gauntlex:learn

gauntlex:learn is a skill for Claude Code from sanjoy1234/gauntlex. It costs 38 tokens per session (268 once invoked), scanned A, original, MIT.

A learning command for GAUNTLEX, an adversarial code-testing tool. It saves the results of a completed run in ChromaDB, a database for storing and searching information, and updates attack effectiveness scores.

In plain words
What is it for?
Use it after reviewing a completed GAUNTLEX run, with its run ID and a pass or fail outcome. It records the attacks, updates rolling scores, and prints a summary of the results.
Why use it?
It keeps useful results from past tests so later GAUNTLEX runs can track which attack types work well for a particular codebase. It can also flag weak attack categories for prompt updates.

Skill for Claude Code

Written for Claude Code: installed under .claude/.

Part of the gauntlex plugin — 7 skills, 1 MCP server shipped together

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 skills/sanjoy1234/gauntlex/learn
Any agent
npx skills add sanjoy1234/gauntlex --skill learn
Clone the repo
git clone --depth 1 https://github.com/sanjoy1234/gauntlex

Made for: Claude Code.

Or install gauntlex, the plugin that ships this one along with the rest of its 7 skills, 1 MCP server.

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 gauntlex:learn

README.md
[![agentmods](https://agentmods.dev/badge/skills/sanjoy1234/gauntlex/learn.svg)](https://agentmods.dev/skills/sanjoy1234/gauntlex/learn)
Your own site
<a href="https://agentmods.dev/skills/sanjoy1234/gauntlex/learn"><img src="https://agentmods.dev/badge/skills/sanjoy1234/gauntlex/learn.svg" alt="Measured on agentmods" height="20"></a>
Per session 38 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 268 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 5.1 $0.00038 $0.00268
Opus 5 $0.00019 $0.00134
Sonnet 5 $0.00008 $0.00054
Haiku 4.5 $0.00004 $0.00027

Measured 5d ago against content hash 8381aca8494b, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

gauntlex:learn 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.

.claude/skills/gauntlex/learn/SKILL.md · 36 lines

What it actually says

/gauntlex:learn

Ingests a run's results into the Knowledge Forge (ChromaDB) and updates the Adaptive Brain.

Usage

/gauntlex:learn --run-id <run_id> --outcome pass|fail

Execution

gauntlex learn --run-id "$RUN_ID" --outcome "$OUTCOME"

What It Does

  1. Reads .gauntlex/reports/<run_id>.json
  2. Writes each attack into ChromaDB with CWE tag + codebase fingerprint
  3. Updates rolling 30-run EMA effectiveness scores per (CWE, fingerprint-cluster)
  4. Schedules Breaker prompt rewrite via meta_agent if any CWE < 0.3 effectiveness over 30 runs
  5. Prints updated effectiveness summary

Output to User

  • Number of attacks ingested
  • Any CWE categories now flagged for prompt rewrite
  • Updated top-5 effective attack categories for this codebase type
  • "Forge entries: total" — shows the brain is growing
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. 5d ago First seen · 36 lines · 38 tokens per session scan A 8381aca8494b

Subscribe to this mod's changes

gauntlex:learn is a skill published in the GitHub repository sanjoy1234/gauntlex (1 stars, last pushed 1mo ago), licensed MIT. It adds 38 tokens to every session and 268 once invoked, about $0.0002 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.