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 commands/aziontech/webkit/spec-creategit clone --depth 1 https://github.com/aziontech/webkitWhat 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 | $0.00044 | $0.01330 |
| Opus 5 | $0.00022 | $0.00665 |
| Sonnet 5 | $0.00009 | $0.00266 |
| Haiku 4.5 | $0.00004 | $0.00133 |
Grade A, and why
spec-create 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 yesterday.
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 — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are creating a new component spec for .specs/<name>.md. Follow the spec-create skill in .claude/skills/spec-create/SKILL.md exactly.
User input: $ARGUMENTS
Two modes — pick one up front
spec-create has two distinct paths. Detect the mode from the arguments (and confirm with the user before drafting):
- Mode A — Figma-driven (
--figma <url>present). The Figma frame is the source of truth: extract tokens/regions/states, map them to DESIGN.md, and ask the user only where the frame is silent (behavior, a11y intent, story list). The component name is derived from the frame's component name (normalized to kebab, confirmed with the user); passing<name>is optional here and, when given, overrides the frame. - Mode B — Name + Category (no
--figma; the user is building without a design). There is no ground truth, so invention risk is highest. This mode is deliberately more blocking: it always asks about every section, one at a time, never infers props/events/tokens, and refuses to write until each mandatory section has a real answer (no silentTBD).--categoryis required here (ask + confirm if missing).
If --figma is absent, you are in Mode B — do not quietly infer from intent. If both a URL and "no figma" intent are ambiguous, ask which mode the user wants before proceeding.
What to do
- Read the skill and the foundations:
.claude/skills/spec-create/SKILL.md— full workflow..specs/_template.md— copy this body structure..specs/_schema.json— frontmatter contract..claude/rules/no-invention.md,naming.md,tokens.md,accessibility.md,bem-testid.md..claude/docs/DESIGN.md— design tokens (authoritative).
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.
- yesterday First seen · 47 lines · 0 tokens per session scan A 4b7691a29192
spec-create is a command published in the GitHub repository aziontech/webkit (2 stars, last pushed 3d ago), licensed MIT. It adds 44 tokens to every session and 1,330 once invoked, about $0.0002 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 commands, from other repositories
speckit.tasks
Generate an actionable, dependency-ordered tasks.md for the feature based on available design artifacts.
diagnose
Run sivtr doctor to check the environment, then investigate any failing checks.
test-changed
Run the full quality gate on only the files changed since branching from main.
route
Route notes from inbox to appropriate vault destinations.
cleanup-transcript
Clean up a video/audio transcript (SRT or plaintext) - identifies structure, fixes transcription errors, asks clarifying questions iteratively.
reflect
Capture learnings from the current conversation through guided reflection.