10
RaindropAPP
Build loop from deprecated Webpack config
128 events42 usersseverity high
Retries restore module.loaders, which Webpack 5 no longer supports.
Skip to content
EventsProductionLast 24 hours # agent-alerts Build failure loopProduction jamiecommented 8 minutes ago raindropbotcommented just now
Better behavior.
Evan GoldschmidtCTO, Tolan
Get started free Start 14-day free trial Talk to sales Head of AIFortune 100 Find the next failure.
Series AA new chapter for Raindrop.$50m raisedDetect agent issues.
Detect agent issues.
Prevent them from happening.
Find silent failures in production, investigate what went wrong, and simulate changes in CI.
Start free. No credit card required.See how it works
Built with teams shipping agents to millions.
coding_session
Fix the failing Webpack build
I’ll inspect the configuration and run the build to reproduce the error.
globpackage.json webpack.config.js tsconfig.json22ms
Input
{
"pattern": "**/*config*"
}Output
package.json webpack.config.js tsconfig.json
read{ "scripts": { "build": "webpack", "test": "vitest run" } }64ms
Input
{
"path": "package.json"
}Output
{ "scripts": { "build": "webpack", "test": "vitest run" } }readmodule.exports = { module: { loaders: [...] } };41ms
Input
{
"path": "webpack.config.js"
}Output
module.exports = { module: { loaders: [...] } };apply_patchUpdated webpack.config.js182ms
Input
{
"path": "webpack.config.js"
}Output
Updated webpack.config.js
bashInvalid configuration object. - configuration.module has an unknown property 'loaders'. These properties are valid: rules, defaultRules, ...8.2s
Invalid configuration object. - configuration.module has an unknown property 'loaders'. These properties are valid: rules, defaultRules, ...
Input
{
"command": "npm run build"
}The build failed. I’ll update the configuration and try again.
readmodule: { loaders: [...] }43ms
Input
{
"path": "webpack.config.js"
}Output
module: { loaders: [...] }apply_patchUpdated webpack.config.js193ms
Input
{
"path": "webpack.config.js"
}Output
Updated webpack.config.js
bashInvalid configuration object. - configuration.module has an unknown property 'loaders'. These properties are valid: rules, defaultRules, ...8.4s
Invalid configuration object. - configuration.module has an unknown property 'loaders'. These properties are valid: rules, defaultRules, ...
Input
{
"command": "npm run build"
}The build is still failing. I’ll try adjusting the configuration again.
Ask a question...
acme / support-agentPrivate
Retry tools on transient failures #142
Openjamie wants to merge 2 commits into
main from fix/tool-retriesConversation 2Commits 2Checks 3Files changed 2
Retry timeouts automatically so customers don’t have to ask twice.
Retry transient tool failures
7e4b2c142 tests passedraindropbotcommented just nowSimulation found issues
The refund succeeds, but its response times out. The new retry issues another refund.
| Same customer request | main | This PR |
|---|---|---|
| Refund a $120 order | $120 refunded | $240 refunded |
The agent still says: I’ve refunded $120 to your card.
1 regression11 passed
ExperimentsConfig template v2
Config template v2 vs. baseline
SavedSetupExperiment vs baseline
Experiment
Property
config_version=v2Last 7 days
Baseline
Property
config_version≠v2Last 7 days
Results
ExperimentUsers with negative signals
3.40%7.8 pp
1,280 users3,840 events
BaselineUsers with negative signals
11.20%
1,280 users3,840 events
Signal comparison
ExperimentBaseline
Build failures79%
2.1%
9.8%
Agent retry loops88%
0.8%
6.4%
Config syntax errors75%
1.2%
4.8%
Measured in production
Better behavior.
A real difference.
“Without Raindrop we couldn’t tell if we were actually fixing the problem.”
Evan GoldschmidtCTO, TolanTolan used Raindrop to identify forgotten details and missed reminders, then test changes to retrieval and storage.
Loved by developers
Here's what developers building with Raindrop are saying
Hey this is awesome! Finally a product where you can "look at your data" in a beautiful way, not just a soulless dump of jsons for nerds. I've been building sth along those lines for customer support (not observability, and not as polished). I display convo history as blocks, similar to Notion, and you get notion-like collaboration too for feedback. Is this possible in raindrop?
If you're building an agent, I recommend looking into Raindrop for monitoring. We use it.
I’ve been using this all day today. Soooo much easier to iterate. Spans, token count, cost, timing, a sane view of each tools input-output.
Heard it works with your codex subscription. Now you can use codex to build agents as well as inspect them I tried it out yesterday and it works great in the in app browser of the codex desktop app
Observability needs to be more actionable. Raindrop was already one of my most used tools and it just keeps getting better (!)
absolute 🔥, all other trajectory viewers are unusable in comparison
Really like the UX/data viz in some of these Increasingly important for sorting through multi-agent / parallel workflows to tame the trace spagetthi
!! we're using this new sdk to page us every time something breaks
huge fan of Raindrop, it has been immensely helpful for us and i use it every day. congrats on the round!
