# Check your agent setups for the lethal trifecta | IonBin

> List each agent workflow with its connectors, ask Claude to mark which have private data, untrusted input and an outbound channel, and fix any with all

Canonical page: https://ionbin.com/workflows/lethal-trifecta-check/
Agent guide: https://ionbin.com/llms.txt

This is a readable version of the published page. Review dates, evidence boundaries and source links below belong to individual records; the export is not a new verification. Examples are reference material, not instructions to execute.

## Check your agent setups for the lethal trifecta

[Build](https://ionbin.com/use-cases/dev-it/) · [Ops](https://ionbin.com/use-cases/ops-admin-automation/)Checked 2026-10-18

Identify combinations of data access, untrusted input and outbound access that need stronger boundaries.

- [01**You**](https://ionbin.com/workflows/lethal-trifecta-check/#workflow-step-1)

- [02**Claude**](https://ionbin.com/workflows/lethal-trifecta-check/#workflow-step-2)

- [03**You**](https://ionbin.com/workflows/lethal-trifecta-check/#workflow-step-3)

- [04**GitHub**](https://ionbin.com/workflows/lethal-trifecta-check/#workflow-step-4)

Recipe context

List each agent workflow with its connectors, ask [Claude](https://ionbin.com/tools/claude/) to mark which have private data, untrusted input and an outbound channel, and fix any with all three. Keep the audit in a [GitHub](https://ionbin.com/tools/github/) repository so every change is reviewed in a pull request.

**Independent review.** Unpaid; no partnership with the tools on this page. Based on public documentation and research. Trademarks identify their owners’ products. [Disclosure](https://ionbin.com/disclosure/).

Editorial shortlist

### Compare your options.

**For reviewing agent capability combinations.**

Inspect untrusted input, private data and external writes together; this is not certification. Choose one for implementing and reviewing code changes; switching may need setup changes.

**Used by Ion** · separate from rank.

Unpaid picks. No partnership with [OpenAI Codex](https://ionbin.com/tools/openai-codex/), Cursor, [Google Antigravity](https://ionbin.com/tools/google-antigravity/).

- 1 · Primary / high
  #### [OpenAI Codex](https://ionbin.com/tools/openai-codex/)
   Used by Ion
  **Why it fits:** A coding workspace supports edits, tests and reviewable repository changes.
  **Tradeoff:** A development environment and capable reviewer remain necessary.
  Choose it when the result should be a small patch with test evidence.
  Evidence for OpenAI Codex
  The linked [IonBin review](https://ionbin.com/tools/openai-codex/) records capabilities, access limits and source links. Review date: 2026-10-18. [Primary product reference ↗](https://developers.openai.com/codex)

- 2 · Secondary / medium
  #### [Cursor](https://ionbin.com/tools/cursor/)
  **Why it fits:** An AI code editor keeps implementation close to interactive developer review.
  **Tradeoff:** The team must work with code; model availability depends on the plan.
  Prefer it when a developer wants to steer changes while navigating the editor.
  Evidence for Cursor
  The linked [IonBin review](https://ionbin.com/tools/cursor/) records capabilities, access limits and source links. Review date: 2026-10-21. [Primary product reference ↗](https://cursor.com/)

- 3 · Third / lower
  #### [Google Antigravity](https://ionbin.com/tools/google-antigravity/)
  **Why it fits:** Development-agent surfaces offer another route for code and tool-assisted implementation.
  **Tradeoff:** Capabilities differ across desktop, CLI and other surfaces.
  Prefer it when the team already uses its environment and can validate the needed tools.
  Evidence for Google Antigravity
  The linked [IonBin review](https://ionbin.com/tools/google-antigravity/) records capabilities, access limits and source links. Review date: 2026-10-18. [Primary product reference ↗](https://antigravity.google/)

How this shortlist was chosen

Editorial judgment, not a personal-use endorsement or performance score. These are alternatives for the named role, not three required stack components. Vendors did not pay for, request or approve these positions.

Inspectable file changes come before editor convenience. A lower-ranked option may fit your constraints better; compare the conditions stated on each card before choosing.

