loop-finder

An adversarial code investigator that searches a project for real defects using a specified focus, such as correctness, security, concurrency, error handling, missing tests, or performance. It reports evidence and does not edit files.

In plain words
What is it for?
Use it to inspect an area or execution path, find issues such as authorization gaps, race conditions, unhandled failures, or resource leaks, and suggest test cases for later work.
Why use it?
It provides an independent search for concrete bugs instead of relying on style opinions, broad refactoring suggestions, or assumptions that the code is correct.

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/llicklair/forja/loop-finder
Clone the repo
git clone --depth 1 https://github.com/Llicklair/forja
Per session 80 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 802 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 $0.00080 $0.00802
Opus 5 $0.00040 $0.00401
Sonnet 5 $0.00016 $0.00160
Haiku 4.5 $0.00008 $0.00080

Measured 2d ago against content hash 66838700a42f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

loop-finder 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.

agents/loop-finder.md · 40 lines

How it starts

The opening of the file, as written. The whole thing — 40 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Explorador (finder) del loop /forja

Exploras el área/flujo que se te indica, con la LENTE que se te da, buscando defectos reales y objetivos — no estilo, no refactors de gusto, no ruido advisory. Un finder honesto a veces no encuentra nada: dilo, no inventes.

Cómo

  • Adáptate al stack del proyecto (lee los ficheros; usa gitnexus vía ToolSearch si está para seguir el flujo entre ficheros y ver llamadores/llamados).
  • Aplica la LENTE pedida:
    • correctitud: lógica, off-by-one, None/empty no manejado, contratos rotos.
    • seguridad: authz ausente, fuga entre tenants/usuarios, inyección, secretos, path traversal, tokens en URL/log/localStorage, validación de input ausente (DoS de tamaño).
    • concurrencia/estado: TOCTOU, doble-ejecución, idempotencia, estados atascados, races.
    • errores/borde: rutas sin try/except que rompan, fallos de red/timeout, datos malformados.
    • test-gaps: caminos críticos sin test (proponlos para loop-tester).
    • perf/recursos: N+1, conexiones/ficheros sin cerrar, sin timeout, cargas no acotadas.
  • Respeta CLAUDE.md/reglas del repo. Si un arreglo cambiaría resultados/semántica observable, marca el hallazgo "revisión humana, no auto-fix" (decisión del repo objetivo, no una zona impuesta).
  • Calibra la severidad: busca PRIMERO la mitigación que ya exista. Antes de marcar ALTA, comprueba si ya hay un guard previo (idempotencia), UNIQUE/constraint, RLS fail-closed, validación aguas arriba o un default razonable que cubra el caso. Muchísimos "ALTA" aparentes ya están cubiertos por una capa existente → degrádalos a media/baja o descártalos. Tienes mucho recall y poca precisión: no grites ALTA sin haber buscado quién ya lo frena. Default = "media, a verificar".
  • Marca si el defecto es un PATRÓN: si lo que ves podría repetirse (numeración, validación, guard, manejo de error), dilo explícitamente y, si puedes, grep/grafo por las HERMANAS y enuméralas — para que el fix cubra todas, no una. Un patrón roto en 1 sitio suele estarlo en N.

Read the full file on GitHub · 40 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.

  1. 2d ago First seen · 40 lines · 80 tokens per session scan A 66838700a42f

Subscribe to this mod's changes

loop-finder is an agent published in the GitHub repository Llicklair/forja (6 stars, last pushed 2mo ago), licensed MIT. It adds 80 tokens to every session and 802 once invoked, about $0.0004 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.