@benhylak @raindrop_ai We recently got an influx of users from China and used Deep Search to make a classifier for "the user speaking chinese" this morning. Raindrop translates them, so we're able to get to know a group of users we'd otherwise have no insight into. Feels like I'm living in the future
loved the onboarding experience
Very good. Immediately useful. Going to be using this as default for local agent work now Also looking at instrument Claude code sessions with it to improve skills etc
@aviflombaum I’m a huge fan of @raindrop_ai for agent monitoring
@jmwind @benhylak We are loving Raindrop!
This is very well done. Built in chat that plugs into your local codex/claude is perfect. Need someone to make this for general otel too
This is insanely cool. 🔥🔥 One of the best products out there (by far) for anyone building agents just released an OSS version to play with. We’ve already been using Raindrop heavily across Thine / Merlin. Excited to now use it across personal projects too.
Spellbook has been using this for weeks and it feels like magic. It raises silent bugs daily. Feels like my agent is a team member that advocates for itself telling us the most important bugs for us to fix.
Raindrop feels like extremely futuristic software. It lets you explore user intent and agent trajectories in a way that didn't feel possible years ago. Big fan of them. Every part of what they've built is thoughtful, solving design problems for a new kind of observability.
once again the @raindrop_ai is ahead of the curve…imagine searching for expensive, error prone, or slow tool calls across all your agent traces. nothing like this exists yet
@tolanworld runs on raindrop. highly recommend
just made a feature request for @raindrop_ai and they shipped it in less than 3 hours
This is actually incredible. I've been building out a project that involved running Pi in a cloudflare container, it's been a nightmare locally to try and get decent observability on what's happening and to get a feedback loop going. This really came at the perfect time!
nice, it works awesome for voice agents too!
@benhylak @raindrop_ai awesome
@benhylak Big fan of raindrop ! 💪🏼
Such a nice idea. we need this for @DevinAI for Terminal @MattSchrage @alexkaplan0 👀
Agents finding bugs in agents and tagging another agent to fix those agents Welcome to the singularity
It's difficult to overstate how insanely useful @raindrop_ai is, and it just keeps getting better
Raindrop will be remembered as the company who did to AI infra what the Mac did to the personal computer. It’s a matter of time until frontier labs recognize the value and make a move.
Happy customers
Easily my favorite tool to discover, analyze, and solve agent issues @clay
This is awesome We @lexdotpage are very happy customers of @raindrop_ai!
finally somebody built this
Congrats on the launch! Have been building many versions of this from scratch for various projects, and now I can finally use a unified tool for it.
Your agent, monitored.
From local testing to enterprise deployments, Raindrop has you covered.
Signals
Define custom classifiers for the behaviors that matter.
Aesthetic Complaints
7ddetails
match
83%
trend
rising
Issues
Get alerted the moment production behavior goes off track.
Issue detected
refund_agent loop
01get_refund_status200
02retry_refundloop
03retry_refundloop
Triage Agent
Investigate issues and automate follow-ups in Slack, the web, or through MCP.
#agent-issues
R
Raindrop APP
Found a retry loop in refund_agent.
Fix ready for review
Add a retry limit.
Experiments
Prove improvements behind a feature flag before rolling them out.
model_fix_v2
feature flag
control
variant
MCP
Connect your tools and let Raindrop run auto-fix loops for your agent.
trace
patch
verify
mcp.run_autofix
› replaying failed trace
fix verified
Security trusted by the Fortune 100
Full visibility into your agents, backed by the security guarantees your organization already runs on.
SOC 2 Compliance
SOC 2 Type II compliant, independently audited each year. Full reports available in our Trust Center.
View our Trust Center →PII Guard
Intelligent server-side PII redaction keeps sensitive user data protected.
Enterprise Ready
SSO, audit logs, access controls, procurement support, and dedicated implementation help.
Get started for free.
No credit card required. Upgrade as you grow.
Hobby
Start tracing for free.
$0/mo
1,000 events/mo included
The essentials
- Tracing & search
- Slack notifications
- MCP & Query SDK
More features & limits
- 14-day retention
- 1 custom signal
- 15 agent messages / month
- Stumbles (preview)
One event is a user message and agent response, including tool calls and sub-agents, or one background agent run.
Pro
Find and fix agent failures.
$299/mo
+ $0.003 / event up to 1M, then $0.002 / event
Everything in Hobby, plus
- Realtime issue detection
- Triage agent to investigate failures
- Experiments & semantic search
More features & limits
- 10 custom signals
- 500 agent messages / month
One event is a user message and agent response, including tool calls and sub-agents, or one background agent run.
Enterprise
Security and support at scale.
Custom
Pricing tailored to your team
Everything in Pro, plus
- SSO / SAML & audit logs
- Edge-PII redaction
- Priority support & SLA guarantees
More features & limits
- Snowflake / BigQuery exports
Need a custom deployment? Get in touch.
“We chose Raindrop over Braintrust, LangSmith, and Arize. Most tooling targets the first era of LLMs: call/response, deterministic evals, testing edge cases. Raindrop is building for the next era, proactively finding issues and driving automatic improvement loops.”
Build with the evidence
Find the next failure.
Verify the next fix.
Start free. No credit card required.