One workflow launch score
Scores side effects, data exposure, untrusted instructions, approval gates, rollback, and launch evidence.
What you get in 10 minutes
A concrete preview of the buyer workflow: score one AI agent launch, turn the risk into a short brief, map tool permissions, and collect evidence without pasting secrets or private customer data.
0-2 min
Run the free risk score against one workflow and identify whether the risky surface is tools, data, memory, public output, or launch evidence.
2-5 min
Generate a non-sensitive launch brief with the workflow boundary, highest side effect, controls already present, and immediate fixes.
5-8 min
Map tool actions to approval gates and pick the prompt-injection or MCP fixture patterns that should be replayed before launch.
8-10 min
Decide whether the digital pack is enough or whether one fixed-scope 24-hour review is useful after safe intake.
Non-sensitive launch brief
Converts a workflow description into top launch questions, immediate fixes, evidence to gather, and material not to send.
Agent action matrix
Maps read, draft, write, command, financial, identity, and credential actions to approvals and launch proof.
Prompt-injection test set
Starts a replayable fixture set for untrusted content, tool output, metadata attack, exfiltration, and approval bypass cases.
Private-data guardrails
Prepares a high-level intake that keeps credentials, payment details, private screenshots, customer records, and full identifiers out.
Fictional sample report
Shows the Markdown delivery format for top risks, severity, rationale, and three prioritized fixes.
Use it when
The pack is designed for builders who need a fast launch screen before a small AI workflow reaches real systems.
- An agent can browse, email, post, edit records, run code, call tools, or touch MCP servers.
- The workflow sees customer context, support records, documents, memory, retrieval, uploads, or tool output.
- You need practical launch evidence before a demo, client handoff, pilot, or public release.
Not for
This is a lightweight pre-launch screen, not a certification or incident response workflow.
Not legal advice, compliance certification, penetration testing, incident response, deep source-code review, or a guarantee that a system is secure.