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.
git clone --depth 1 https://github.com/dwarvesf/dwarves-kitWrote 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/commands/dwarvesf/dwarves-kit/test-write)<a href="https://agentmods.dev/commands/dwarvesf/dwarves-kit/test-write"><img src="https://agentmods.dev/badge/commands/dwarvesf/dwarves-kit/test-write/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/commands/dwarvesf/dwarves-kit/test-write"><img src="https://agentmods.dev/badge/commands/dwarvesf/dwarves-kit/test-write.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 | $0.00047 | $0.01342 |
| Opus 5.5 | $0.00019 | $0.00537 |
| Sonnet 5.5 | $0.00009 | $0.00268 |
| Haiku 4.5 | $0.00005 | $0.00134 |
Grade A, and why
test-write 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 1mo 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a test-write dispatcher. Your job is to take a spec's ALREADY-REVIEWED ## Test plan (post /kit:test-plan-review-team, verdict SOLID) and turn it into real, executing test code by dispatching kit:test-writer. You do no test-design work yourself , resolution and gating only, same coordinator shape as test-harden.md.
Process
Bracket the phase for timing before starting: bash lib/gate/gate-ledger.sh outcome <rid> test-write start.
Step 1: Find the active spec
Detect the active docs/specs/SPEC-NNN-<slug>.md the same branch-aware way /kit:test-plan uses. If several specs match, ask the user which one, do not auto-pick.
Step 2: Require a fresh SOLID verdict
Read the spec's ## Test plan and ## Test plan critique sections. All of the following must hold, or you stop (see Step 3):
- Present. The spec has both a
## Test planand a## Test plan critiquesection. Missing either -> stop. - Exactly one match.
grep -c "^## Test plan critique"on the spec is exactly1. Zero means no critique ever ran; more than one means the replace-not-stack rule (test-plan-review-team.mdStep 5) was violated somewhere and the file is not trustworthy , either way, stop rather than guess which copy is current. - Verdict is SOLID. The critique's
### Verdict:line reads exactlySOLID, notREVISEorRECONSIDER, and there is exactly one### Verdict:line (more than one is the same untrustworthy-file case as check 2). - Fresh, not stale. Compare the
Date:line under## Test planagainst theDate:line under## Test plan critique. The critique's date must be on or after the plan's date. If the plan's date is later, the plan was rewritten after the critique ran (e.g. a re-run of/kit:test-plan) and the SOLID verdict no longer describes the matrix in front of you , treat as stale, same as missing.
This is the same freshness/exactly-one-match discipline /kit:test-harden codifies for its own verdict read, reused inline here rather than re-derived, since test-harden.md had not landed in this repo as of this writing.
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.
- 1mo ago Changed 7aa39a34feaa
- 1mo ago First seen · 61 lines · 47 tokens per session scan A e25ecbfc8a66
test-write is a command published in the GitHub repository dwarvesf/dwarves-kit (12 stars, last pushed yesterday), licensed MIT. It adds 47 tokens to every session and 1,342 once invoked, about $0.0002 per session on Opus 5.5. 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 2026-09-04.
Other commands, from other repositories
spec-impl
Execute spec tasks using TDD methodology.
cloudflare-durable-objects:setup
Interactive Durable Objects project initialization. Scaffolds new DO project with proper bindings, migrations, and boilerplate code.
api-aqa-flow
Workflow for backend API test automation: TMS / Issue Tracker test cases → automated API tests, HITL-gated.
speckit.companion.living-validate
Check the shape of living specs and a feature spec's deltas — a requirement with no scenario, a scenario missing WHEN or THEN, a duplicate heading, a delta pointing at nothing (opt-in, read-only, never halts).
test-driven-development
Use when implementing any feature or bugfix, before writing implementation code - write the test first, watch it fail, write minimal code to pass; ensures tests actually verify behavior by requiring failure first.
verification-before-completion
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always.