halt-check

halt-check is a command for Claude Code from bookedsolidtech/helixir. It costs 20 tokens per session (961 once invoked), scanned A, original, MIT.

A read-only smoke test for a HALT kill switch, which is a file-based control that should pause governed automation. It checks hooks, middleware, and tool-call restrictions across the system.

In plain words
What is it for?
Use it after changing hooks, middleware, or Claude settings to verify HALT behavior end to end.
Why use it?
It helps detect parts of the automation that continue running when the emergency stop is enabled. This provides a release-time check for incomplete enforcement.

Command for Claude Code

Install

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.

agentmods
npx agentmods add commands/bookedsolidtech/helixir/halt-check
Clone the repo
git clone --depth 1 https://github.com/bookedsolidtech/helixir

Made for: Claude Code.

Wrote 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.

agentmods badge for halt-check

README.md
[![agentmods](https://agentmods.dev/badge/commands/bookedsolidtech/helixir/halt-check.svg)](https://agentmods.dev/commands/bookedsolidtech/helixir/halt-check)
Your own site
<a href="https://agentmods.dev/commands/bookedsolidtech/helixir/halt-check"><img src="https://agentmods.dev/badge/commands/bookedsolidtech/helixir/halt-check.svg" alt="Measured on agentmods" height="20"></a>
Per session 20 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 961 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00020 $0.00961
Opus 5 $0.00010 $0.00481
Sonnet 5 $0.00004 $0.00192
Haiku 4.5 $0.00002 $0.00096

Measured 4d ago against content hash 7a3b25ed5f19, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

halt-check 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 4d 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.

.claude/commands/halt-check.md · 121 lines

How it starts

The opening of the file, as written. The whole thing — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/halt-check — Kill Switch Smoke Test

Verifies that the REA HALT kill switch is actually enforced end-to-end: hooks check for .rea/HALT at their top, middleware short-circuits on kill-switch, and governed tool calls are denied when the file is present. Run this after any change to hooks, middleware, or .claude/settings.json. Run it at least once per release.

This command is advisory and read-only. It writes a temporary HALT file, observes behavior, and removes it — but the observation is all through safe, declarative checks. It never issues destructive tool calls to probe the system.

Preflight

  1. Read .rea/policy.yaml — confirm REA is installed
  2. Check if .rea/HALT already exists. If it does, report "Session already frozen — cannot run smoke test without disturbing existing HALT. Unfreeze first." and stop.
  3. Verify the user has permission — this is an L1+ operation because it writes .rea/HALT (briefly).

Step 1 — Baseline check

Before writing HALT, capture the baseline:

npx rea check --json

Record:

  • halt_present: false (must be)
  • autonomy_level
  • profile
  • hook_count from .claude/settings.json
  • middleware_enabled list

If the baseline reports halt_present: true, stop — the smoke test requires a clean starting state.

Step 2 — Write the test HALT

printf '{"reason":"halt-check smoke test","timestamp":"%s","invoker":"halt-check"}\n' "$(date -u +%FT%TZ)" > .rea/HALT

Confirm the file exists and is readable:

ls -la .rea/HALT

Step 3 — Verify denial paths

Run the diagnostic CLI which simulates representative tool calls through the middleware without actually executing them:

npx rea check --simulate

Expected output: each simulated call returns denied: kill-switch. Record the results.

Then inspect the hook library — every hook should source hooks/_lib/halt-check.sh:

for hook in .claude/hooks/*.sh; do
  if ! grep -q 'halt-check.sh' "$hook"; then
    echo "MISSING HALT CHECK: $hook"
  fi
done

Read the full file on GitHub · 121 lines

Changes

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.

  1. 4d ago First seen · 121 lines · 20 tokens per session scan A 7a3b25ed5f19

Subscribe to this mod's changes

halt-check is a command published in the GitHub repository bookedsolidtech/helixir (5 stars, last pushed 2mo ago), licensed MIT. It adds 20 tokens to every session and 961 once invoked, about $0.0001 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.