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.
npx agentmods add agents/liatrio-labs/claude-code-gauntlet/claudegit clone --depth 1 https://github.com/liatrio-labs/claude-code-gauntletWrote 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/agents/liatrio-labs/claude-code-gauntlet/claude)<a href="https://agentmods.dev/agents/liatrio-labs/claude-code-gauntlet/claude"><img src="https://agentmods.dev/badge/agents/liatrio-labs/claude-code-gauntlet/claude.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.00881 |
| Opus 5 | $0.00000 | $0.00441 |
| Sonnet 5 | $0.00000 | $0.00176 |
| Haiku 4.5 | $0.00000 | $0.00088 |
Grade A, and why
CLAUDE 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.
This is a copy
86% identical to AGENTS — 4 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 45 lines — stays where its author put it; the contents beside it link to each section on GitHub.
agents/
Subagent contracts. Each .md is a system prompt with enforced YAML frontmatter.
- Frontmatter is system-enforced.
tools,effort,modelandcolorare not advisory — Claude Code enforces them. - LSP-first investigation. Agents prefer
goToDefinition/findReferences/hover, with Grep as fallback. - A finding field is one entry in
workflows/src/registry.jsplus the owning agent's output block. A field no schema declares is rejected at dispatch — the item schema is closed (additionalProperties: false, issue #53) — so emitting it burns schema retries and, persisted, fails the agent.tests/test_dimensions_registry.pyfails the build when the two drift. - Declared-but-optional fields are not nullable. A not-applicable value is omitted, never
null; the platform types each property to a single type and a null burns retries. requiredisFINDING_REQUIREDplus per-rowrequiredExtra. Required only when the contract emits a field unconditionally (no OMIT branch); a multi-dimension agent requires it only when every sibling dimension does too. A field required only in its OWN dimension (siblings OMIT it —claude_md_rule,spec_text) goes inrequiredWhenDimensioninstead, schema-enforced there on a first-party-direct dispatch only; contract prose is the floor elsewhere.dimension— short name from agent output:"bug","security","cross_file_impact","test_coverage","convention","intent","comment_accuracy","type_design","simplification". Never the agent name.agentis injected by the orchestrator at merge; agents do not emit it.- Canonical fields — every dispatch schema declares exactly these:
id,file,line_start,line_end,title,description,severity,confidence,dimension,origin,evidence,suggestion,claude_md_rule,cross_file_refs,suggested_fix_code. - Per-dimension extras — one entry on the owning registry row:
hidden_errors(bug),attack_vector(security),affected_consumers(cross_file_impact),criticality+failure_scenario(test_coverage),spec_text(intent),invalid_state_example(type_design),behavior_preserved(simplification). - The false-positive exclusion list and the complete-read contract are intentionally duplicated
into every agent that needs them, so the guarantee survives a failed file read. Each copy carries
a
<!-- Canonical source: ... -->pointer. Do not refactor them into a shared read.tests/test_agent_contracts.py::TestCompleteReadContractasserts the complete-read copies are byte-identical;TestPromptInjectionArtifactsMirrorasserts the 7 exclusion-list copies are byte-identical to each other (not to the canonical skills file, which deliberately uses a different person/voice). - The challenger is structurally blind by design — title, description, location, and the code it opens itself. It is never given the shared context path.
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 First seen · 45 lines · 0 tokens per session scan A 6c73078cf906
CLAUDE is an agent published in the GitHub repository liatrio-labs/claude-code-gauntlet (12 stars, last pushed today), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 881 tokens. A static security scan graded it A with 0 findings. It is 86% identical to AGENTS, differing in 4 lines, and is treated as a copy.
Other agents, from other repositories
repo-evaluator
Use this agent when evaluating repository health before contributing, analyzing maintainer responsiveness, or deciding if a repo is worth investing time in. Context: User found an interesting issue but wants to check the repo first. user: "Is this repository worth contributing to?" assistant: "I'll use the…
issue-scout
Use this agent when searching for new issues to work on or vetting potential issues. This agent finds and evaluates good contribution opportunities. Context: User wants to find issues to contribute to. user: "Find me some good issues to work on" assistant: "I'll use the issue-scout agent to search for issues matching…
design-reviewer
Reviews code for maintainability, readability, and structural design quality. Read-only.
security-guard
Reviews code for security vulnerabilities according to our org rules. Read-only.
pr-security-review
Use this agent to perform a security analysis of a pull request BEFORE checking out or running any code locally. Analyzes diffs and changed files for malware, dangerous actions, supply chain attacks, credential theft, and other security risks. Returns a verdict of SAFE, REVIEW NEEDED, or BLOCK. Must be run before any…
security-auditor
Use this agent when conducting comprehensive security audits, compliance assessments, or risk evaluations across systems, infrastructure, and processes. Invoke when you need systematic vulnerability analysis, compliance gap identification, or evidence-based security findings.