Demo workspace Insights
Three days of agent traffic, audited
Five coding sessions across Claude Code, Cursor, Codex and Aider. Open any session to replay every request, the tokens it cost and the secrets that were masked before they reached the model.
Requests
15
5 sessions
Tokens
363.9K
350.2K in · 13.6K out
Est. cost
$1.18
across all sessions
Sensitive values
13
40% of requests flagged
Blocked
0
none so far
Traffic
Requests per day, with flagged volume
15 requests
13 flagged
Oct 12Oct 25
Requests by provider
Share of this window
Providerreqscost
- OpenAI47%7$0.333
- Anthropic40%6$0.845
- Ollama13%2$0.00
What was detected
Kinds of sensitive value
- EMAIL3
- SECRET2
- OPENAI_KEY1
- ANTHROPIC_KEY1
- DATABASE_URI1
- CARD1
- IBAN1
- BEARER1
Try the masking engine
It runs locally in your browser
Input — what the agent would send
Runs locally in your browser. Nothing here is sent anywhere.
Output — what the model receives6 masked
User report — please summarise before I paste it to the model. Account: [EMAIL_1] (preferred contact) Card on file: [CARD_1] IBAN: [IBAN_1] 00 OPENAI_API_KEY=[OPENAI_KEY_1] Authorization: Bearer [JWT_1] Postgres: [DATABASE_URI_1]
CARDDATABASE_URIEMAILIBANJWTOPENAI_KEY
Session replay
Select a session, then step through its requests
Anthropicclaude-sonnet-4.6Passed/v1/messages
Status200
Latency1.84 s
Tokens18.4K / 640
Cost$0.064
0 detection types · masked before sending
—Request (after masking)
tool_use: read_file apps/api/src/auth/middleware.ts (412 lines)
Response
The middleware validates the session cookie but never checks that the token audience matches the API.
Session “Refactor auth middleware” · Anthropic · masked values are stored as tokens, never as the original text.
Every request
Filter the full audit log
15 / 15
| Time | Session | Request | Detections | Tokens | Cost |
|---|---|---|---|---|---|
| 20010:47:13 | Refactor auth middlewareAnthropic | /v1/messagesopus-4.5 | EMAIL×2 Passed | 44.2K | $0.244 |
| 20010:40:13 | Refactor auth middlewareAnthropic | /v1/messagessonnet-4.6 | — | 34.8K | $0.121 |
| 20010:33:13 | Refactor auth middlewareAnthropic | /v1/messagessonnet-4.6 | OPENAI_KEYANTHROPIC_KEYDATABASE_URI Masked | 26.1K | $0.092 |
| 20010:29:13 | Refactor auth middlewareAnthropic | /v1/messagessonnet-4.6 | — | 19.1K | $0.064 |
| 42908:03:13 | Fix flaky billing testsOpenAI | /v1/chat/completionsgpt-5.2 | — | 400 | $0.00 |
| 20007:55:13 | Fix flaky billing testsOpenAI | /v1/chat/completionsgpt-5.2 | — | 23K | $0.104 |
| 20007:47:13 | Fix flaky billing testsOpenAI | /v1/chat/completionsgpt-5.2 | CARDIBAN Masked | 16.5K | $0.078 |
| 20007:41:13 | Fix flaky billing testsOpenAI | /v1/chat/completionsgpt-5.2 | — | 9.5K | $0.041 |
| 20010:11:13 | Index the docs siteOpenAI | /v1/chat/completionsgpt-5.2-mini | — | 30.6K | $0.037 |
| 20010:04:13 | Index the docs siteOpenAI | /v1/chat/completionsgpt-5.2-mini | BEARER Masked | 49K | $0.062 |
| 20009:55:13 | Index the docs siteOpenAI | /v1/chat/completionsgpt-5.2-mini | — | 6.4K | $0.011 |
| 20007:59:13 | Local embedding passOllama | /v1/chat/completionsllama3.3:70b | — | 40.8K | $0.00 |
| 20007:35:13 | Local embedding passOllama | /v1/chat/completionsllama3.3:70b | EMAILPHONE Masked | 12.7K | $0.00 |
| 20009:25:13 | Onboard the data pipelineAnthropic | /v1/messagesopus-4.5 | — | 20.4K | $0.126 |
| 20008:55:13 | Onboard the data pipelineAnthropic | /v1/messagesopus-4.5 | AWS_KEYSECRET×2 Masked | 30.3K | $0.198 |
Run this on your own machine
The dashboard is the same open-source app. Self-host it, or let us host it for €5 per seat.