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/sazardev/shiki/agents-mdgit clone --depth 1 https://github.com/sazardev/shikiWhat 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 | $0.01429 | $0.01429 |
| Opus 5 | $0.00714 | $0.00714 |
| Sonnet 5 | $0.00286 | $0.00286 |
| Haiku 4.5 | $0.00143 | $0.00143 |
Grade A, and why
shiki 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.
How it starts
The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Read this first
CLAUDE.mdis the exhaustive source of truth (layout, keybindings, config schema, git sync, release automation, per-file design rationale). Read the relevant section before any architectural change.IDEA.mdis the design spec.AGENTS.mdonly covers what you'd otherwise guess wrong.- This is a Rust Cargo workspace (
shiki-core→shiki-config→shiki-tui→shiki-cli, strict one-way deps). Binary name isshiki(inshiki-cli/src/main.rs), launched with no args = TUI,-- <args>= CLI subcommands.
Commands
cargo check --workspace # fast iteration
cargo test --workspace # the whole suite
cargo clippy --workspace --all-targets -- -D warnings # CI gate — the -D warnings matters
cargo fmt --all # then re-clippy
cargo audit # CI runs this; ignore list lives in .cargo/audit.toml
cargo run -p shiki-cli -- --help # CLI; no args launches the TUI
- CI (
ci.yml) runsfmt --check(ubuntu only),clippy -D warningsandcargo teston all 3 OSes. Clippy is matrixed becauseshiki-core/src/editor.rshas#[cfg(target_os = ...)]blocks. - Never commit a clippy failure;
-- -D warningsis enforced in CI.
Testing
- There are ~293
#[test]s (134 inshiki-core, 17shiki-config, 129shiki-tui, 13shiki-cli) — CLAUDE.md's "almost no tests yet" paragraph is stale. They are all inline#[cfg(test)]modules inside source files (notests/dirs, no#[ignore], no fixture setup).cargo test --workspaceis green. - To exercise the CLI/TUI without touching real user data, override XDG dirs (used via
directories::ProjectDirs::from("", "", "shiki")):
XDG_CONFIG_HOME=/tmp/shiki-test-config XDG_DATA_HOME=/tmp/shiki-test-data cargo run -p shiki-cli -- notebook create personal
- For
shiki-tuilogic, prefer pure functions over constructing a fullAppin a test — the pattern ispanel_drawer::drawer_hit_at(notebook_count, area, column, row)(takes plain numbers, not&App).
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 · 58 lines · 1,429 tokens per session scan A b4dfd2b7497d
shiki AGENTS.md is an instructions file published in the GitHub repository sazardev/shiki (289 stars, last pushed 5d ago), licensed MIT. It adds 1,429 tokens to every session, about $0.0071 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
kimun CLAUDE.md
Instructions for nico2sh/kimun, covering kimun, docs and rules.
reflect-open AGENTS.md
AGENTS.md instructions for team-reflect/reflect-open, covering purpose, what is reflect, product principles, agent workflow and development workflow.
siyuan AGENTS.md
AGENTS.md instructions for siyuan-note/siyuan, covering agents.md, 1. non-negotiable constraints, do not hand-edit, verification and prohibited operations and 2. project-specific rules.
reflect-open CLAUDE.md
Claude Code instructions for team-reflect/reflect-open: See AGENTS.md for the full project overview, tech stack, database tables, code conventions, and development cycle.
zen CLAUDE.md
Claude Code instructions for sheshbabu/zen, covering claude.md, development commands, build commands, dependencies and go build tags.
hippocampus CLAUDE.md
Instructions for sturlese/hippocampus, covering layout, frontmatter schema, conventions, answering questions from the vault (read order) and contradictions.