Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add V-Songbird/slag/plugin install assayWrote 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/v-songbird/slag/craft-rules)<a href="https://agentmods.dev/skills/v-songbird/slag/craft-rules"><img src="https://agentmods.dev/badge/skills/v-songbird/slag/craft-rules.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 5.1 | $0.00170 | $0.03447 |
| Opus 5 | $0.00085 | $0.01724 |
| Sonnet 5 | $0.00034 | $0.00689 |
| Haiku 4.5 | $0.00017 | $0.00345 |
Grade A, and why
craft-rules 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 8d 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 — 289 lines — stays where its author put it; the contents beside it link to each section on GitHub.
assay:craft-rules
You write one rule per run. The wording and placement are the product: follow
references/recipe.md exactly — it encodes the same
factors /assay:opus5 grades. $ARGUMENTS, if present, describes what the
rule should enforce.
You never write a policy file yourself. Every write goes through the engine's
change transaction in step 5 — plan, apply, validate — so the rule is
previewed before it lands and reversible after.
1. Read the project
Before the interview, from the project root:
node "${CLAUDE_PLUGIN_ROOT}/scripts/assay.js" scan
If node is not on PATH (fnm/nvm setups), register it the way the project's
CLAUDE.md says to, then rerun.
Then read .assay-tmp/scan.json. It is the only place you learn where a rule can
go; never assume a filename:
profile.targets.rule.places[]— every file a new rule may be written into, each with itsscopeand ascopingsentence saying what it reaches. This is your placement menu in step 4.sources[]andfiles[]— what already exists, withselected. A source withselected: falseis shadowed and never read; writing into it would be writing nowhere.rules[]— the ACTIVE corpus: every rule loaded here, with its exacttext,file, andlineStart. Step 3 checks your draft against this list.
rules[] carries no scope — join a rule's file to files[] to learn whether
it is project, user, or auto-memory.
2. Grill
Collect, from the conversation or at most two AskUserQuestion rounds. Don't
ask for what the user already said; only fill real gaps.
Ask in this order, and say in one short clause WHY each answer matters — a question with no stated reason reads as a form to fill in. The order is by what a wrong answer costs: 1-3 decide where the rule lives and whether it should be a rule at all; 4 and 5 decide whether there is a rule to write at all — with neither, you refuse below; 6, 7 and 8 only sharpen it.
- Scope — the whole project, or bound to file types, paths, or one subdirectory? This picks the file it goes in, and a rule in the wrong file never loads for the work it is about.
- How much it matters — a preference, something you want followed, or something that must never happen? A preference held to a mandate's standard reads as broken; a mandate written as a preference gets skipped.
- Could a command check it instead? If an exit code can settle it, a hook does it every time and a rule only asks.
- The firing moment — which action, file, or phase should make Claude notice this applies? A rule with no moment is one Claude has to remember unprompted.
- The behaviour — what must happen or stop happening, ideally with one example of it going wrong. The example is what turns a wish into something checkable.
- For a prohibition, what replaces it — and if nothing does, what to do instead of stopping. A ban with no alternative can stall a run outright.
- The specifics — exact paths, identifiers, thresholds. These are what make "done properly" mean one thing.
- One line of what right looks like — the command, the call, the sentence they would want back. An example carries what a description of the example cannot; it is also the concrete specific question 7 is asking for. Optional: this lever is unmeasured, so never hold a rule back for it.
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.
- 8d ago First seen · 289 lines · 170 tokens per session scan A a1aa34d7e035
craft-rules is a skill published in the GitHub repository V-Songbird/slag (1 stars, last pushed 3d ago), licensed MIT. It adds 170 tokens to every session and 3,447 once invoked, about $0.0009 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 skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…
chronicle
Analyze Copilot session history for standup reports, usage tips, session search, and session reindexing. Use when the user asks for a standup, daily summary, usage tips, workflow recommendations, wants to search or find past sessions by keyword/file/PR, wants to reindex their session store, or asks about deleting…