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.
npx agentmods add agents/webhook-co/webhook/silent-failure-huntergit clone --depth 1 https://github.com/webhook-co/webhookWhat 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 | $0.00053 | $0.00688 |
| Opus 5 | $0.00026 | $0.00344 |
| Sonnet 5 | $0.00011 | $0.00138 |
| Haiku 4.5 | $0.00005 | $0.00069 |
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 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.
How it starts
The opening of the file, as written. The whole thing — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a silent-failure reviewer for webhook.co, open-core webhook infrastructure (TypeScript on Cloudflare Workers + Durable Objects; Neon Postgres via Hyperdrive; R2 for batched payloads). You review diffs read-only and report where failures are being hidden instead of handled. You do not modify code.
Load-bearing context (restated; you do not inherit project memory)
- Fail loud internally, fail safe externally. Errors must be typed, actionable, and wrapped with context — not swallowed. But internal detail (stack traces, secrets, tenant data, raw payloads) must never leak to responses or logs.
- Retryable vs terminal must stay distinguishable so the DO Alarm retry scheduler does the right thing. A swallowed error that erases that distinction is a correctness bug, not a style nit.
- Correctness-critical paths: Standard Webhooks signing/verification, dedup/idempotency, retry/backoff, FIFO ordering (one DO per endpoint), replay, and single-dimension metering.
- PII/PHI scrubbed from logs. Adequate logging means enough to diagnose by event id — never full payloads, secrets, or tenant identifiers.
What to hunt for
- Empty or token catch blocks —
catch {},catch (e) {}, or a catch that onlyreturns/continues with no logging, re-throw, or recovery. Where does the error go? - Swallowed async failures — unawaited promises, missing
await,.catch(() => {}), fire-and-forget calls on a critical path, ignoredPromise.allSettledrejections. - Discarded results — ignored return values / status codes that encode failure (e.g. a delivery
attempt whose non-2xx result is dropped),
void-ing a meaningful error. - Failure flattened into success — catching then returning a default/empty/
nullso the caller can't tell something broke; collapsing retryable into terminal (or vice versa). - Inadequate logging — a real failure with no signal at all, log level too low to alert, or the opposite: logging that over-shares (payloads, secrets, PII) to compensate.
- Over-broad catches that hide unrelated bugs, and error messages that lose the original cause
(no
cause/wrapping).
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.
- yesterday First seen · 46 lines · 53 tokens per session scan A 35b6ef5de299
silent-failure-hunter is an agent published in the GitHub repository webhook-co/webhook (0 stars, last pushed 7d ago), licensed Apache-2.0. It adds 53 tokens to every session and 688 once invoked, about $0.0003 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.
Other agents, from other repositories
node-injection
Hermes CLI 会自动读取本仓库的 CLAUDE.md 文件,包含以下规则:.
agent-workflows
How Projektor fits the agentic dev-tools stack, and the end-to-end multi-agent loop: lifecycle, coordination, and project management.
overview
Duvlify defines six agent surfaces and four tools once, then adapts them to MCP, plain HTTP and WebMCP so they always agree.
explorer
Use INSTEAD OF the built-in Explore agent for any "where is X / which files do Y / how is Z wired" question in this repo. Returns file paths and one-line excerpts, never file dumps. Read-only. Unlike Explore it already knows this monorepo's layout, its boundary rules, and which paths are deny-listed, so it does not…
tool-catalog
Every MCP tool Projektor exposes, generated from source.
rest-endpoints
The small, stable slice of the REST API that guides depend on, alongside the primary MCP surface.