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 skills add retz8/pastiche --skill grill-to-specgit clone --depth 1 https://github.com/retz8/pasticheWrote 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/retz8/pastiche/grill-to-spec)<a href="https://agentmods.dev/skills/retz8/pastiche/grill-to-spec"><img src="https://agentmods.dev/badge/skills/retz8/pastiche/grill-to-spec/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/retz8/pastiche/grill-to-spec"><img src="https://agentmods.dev/badge/skills/retz8/pastiche/grill-to-spec.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00061 | $0.00662 |
| Opus 5 | $0.00030 | $0.00331 |
| Sonnet 5 | $0.00012 | $0.00132 |
| Haiku 4.5 | $0.00006 | $0.00066 |
Grade A, and why
grill-to-spec 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 10d 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 — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Grill to Spec
Convert a just-completed grill-me session into a spec file under _dev/docs/spec/.
File naming
Number first. Filename comes from the TODO number when available, otherwise name only. Do not ask the user about the path — derive it and proceed.
- Task (
**N.M**in TODO):_dev/docs/spec/task-<N.M>-<short-name>.md - Phase:
_dev/docs/spec/phase-<N>-<short-name>.md - No task number available:
_dev/docs/spec/<short-name>.md
Hard rules
- Spec, not plan. Capture what was decided, not how to implement it. No task breakdown, no step-by-step, no file-by-file instructions.
- Only what was locked in the grill. If a decision did not come up explicitly in the grilling conversation, do not include it. Do not infer, do not extend, do not add "for completeness."
- No invented decisions. If the user did not state a preference on something, it is not in the spec. Surface it as an open item only if the user flagged it as open during grilling.
- No implementation details. No code, no command lines, no file syntax, no per-step procedure. The spec is a high-level overview of the locked task scope and decisions.
Required structure
# <Task title>
<One- or two-sentence statement of what this spec covers. Reference the parent TODO item, phase, or OSS_SPEC section.>
## Scope
<Bulleted or short-prose statement of what is in scope for this task. Drawn from the grill, not invented.>
## Locked decisions
### 1. <Decision title>
<One short paragraph stating the decision as locked in the grill.>
### 2. <Decision title>
<...>
## Invariants
<Optional. Constraints the grill surfaced that apply throughout the task. Skip if none were raised.>
## Open items
<Optional. Only items the user explicitly flagged as deferred or unresolved during the grill. Skip if none.>
Process
- Re-read the grilling conversation in context. Identify every decision the user explicitly accepted, rejected, or modified.
- Derive the target path per the File naming rules above. Do not ask, do not confirm.
- Write the spec directly to that path using the required structure. Do not display the draft to the user beforehand — just write it.
- Report the written path in one line. The user will redirect if revisions are needed.
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.
- 10d ago First seen · 69 lines · 61 tokens per session scan A 4da842a0a8c8
grill-to-spec is a skill published in the GitHub repository retz8/pastiche (4 stars, last pushed 3mo ago), licensed MIT. It adds 61 tokens to every session and 662 once invoked, about $0.0003 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
design-system
Token architecture and component specs: three-layer tokens (primitive→semantic→component), CSS variables, spacing/typography scales, slide generation. Use for design tokens, systematic design, brand-compliant presentations.
codex
Run the Codex CLI directly in the user's checkout for code analysis, refactoring, or automated editing without Claude Architect's verified delegation lifecycle.
commit
A guided process for committing code changes to Git. It checks the working tree, follows a conventional commit format, links a task when available, reviews the changes, and asks for approval.
orient
Internal skill for commands. Render the shared orientation snapshot - active task, what's in flight, what's next - tracker-universal. Do not trigger on user conversation; only when a command needs the snapshot.
x-multi-llm-align
A review process in which two sub-agents examine an API, data format, event schema, or workflow from their separate implementation perspectives. The user passes documents and feedback between them over multiple rounds.
x-cr
A software-correctness investigation skill for finding why code behaves differently from what was expected. It uses evidence from code paths, specifications, tests, logs, and changes to assess possible causes.