Borrowing it
Nothing to install: this file belongs to ractive/hyalo. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/ractive/hyalo/main/.claude/skills/dogfood/SKILL.mdgit clone --depth 1 https://github.com/ractive/hyaloWrote 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/ractive/hyalo/dogfood)<a href="https://agentmods.dev/skills/ractive/hyalo/dogfood"><img src="https://agentmods.dev/badge/skills/ractive/hyalo/dogfood/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/ractive/hyalo/dogfood"><img src="https://agentmods.dev/badge/skills/ractive/hyalo/dogfood.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00102 | $0.01956 |
| Opus 5 | $0.00051 | $0.00978 |
| Sonnet 5 | $0.00020 | $0.00391 |
| Haiku 4.5 | $0.00010 | $0.00196 |
Grade A, and why
dogfood scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Dogfood Session
A dogfood session exercises hyalo against real knowledgebases to find bugs, verify fixes, assess UX, and discover improvement opportunities. Sessions should feel like a real user working with the tool — not a mechanical checklist. Be curious, follow interesting leads, try edge cases that occur to you naturally.
Phase 1: Prepare
- Build and install from source, then prove the
hyaloon PATH is this tree. The PATH binary is a cargo install and goes stale silently: on 2026-09-04 it was three days and eleven PRs behindtarget/release/hyalo, and a DEC-284 check against it nearly got filed as a bug.cargo build --release cargo install --path crates/hyalo-cli --target-dir target --force test "$(hyalo --version | awk '{print $3}' | tr -d '(' | sed 's/+dirty$//')" = "$(git rev-parse --short=12 HEAD)" \ && echo "hyalo on PATH is HEAD" || { echo "STALE hyalo on PATH: $(which hyalo)"; exit 1; }--target-dir targetreuses the release build so the install takes seconds; the version string ishyalo X.Y.Z (<12-char hash>[+dirty] <date>), and+dirtyis fine — it only means uncommitted edits such as the report you are about to write. If the check fails, fix the PATH before doing anything else — every later command in this skill, and every explorer agent you spawn, must run the same binary. Never fall back to./target/release/hyalofor some commands andhyalofor others in one session. - Find the last dogfood report to establish the baseline:
Read it to learn: which version was tested, which bugs were found, what was verified.hyalo find --tag dogfooding --sort date --reverse --limit 1 --format text - Discover what changed since then: Use
git log --onelinefrom that report's date to HEAD. Look for merged iteration branches (iter-N/...). For each, read the iteration file in the knowledgebase to understand what was implemented:
Thenhyalo find --tag iteration --property "date>=YYYY-MM-DD" --sort date --format texthyalo read <iteration-file> --format textfor each to understand the features, bug fixes, and UX changes. These are your priority test targets. - Check prior dogfood reports for open bugs to re-verify:
Read the most recent reports and collect any bugs marked as found (not yet verified fixed).hyalo find --tag dogfooding --sort date --reverse --limit 5 --format text - Pick target knowledgebases: Use a mix of:
- Own KB (
hyalo-knowledgebase/, ~250 files) — well-structured, has schemas/views - MDN Web Docs (
../mdn/files/en-us/, ~14K files) — large, stress-tests performance - GitHub Docs (
../docs/content/, ~3.5K files) — complex nested YAML frontmatter - VS Code Docs (
../vscode/, ~1K files) — mid-size, different structure
- Own KB (
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 Changed · +14 lines fb2503d2a60e
- 11d ago First seen · 167 lines · 102 tokens per session scan A 1af4f0d5e48a
dogfood is a skill published in the GitHub repository ractive/hyalo (24 stars, last pushed yesterday), licensed MIT. It adds 102 tokens to every session and 1,956 once invoked, about $0.0005 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 skills, from other repositories
dd-triage-flaky-test
Load when investigating a specific flaky test. Gets history, failure pattern, and category, then recommends fix, quarantine, or escalate.
dd-unblock-pr
Load when investigating a failing PR CI pipeline or checking PR health. Attributes each CI failure as flaky, infra, or regression, proposes a targeted action, and reports code coverage.
agent-inspect
Local evidence debugger and trajectory-test toolkit for TypeScript AI agents. Use when capturing framework-faithful traces, asserting TraceContract/TraceFacts, packaging Evidence v2, or inspecting local runs over read-only MCP (gettracefacts).
confirm-failures-are-causally-linked-to-the-task-before-reportin
When delegating a task affected by this skill, include.
isolate-the-specific-failure-mode-before-proposing-a-fix
When delegating a task affected by this skill, include.
qa-sweep
Codex adapter for independent QA review of opencode-swarm changes. Use after implementation or debugging when the user asks for a QA sweep, regression check, review pass, quality gate, or confidence check before publishing.