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/ihuzaifashoukat/splicedeck/agents-mdgit clone --depth 1 https://github.com/ihuzaifashoukat/splicedeckWrote 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/instructions/ihuzaifashoukat/splicedeck/agents-md)<a href="https://agentmods.dev/instructions/ihuzaifashoukat/splicedeck/agents-md"><img src="https://agentmods.dev/badge/instructions/ihuzaifashoukat/splicedeck/agents-md.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 | $0.01765 | $0.01765 |
| Opus 5 | $0.00882 | $0.00882 |
| Sonnet 5 | $0.00353 | $0.00353 |
| Haiku 4.5 | $0.00177 | $0.00177 |
Grade A, and why
splicedeck 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 5d 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 — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
splicedeck — agent contract
Read by Codex, Claude Code, Cursor, Gemini CLI and any agent that honours
AGENTS.md. Read this before touching anything.
A local-first video editor that an AI agent drives. One pass over a source produces both a cleaned long-form master and vertical clips. It remembers how each party likes to be cut.
The governing idea
A bookmark is the base. The casebook is the delta. findings.lock.txt is the
receipt.
A style bookmark supplies opening settings. Per-party memory overrides individual
fields where a human shipped something different and kept it. The resolver writes
its answer to a committed lock file, so git diff reads as a behaviour changelog
and git revert is a real behavioural rollback.
Hard rules
- Pure-Python dependencies only. No compiled extension on any default path,
ever, including transitively. On a clean Windows box with no Microsoft Visual
C++ redistributable,
ctranslate2andonnxruntimeboth fail to load their DLLs. Speech recognition runs as a subprocess against an external binary, never as an imported wheel. A pull request that adds a wheel with a binary artifact on the default path is rejected regardless of what it does. - Local-complete, cloud-optional. A fresh clone with no API keys and no cloud account must render a finished, delivered file. Cloud tiers are opt-in accelerators behind the same interface. Nothing may become required.
- ffmpeg is a child process. Never linked, never vendored, never bundled.
Every filter-embedded path is escaped or made relative, because a Windows
drive letter breaks filter parsing:
:is the filter option separator. - Capability is proven by trial, never by feature list.
ffmpeg -encoderslists encoders the hardware cannot run. On the reference machineh264_nvencis listed and fails at runtime because the GPU has no encoder silicon, whileh264_qsvsucceeds. Detect by test-encoding. weld/is pure. It builds argv lists and filtergraph strings and returns them. No subprocess, no disk, no network, no clock. This is what makes the whole decision layer testable in CI with no ffmpeg and no GPU. Do not put I/O in it "just this once".- All time is an integer tick. No float ever enters the time domain. Ranges
are
in/out/at, never origin-plus-extent, so moving a cut is one number changing. - Nothing durable is authored by the model. Every memory tool argument is an identifier the pipeline minted — never a value, a string, or a path. A case records something a human did; the pipeline writes every field from its own ledger.
- The server may write only
casebook/pending/. It may readcasebook/parties/,casebook/schema/,bookmarks/andledger/. Skills, prompts and executables are unreachable because nothing can reach them, not because something refuses. - A cut never lands inside a word. Snap to word boundaries from the index.
- Describe the look, never the source. No preset, bookmark, axis value, element, directory, file, party identifier, commit message or documentation line may be named after — or documented as targeting — any real person living or dead, or any third-party brand, product, channel, publication or trade dress. This binds the casebook write gate too.
- Media never enters git. Sources are content-addressed; the repository holds pointers. The hash is authoritative, the path is advisory.
- Every failure is typed, and every refusal carries its correction as data — a list of ready-to-send calls, not prose the caller must parse.
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.
- 5d ago First seen · 137 lines · 1,765 tokens per session scan A befa1e9296b5
splicedeck AGENTS.md is an instructions file published in the GitHub repository ihuzaifashoukat/splicedeck (3 stars, last pushed 17d ago), licensed Apache-2.0. It adds 1,765 tokens to every session, about $0.0088 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 instructions, from other repositories
intelligent-terminal rust.instructions.md
Concise Rust coding conventions for this repository.
Browser4 CLAUDE.md
Claude Code instructions for platonai/Browser4, covering browser4 — project context for claude, architecture, key dispatch chain (cli → browser), batch commands and e2e test structure.
apex-accelerator vendor-prompting.instructions.md
Vendor prompting best-practice rules for Anthropic Claude and OpenAI GPT-5.6-Terra agents and prompts. Each rule cites a rule ID in the vendor-prompting skill rules.json registry. Validator: npm run lint:vendor-prompting.
autopus-adk GEMINI.md
Gemini CLI instructions for Insajin/autopus-adk, covering autopus-adk harness, 스킬 디렉터리, core guidelines, subagent delegation and file size limit.
agent-ship CLAUDE.md
Claude Code instructions for Agent-Ship/agent-ship, covering claude.md, project overview, commands, docker development (recommended) and local development (no docker).
clawock AGENTS.md
AGENTS.md instructions for KCNyu/clawock, covering agents.md - your workspace, every session, kcn 偏好, git hook (one-time setup per clone) and git auto-commit rules.