silent-failure

silent-failure is an agent for coding agents from andyzengmath/soliton. It costs 31 tokens per session (1,931 once invoked), scanned A, original, MIT.

A code-review agent that looks for errors hidden by code that fails quietly, such as ignored exceptions, rejected promises that are not handled, and tests without checks that could pass incorrectly.

In plain words
What is it for?
Use it when reviewing changes involving error handling, promises, optional values, fallback returns, error logging, or test mocks. It must be enabled in the stated Soliton configuration.
Why use it?
It helps find cases where software appears to work because an error was swallowed or a missing value was silently replaced, problems that ordinary tests may miss.

Agent

Part of the soliton plugin — 1 skill, 3 commands, 13 agents 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 agents/andyzengmath/soliton/silent-failure
Clone the repo
git clone --depth 1 https://github.com/andyzengmath/soliton

Or install soliton, the plugin that ships this one along with the rest of its 1 skill, 3 commands, 13 agents.

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 silent-failure

README.md
[![agentmods](https://agentmods.dev/badge/agents/andyzengmath/soliton/silent-failure.svg)](https://agentmods.dev/agents/andyzengmath/soliton/silent-failure)
Your own site
<a href="https://agentmods.dev/agents/andyzengmath/soliton/silent-failure"><img src="https://agentmods.dev/badge/agents/andyzengmath/soliton/silent-failure.svg" alt="Measured on agentmods" height="20"></a>
Per session 31 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,931 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. 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 $0.00031 $0.01931
Opus 5 $0.00015 $0.00966
Sonnet 5 $0.00006 $0.00386
Haiku 4.5 $0.00003 $0.00193

Measured 4d ago against content hash 8ca1af71cf80, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

silent-failure scanned grade A with 1 finding 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 4d 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.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

try { await fetch(url); }
agents/silent-failure.md · 216 lines

How it starts

The opening of the file, as written. The whole thing — 216 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Silent Failure Agent

You are a specialist reviewer for silent error-handling failures — the class of bugs where code "works" in the sense of not throwing, but swallows errors that would indicate real problems. This is one of the highest-signal classes of AI-generated-code defects because LLMs frequently "paper over" runtime errors rather than expose them.

Default: OFF as of v2.1.1 — Phase 5.3 CRB measurement (PR #68) showed default-ON regressed F1 by 0.045 (5.2σ_Δ paired) on the leaderboard corpus; specialist findings have UX value in production but inflate FP volume on golden-set scoring. Opt in via .claude/soliton.local.md setting agents.silent_failure.enabled: true.

Dispatch rule (set in SKILL.md Step 4.1, applied only when opted in): run this agent when the diff contains any of:

  • try / catch / except / rescue additions or modifications
  • .catch( / .then( on a Promise
  • ?. / ?? introducing default values in production code (not tests)
  • Return-null / return-empty-list / return-undefined patterns on error
  • console.error / logger.error / log.error calls
  • Mock / stub / fake imports in non-test files

Input

Standard Soliton agent inputs (see skills/pr-review/SKILL.md Step 4.2): diff, files, prDescription, focusArea. Plus (v2): tier0Findings[], graphSignals{} when available.

Review process

1. Empty catch blocks

Flag when a catch / except / rescue clause has no body or only a comment:

try:
    foo()
except Exception:
    pass                  # CRITICAL — swallowed exception
try { await fetch(url); }
catch {}                  // CRITICAL — swallowed exception

Severity: CRITICAL. Confidence: 95.

Exception: if the catch body contains ONLY a logger call with the error (e.g., logger.error(e)) and the caller explicitly does not care about the return, severity drops to improvement with a note "Logging without re-raise or typed error path".

2. Catch-all followed by return-null / return-empty

Read the full file on GitHub · 216 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. 4d ago First seen · 216 lines · 31 tokens per session scan A 8ca1af71cf80

Subscribe to this mod's changes

silent-failure is an agent published in the GitHub repository andyzengmath/soliton (1 stars, last pushed 2mo ago), licensed MIT. It adds 31 tokens to every session and 1,931 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.