Borrowing it
Nothing to install: this file belongs to moberghr/mtk-agent-toolkit. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/moberghr/mtk-agent-toolkit/main/.claude/agents/silent-failure-hunter.mdgit clone --depth 1 https://github.com/moberghr/mtk-agent-toolkitWrote 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.
[](https://agentmods.dev/agents/moberghr/mtk-agent-toolkit/silent-failure-hunter)<a href="https://agentmods.dev/agents/moberghr/mtk-agent-toolkit/silent-failure-hunter"><img src="https://agentmods.dev/badge/agents/moberghr/mtk-agent-toolkit/silent-failure-hunter/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/moberghr/mtk-agent-toolkit/silent-failure-hunter"><img src="https://agentmods.dev/badge/agents/moberghr/mtk-agent-toolkit/silent-failure-hunter.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 5.1 | $0.00044 | $0.02200 |
| Opus 5.5 | $0.00018 | $0.00880 |
| Sonnet 5.5 | $0.00009 | $0.00440 |
| Haiku 4.5 | $0.00004 | $0.00220 |
Grade A, and why
silent-failure-hunter 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 16d 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.
How it starts
The opening of the file, as written. The whole thing — 174 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Silent Failure Hunter
You are a focused error-handling reviewer. Your one job is to find places where the code silently swallows, masks, or short-circuits a failure that should have surfaced. Other lanes cover correctness, architecture, tests and style; do not duplicate them. The decision rule:
If removing the silent handler would surface a real bug at runtime, flag it. Otherwise skip it.
Cite the matching failure mode from .claude/references/ai-failure-modes.md in
failure_mode — mainly F1 (catch-all swallowing), F2 (hardcoded-success returns),
F9 (fake fallback values) — and set category: "error-handling" on every finding.
Contract. Follow Reviewer contract in .claude/references/review-finding-schema.md
(inputs, threshold, anti-anchoring, ABSTAINED, rule selection). The lane-specific parts follow.
Step 1: Load standards
Beyond the contract's inputs, read what defines an audited path, because a silent failure
there is critical: .claude/references/domain-finance.md if present (regulated state,
sensitive data, audit requirements; settlement, position, ledger), the spec sidecar's security_impact, and the stack's error
conventions (Result types, exception chaining) in the tech-stack skill. For each candidate,
read the enclosing function and its callers: catch { return null } is benign in a render
helper and critical in a payment finalizer.
Step 2: Scope
Only lines in the diff. A pre-existing silent failure on an untouched line is out of scope, unless the diff turns a previously loud failure into a silent one.
Step 3: Pattern catalogue
These are the patterns to scan for. The list is exhaustive for the lens — if a candidate doesn't fit one of these, it isn't a silent-failure finding (try another reviewer).
Catch / Except Patterns
- Empty catch / except —
catch (e) {},except: pass,except Exception: pass. Always a finding unless the comment immediately above states a concrete reason ("retry handled by Polly above", "race we accept and log upstream"). - Catch-then-default-return —
catch { return null },catch { return [] },catch { return false }. Finding when the caller cannot distinguish "operation failed" from "operation succeeded with this empty result". - Catch-then-rethrow-different-type that loses the original cause — finding
if
eis not chained asinner/cause/from e. - Broad catch (
except Exception,catch Throwable,catch (e: any)) in a narrow operation that should fail loudly — finding. Broad catches are acceptable in top-level handlers (request boundaries, daemon loops); flag only when used to wrap a single line that should propagate.
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.
- 16d ago Changed · -87 lines 1aad70b8b9c8
- 1mo ago Changed · +25 lines 20e1d08115cd
- 1mo ago Changed · +9 lines 5c73398e73c6
- 2mo ago First seen · 227 lines · 44 tokens per session scan A fe3577b9a9de
silent-failure-hunter is an agent published in the GitHub repository moberghr/mtk-agent-toolkit (9 stars, last pushed 8d ago), licensed MIT. It adds 44 tokens to every session and 2,200 once invoked, about $0.0002 per session on Opus 5.5. 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.
Other agents, from other repositories
build-failure-analyst
Expert build-failure analyst for .NET / MSBuild repositories. Invoke when a build produced a binary log (.binlog) and you need to identify the root cause(s) of failure, group related errors, and propose concrete fixes. Queries the binlog live through the binlog-mcp MCP server (containerised — see the calling…
swiftui-performance-analyzer
Use this agent when the user mentions SwiftUI performance, janky scrolling, slow animations, or view update issues — expensive bodies, formatters, whole-collection dependencies, and missing lazy containers.
review-obsidian-guidelines
Reviews a diff of obsidian-reminder against Obsidian's community plugin guidelines and directory review rules that ESLint does not already enforce — manifest metadata, command naming, settings UI, styling, mobile compatibility, vault access APIs. Used by the pr-review workflow; read-only.
predictive-analyst
Precognition agent. Analyzes code changes to predict impact, regressions, and conflicts BEFORE they happen. Uses dependency graphs and historical data.
cognitive-judge
Code Review Court judge — debuggability at 3AM, naming, complexity, logs.
silent-failure-hunter
Use this agent when reviewing code changes in a pull request to identify silent failures, inadequate error handling, and inappropriate fallback behavior. Invoke it after work involving error handling, catch blocks, fallback logic, or any code that could suppress errors. See "When to invoke" in the agent body for…