What is Agent Author?

An agent specification that turns one past security incident into a reusable detector for the same coding mistake. SAST means static application security testing: reviewing source code for security problems without running it.

Use it to read a past security report, confirm the project's conventions and helpers, and create an agent that flags similar new code.

It helps prevent a fixed security bug from returning in the same codebase. The detector focuses on the underlying pattern instead of copying the original incident.

Agent Author — Agent by agentgg-dev

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/agentgg-dev/agentgg/create

Clone the repo

git clone --depth 1 https://github.com/agentgg-dev/agentgg

Per session 30 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 1,502 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.

Model	Per session	Once invoked
Fable 5	$0.00030	$0.01502
Opus 5	$0.00015	$0.00751
Sonnet 5	$0.00006	$0.00300
Haiku 4.5	$0.00003	$0.00150

Measured 2d ago against content hash caea04e18721, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

Agent Author 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 2d 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.

packages/cli/src/agents/create.md · 131 lines

How it starts

The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the agent-author. You are NOT looking for new bugs. Your job is to read ONE past security report describing an incident that happened in this codebase, understand why it happened at the code level, and produce a reusable agentgg agent spec that catches the same anti-pattern if it surfaces in this codebase again.

What you are producing (and what you are NOT producing)

You are producing one detector template that future agentgg scan runs will execute over this codebase. It is a generalized pattern detector, not a hunt for the specific past instance.

NOT "find the exact lines from the report." That bug is already fixed.
NOT "find every place where this CWE could theoretically apply" in any codebase. Scope to this codebase's conventions, helpers, and file layout, which you confirm with your tools.
YES "if someone writes new code in this repo with the same shape of mistake (same anti-pattern, same dangerous helper, same missing guard), this agent flags it."

A good agent generalizes the shape of the mistake; a bad agent overfits to the exact file or string from the report.

How to work

You have Read, Glob, and Grep. Your working directory is the repository root. The past report is included verbatim in the prompt below.

Read the report carefully. Identify:
- The vulnerability class (SQLi, SSRF, IDOR, auth bypass, prototype pollution, etc.).
- The unsafe code element (an API, a helper, a flag, a missing check).
- The trust boundary that was crossed.
Find the past bug in the code. Grep / Read the cited files. Confirm how it manifested here: which module, which framework method, what naming convention, what helper functions are involved. If the fix has already landed, read the pre-fix code (the report usually quotes it) and look at the surrounding patterns to see where else the same helper is used or where the same guard is missing.
Generalize. What is the smallest, sharpest pattern that would have caught the past bug AND would catch a re-introduction of the same anti-pattern by a different author in a different file? Examples: "any call to db.rawQuery whose argument is built with + from a request property", "any controller that returns a Tenant without first calling assertTenantMatchesRequest", "any URL fetcher missing the validateInternalUrl wrapper."
Scope tightly. Pick where.extensions and where.preFilter regexes that anchor the shape of the mistake, not literal strings from the past file. Add excludePatterns for tests / fixtures / generated code unless the bug class genuinely lives there.

Read the full file on GitHub · 131 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.

2d ago First seen · 131 lines · 30 tokens per session scan A caea04e18721

Subscribe to this mod's changes

Agent Author is an agent published in the GitHub repository agentgg-dev/agentgg (194 stars, last pushed 3d ago), licensed Apache-2.0. It adds 30 tokens to every session and 1,502 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 2026-08-30.

Code review LLM Security Testing ai-agents appsec cli sast

threat-modeler

Use this agent when the user asks to "create a threat model", "analyze threats", "STRIDE analysis", "what are the threats", "threat modeling", "identify attack vectors", "map attack surface", or needs systematic threat identification with data flow diagrams.

allsmog/vuln-scout · 60 tokens

false-positive-verifier

Use this agent to verify security findings and eliminate false positives. Analyzes code context, data flow paths, and exploitability with structured evidence to determine if a finding is a true positive or false positive.

allsmog/vuln-scout · 46 tokens

poc-developer

Use this agent when the user wants to "write an exploit", "create a PoC", "develop proof of concept", "automate the attack", or needs help creating exploit scripts during Phase 3 of whitebox security review.

allsmog/vuln-scout · 52 tokens

app-mapper

Use this agent when the user asks to "understand the application", "map the codebase", "analyze the architecture", "identify trust boundaries", "map user roles", or needs to build comprehensive application understanding before vulnerability hunting.

allsmog/vuln-scout · 51 tokens

code-reviewer

Use this agent when the user asks to "review code for security", "find vulnerabilities", "security audit", "analyze for security issues", or when exploring a codebase with security concerns.

allsmog/vuln-scout · 43 tokens

attack-researcher

Autonomous attack vector exploration agent that hypothesizes novel attack vectors, tests them against the codebase, and iterates. Use when the standard scan pipeline has completed and you want deeper, creative vulnerability research beyond pattern matching.

allsmog/vuln-scout · 50 tokens