security

A set of rules and tools for handling secrets, local data, filesystem paths, and public-repository safety. It defines how credentials are stored in the operating system's keychain and what must not be placed in code, logs, databases, or a public repository.

In plain words
What is it for?
Use it when reading or storing credentials, handling files and paths, working with local memory data, writing repository examples, or reviewing code for public-release safety.
Why use it?
It reduces the risk of leaking API keys or personal information, sending local data elsewhere, using unsafe paths, or overwriting user changes without a backup.

Agent

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 agents/gingugu/gingugu/security
Clone the repo
git clone --depth 1 https://github.com/gingugu/gingugu
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 411 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 $0.00000 $0.00411
Opus 5 $0.00000 $0.00205
Sonnet 5 $0.00000 $0.00082
Haiku 4.5 $0.00000 $0.00041

Measured yesterday against content hash c81087202818, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

security 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 yesterday.

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.

.ai/agents/security.md · 36 lines

What it actually says

Agent: Security

Scope

Secrets handling, the credential vault, data locality, and the fact that this is a public repository.

Rules

  • Never commit secrets — credentials, API keys, tokens. The .gitignore blocks common token files; do not override it.
  • Credential vault = OS keychain only. credentials.py reads/writes secret values in the keychain. Secret values never touch the SQLite DB, log files, chat, or the repo. credential_list returns only non-secret metadata.
  • Local data stays local. The memory DB lives in the platform data dir, never in the repo. No telemetry, no cloud calls, no phone-home.
  • Public repo hygiene. Never embed internal URLs, customer names, employer details, or PII in code, docs, examples, .ai/, or test fixtures. This repo is open source.
  • Path safety. Validate/normalize any filesystem path derived from user content — no traversal.
  • Never destroy a user's bytes without a copy. Any code path that overwrites a file in the user's repo or home directory writes <name>.bak first whenever the content would change. Ownership is not the test: a file gingugu wrote is still full of edits the user made afterwards, and keying the backup off "did we write this?" is how init --force came to destroy customized hooks silently for three releases. Destructive flags earn a net, not a shortcut.
  • External writes are approval-gated. GitHub API, PyPI, npm publishes: present the exact command + blast radius, wait for explicit approval.

Definition of done

  • No secret or private detail in the diff (grep before commit).
  • Vault changes keep secret values out of the DB and logs.
  • Any publish/mutation was explicitly approved.
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. yesterday First seen · 36 lines · 0 tokens per session scan A c81087202818

Subscribe to this mod's changes

security is an agent published in the GitHub repository gingugu/gingugu (4 stars, last pushed yesterday), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 411 tokens. 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.