A verification check that runs in the main coding session while work is happening. It checks tests, sensitive files, adherence to working rules, and newly discovered traps, then warns or blocks when needed.

Use it before committing changes, to run tests, block sensitive-file problems, check required working practices, and flag unexpected risks during implementation.

It catches common operational mistakes before they become part of the work. It focuses on immediate checks rather than recording findings or judging whether the final result matches the user's intention.

guard — Skill for Claude Code by jun0-ds/sonmat

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.

Any agent

npx skills add jun0-ds/sonmat --skill guard

Clone the repo

git clone --depth 1 https://github.com/jun0-ds/sonmat

Made for: Claude Code.

Or install sonmat, the plugin that ships this one along with the rest of its 6 skills, 3 agents, 1 hook.

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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/jun0-ds/sonmat/guard/github.svg)](https://agentmods.dev/skills/jun0-ds/sonmat/guard)

Your own site

<a href="https://agentmods.dev/skills/jun0-ds/sonmat/guard"><img src="https://agentmods.dev/badge/skills/jun0-ds/sonmat/guard/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/jun0-ds/sonmat/guard"><img src="https://agentmods.dev/badge/skills/jun0-ds/sonmat/guard.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 48 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,242 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00048	$0.01242
Opus 5.5	$0.00019	$0.00497
Sonnet 5.5	$0.00010	$0.00248
Haiku 4.5	$0.00005	$0.00124

Measured today against content hash e05c7da04885, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

guard 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 today.

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.

skills/guard/SKILL.md · 92 lines

How it starts

The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Guard — Main-Side Verification Checks

Automatic verification layer running in the main session (System 1). No agent spawn.

Guard covers verification checks that main can reliably do on itself: sensitive files, test execution, discipline conformance, and detection of novel traps during work. Guard is pure verify-and-flag — it detects, it warns, it blocks. It does not record, accumulate, or persist findings; those belong to scribe (post-work persistence). It also does not verify whether the artifact matches the user's intent; that check is structurally unreliable when done inside the agent that produced the artifact, and is delegated to witness.

Scope boundaries:

Check kind	Owner	Why
Test execution, sensitive file blocking, discipline conformance, novel-trap detection	guard	Real-time verification during work; synchronous check-and-block at decision points
Scope match (is this within what was asked?), content match (does it do what was asked?), framing-derived scope	witness	Intent-artifact comparison — requires protocol isolation from main's reasoning (see witness.md §Isolation stack for what "isolation" means on current Claude Code)
Recording novel traps, writing project rules to AGENTS.md, journaling verdicts, bridge notes, progress tracking	scribe	Post-work persistence — what the session learned that should outlive it

Guard detects and flags. Scribe persists what was flagged. Witness checks intent-artifact match in isolation. Three different axes on the same work.

1. Pre-Commit Verification

Before any commit, check in order:

Test pass confirmation

If tests exist, run them before commit. Fail → block commit.
"No tests, so just commit" is not allowed.

Claims backed by execution

"It works" must be backed by actual output (logs, test results). Not by speculation.

Sensitive file blocking

.env, credentials.*, secrets.*, *.pem, *.key in staging → immediate block.
Also check .gitignore coverage.

Read the full file on GitHub · 92 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.

today Changed e05c7da04885
1mo ago Changed a8f06b7a6f14
2mo ago First seen · 92 lines · 48 tokens per session scan A fb6f1108da69

Subscribe to this mod's changes

guard is a skill published in the GitHub repository jun0-ds/sonmat (5 stars, last pushed yesterday), licensed BSD-3-Clause. It adds 48 tokens to every session and 1,242 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 2026-08-31.

Claude Code Security Testing ai-tools ai-verification anthropic autonomous-agent claude

guard next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
guardjun0-ds/sonmat	Skill	5	yesterday	BSD-3-Clause	A	48	Claude Code
huedominikmartn/hue	Skill	900	yesterday	MIT	A	84	Claude Code, Codex
example-feature-generatorqf-studio/navigator	Skill	356	2d ago	MIT	A	33	Claude Code
authentication-patternstravisjneuman/.claude	Skill	108	yesterday	MIT	A	28	Claude Code
electron-desktoptravisjneuman/.claude	Skill	108	yesterday	MIT	A	38	Claude Code
content-repurposertravisjneuman/.claude	Skill	108	yesterday	MIT	A	43	Claude Code
impact-report-writertravisjneuman/.claude	Skill	108	yesterday	MIT	A	43	Claude Code