listener most likely to struggle)
Directory structure
Skills resolve their install location in this order: CODECK_SKILL_DIR / CODECK_DESIGN_DIR / CODECK_EXPORT_DIR env overrides → ${CLAUDE_PLUGIN_ROOT}/skills/* (Claude Code plugin install) → plugin cache globs (~/.claude/plugins/cache/*/codeck/*/skills/*, ~/.codex/plugins/cache/*/codeck/*/skills/*) → flat installs (~/.agents/skills/codeck*/, ~/.codex/skills/codeck*/, ~/.claude/skills/codeck*/). The first match wins, so the same SKILL.md works as a Claude Code plugin, a Codex plugin, or a flat skills.sh install. Sibling lane directories are derived by codeck/scripts/resolve-dirs.sh (siblings hold in every layout), sourced from each skill's unified Setup bootstrap block.
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 First seen · 78 lines · 1,088 tokens per session scan A 758f1c252557
codeck AGENTS.md is an instructions file published in the GitHub repository hiyeshu/codeck (199 stars, last pushed 24d ago), licensed MIT. It adds 1,088 tokens to every session, about $0.0054 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 2026-08-30.
Other instructions, from other repositories
ppt-master AGENTS.md
Instructions for hugohe3/ppt-master, covering agents.md, project overview, execution requirements, required conventions and compatibility boundary.
ppt-master CLAUDE.md
Instructions for hugohe3/ppt-master: Claude Code entry point. To avoid maintaining two copies, all project instructions live in a single source of truth — AGENTS.md — imported below (Claude Code inlines @-referenced files into context). Always edit AGENTS.md; never duplicate its content here.
PPTAgent AGENTS.md
Instructions for icip-cas/PPTAgent, covering project and user background, purpose, ground truth, core philosophy and 1. good taste first.
genoffice CLAUDE.md
Instructions for genspark-ai/genoffice, covering claude.md, theming rules (mandatory) and build gotchas.
office-open-xml-viewer AGENTS.md
Instructions for yukiyokotani/office-open-xml-viewer, covering agent instructions, project, standard development flow, git and pr rules and verification.
office-open-xml-viewer CLAUDE.md
Instructions for yukiyokotani/office-open-xml-viewer, covering claude.md, worktree startup checklist, claude worktree roles, parallel session safety and claude autonomous work.