Agent Failure Archive
186 real AI agent post-mortems, 107 of them measurement failures. Free tools, paid via x402.
我该使用它吗
质量与安全性
发现（2）
- LOWTool 'sample' description lacks action verb在 sample 中
- LOWTool 'catalog' description lacks action verb在 catalog 中
基于对工具定义和协议合规性的自动分析。
上下文开销
这是每次将服务器的工具加载到模型上下文窗口时所消耗的大致 token 数。数值越高，可用于其他任务的注意力就越少。
安装
一键安装
将以下内容添加到你的 `claude_desktop_config.json` 文件中：
{
"mcpServers": {
"agent-failure-archive": {
"url": "https://desktop-ai2ata5-1.tailfeb765.ts.net/mcp"
}
}
}远程端点
https://desktop-ai2ata5-1.tailfeb765.ts.net/mcpstreamable-http它能做什么
工具清单
工具（9）
🟢precheck(claim, evidence)
Free, no wallet. Check a conclusion against nine known ways of fooling yourself and get back which checks it trips plus the question each one asks. Use it before writing 'we found that'. The paid audit tool adds why each matters, the real incident with the numbers measured at the time, and what to run.
输入模式
{
"type": "object",
"properties": {
"claim": {
"title": "Claim",
"type": "string"
},
"evidence": {
"default": "",
"title": "Evidence",
"type": "string"
}
},
"required": [
"claim"
],
"title": "precheckArguments"
}输出模式
{
"type": "object",
"properties": {
"result": {
"title": "Result",
"type": "string"
}
},
"required": [
"result"
],
"title": "precheckOutput"
}🟡sample
Free, no wallet. Two complete post-mortems from the archive.
输入模式
{
"type": "object",
"properties": {},
"title": "sampleArguments"
}输出模式
{
"type": "object",
"properties": {
"result": {
"title": "Result",
"type": "string"
}
},
"required": [
"result"
],
"title": "sampleOutput"
}⚪catalog
Free, no wallet. What the archive holds, what each paid tool costs, and how payment works.
输入模式
{
"type": "object",
"properties": {},
"title": "catalogArguments"
}输出模式
{
"type": "object",
"properties": {
"result": {
"title": "Result",
"type": "string"
}
},
"required": [
"result"
],
"title": "catalogOutput"
}🟢contents(theme)
Free, no wallet. Every case title in the archive, tagged with the trap it illustrates. Titles only, no bodies. Read this to see what the paid archive actually contains before deciding it is worth a dollar.
输入模式
{
"type": "object",
"properties": {
"theme": {
"default": "",
"title": "Theme",
"type": "string"
}
},
"title": "contentsArguments"
}输出模式
{
"type": "object",
"properties": {
"result": {
"title": "Result",
"type": "string"
}
},
"required": [
"result"
],
"title": "contentsOutput"
}🟢audit(claim, evidence)
$0.02. Full audit of a claim: why each tripped check matters, the incident behind it with measured numbers, and what to run.
输入模式
{
"type": "object",
"properties": {
"claim": {
"title": "Claim",
"type": "string"
},
"evidence": {
"default": "",
"title": "Evidence",
"type": "string"
}
},
"required": [
"claim"
],
"title": "auditArguments"
}🟡search(q)
$0.01. Three real agent post-mortems matching a symptom, each with root cause, the fix that worked, and the prevention rule.
输入模式
{
"type": "object",
"properties": {
"q": {
"title": "Q",
"type": "string"
}
},
"required": [
"q"
],
"title": "searchArguments"
}⚪brief(action)
$0.05. Pre-flight risk brief before an irreversible action, drawn from the ways that class of action actually failed.
输入模式
{
"type": "object",
"properties": {
"action": {
"title": "Action",
"type": "string"
}
},
"required": [
"action"
],
"title": "briefArguments"
}🟢research(q)
$0.25. The 107 measurement failures from an eight-month attempt to measure one person's individuation with embeddings.
输入模式
{
"type": "object",
"properties": {
"q": {
"default": "",
"title": "Q",
"type": "string"
}
},
"title": "researchArguments"
}⚪archive
$1.00. Every case in one response. One payment, no subscription, no account.
输入模式
{
"type": "object",
"properties": {},
"title": "archiveArguments"
}社区
证据