Borrowing it
Nothing to install: this file belongs to incubateur-ademe/benefriches. 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/incubateur-ademe/benefriches/main/.claude/commands/adw-build.mdgit clone --depth 1 https://github.com/incubateur-ademe/benefrichesWrote 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/incubateur-ademe/benefriches/adw-build)<a href="https://agentmods.dev/commands/incubateur-ademe/benefriches/adw-build"><img src="https://agentmods.dev/badge/commands/incubateur-ademe/benefriches/adw-build/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/incubateur-ademe/benefriches/adw-build"><img src="https://agentmods.dev/badge/commands/incubateur-ademe/benefriches/adw-build.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 | $0.00020 | $0.00547 |
| Opus 5.5 | $0.00008 | $0.00219 |
| Sonnet 5.5 | $0.00004 | $0.00109 |
| Haiku 4.5 | $0.00002 | $0.00055 |
Grade A, and why
adw-build 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 9d 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.
What it actually says
Implement the spec at $ARGUMENTS on the current branch, in the current worktree. If the spec starts with BLOCKED, stop and report it unchanged.
- Read the spec, then the AGENTS.md files and rules for every file listed under Files. The spec says what to build; they say how it is done in this repo.
- Invoke a project skill with the Skill tool before doing the work it covers, every time it applies, even when you think you know how:
- database schema or data change:
create-database-migration - new or changed e2e test:
create-e2e-test, thenrun-e2e-teststo run it - multi-step form flow or step:
wizard-form - React components, hooks or Redux:
react-best-practices - an architectural decision or new pattern:
adr
- database schema or data change:
- Implement the spec. When the spec is wrong or incomplete, prefer what the code and rules require, and note the deviation in your report.
- Run the spec's Validation commands and
pnpm verify. Fix every failure and rerun until all are green. Never weaken, skip or delete a test or a check to make it pass; if a check fails for a reason outside the spec (environment, an unrelated flaky test), keep its full output and report it rather than working around it. - Commit as you go, one small commit per coherent step, with a conventional commit message (the
generate-commit-messageskill writes one when it is available). Leave no uncommitted changes.
The runner may resume this session with the failing output of its acceptance check, or with a JSON list of review blockers. Fix each one as in steps 3 to 5, then end with the same report.
Never push, merge, rebase onto another branch or open a pull request: the branch leaves the machine through the runner and a human.
If a required input is missing (user-facing copy, a design, a business rule, an endpoint) and neither the spec nor the code provides it, do not guess: commit only what is complete and verified, then start your final message with the line BLOCKED followed by one bullet per missing input.
Otherwise end with a short report: commits made (git log --oneline since you started), skills invoked, the result of each Validation command and of pnpm verify, and any deviation from the spec.
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.
- 9d ago First seen · 27 lines · 20 tokens per session scan A 2efbd2662daf
adw-build is a command published in the GitHub repository incubateur-ademe/benefriches (45 stars, last pushed today), licensed MIT. It adds 20 tokens to every session and 547 once invoked, about $0.0001 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 2026-10-21.
Other commands, from other repositories
check
Run project checks and fix any errors without committing.
act
Follow RED-GREEN-REFACTOR cycle approach for test-driven development.
implement
Execute all implementation tasks from tasks.md with test-driven development, parallel batching, and atomic commits.
git-workflow
Run review, tests, and commit steps, then prepare a pull request.
fix-tests-execute
Phase 2 of fixing-tests: Fix Execution - investigate, classify, fix, verify, and commit each work item.
check-fix
Execute quality gates, fix any issues found, and create a single high-quality Conventional Commit summarizing all changes made.