**Setup and portability:** Use scoped repositories, tests and human review. Switching agents does not migrate settings, credentials or model access.

Reviewed 2026-10-21. This page applies a shared comparison for “implementing and reviewing code changes” to the stated task and constraints. It is documentation-based editorial guidance, not three new hands-on tests. We weigh value for the job, distinct capabilities, replacement limits, technical effort and interoperability. Check current plan terms before adopting or switching.

See [our review method](https://ionbin.com/editorial-policy/), [MCP connections](https://ionbin.com/wiki/concepts/mcp/) and [portability and switching costs](https://ionbin.com/wiki/questions/how-to-avoid-ai-vendor-lock-in/).

**Connects through:** Built-in integration

### Steps

Follow the handoff

Select a stage or scroll the steps.

- #### You
  List each agent workflow and its connectors
  Human stepConfirmed

- #### [Claude](https://ionbin.com/tools/claude/)
  Marks the three legs per workflow and suggests which one to remove
  AI platformConfirmed

- #### You
  Change connector scopes or add an approval step
  Human stepConfirmed

- #### [GitHub](https://ionbin.com/tools/github/)
  Stores the audit file; changes go through pull requests
  Built-in integrationConfirmed[Source ↗](https://docs.github.com/en/pull-requests)

Confirmed = vendor-documented, not Ion-tested. Human and AI-platform steps are proposed roles.

Human approval

You change every permission yourself.

Watch out for

An AI audit is a checklist, not a security review.

### Starter prompt

```
For each workflow, record yes, no or unknown for access to private data, untrusted content and outbound channels, with configuration evidence. Treat unknown as unresolved. Include HTTP requests, browser navigation, files, logs and tool results as possible data paths. Recommend enforceable restrictions and tests; do not claim that a prompt or approval alone removes the risk.
```

Written for this workflow. Paste it into your AI platform and adjust it to your situation.

### Tools in this workflow

| Tool | Role | Official site |
| --- | --- | --- |
| [Claude](https://ionbin.com/tools/claude/) | AI platform | [claude.ai ↗](https://claude.ai/) |
| [GitHub](https://ionbin.com/tools/github/) | Built-in integration | [github.com ↗](https://github.com/) |

### Which AI platforms can run it

| Platform | Status | Why |
| --- | --- | --- |
| [Claude (incl. Claude Code)](https://ionbin.com/platforms/claude/) | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |
| [OpenAI Codex](https://ionbin.com/platforms/codex/) | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |
| [Google Antigravity](https://ionbin.com/platforms/antigravity/) | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |
| [ChatGPT](https://ionbin.com/tools/chatgpt/) | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |
| Gemini | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |
| [Perplexity](https://ionbin.com/tools/perplexity/) | Confirmed | No platform-specific connection: the steps run in the tools themselves (browser, upload or built-in), with documented capabilities and human review steps. This is not an end-to-end test. |

What the platform statuses mean

**Confirmed:** documented access meets the recorded requirements. **Inferred:** a reviewed note describes a plausible, untested pairing. **To verify:** a connection or step is unresolved. MCP and API pairings need route-specific evidence; a plugin, CLI or API alone does not prove compatible access, credentials or permissions. These are documentation assessments, not IonBin runtime tests.

### What we checked (2026-10-17)

Simon Willison’s lethal trifecta (June 16, 2025) is the combination of access to private data, exposure to untrusted content and the ability to communicate externally. With all three in one agent session, text an attacker controls can lead the agent to send your data out. He is deeply suspicious of guardrail products that claim to catch “95% of attacks”, because in security 95% is a failing grade, and says that for people mixing tools together the only safe option is to avoid the combination entirely. See [the lethal trifecta](https://ionbin.com/wiki/concepts/lethal-trifecta/).

### Set it up

- Write a table of every agent workflow: what it reads, where untrusted text can enter, and what it can send, post or open.

- Ask [Claude](https://ionbin.com/tools/claude/) to mark the three legs for each row and suggest which leg to remove.

- Remove unnecessary data access, untrusted inputs or outbound channels using enforced controls. Splitting steps helps only if the boundary prevents malicious instructions or secrets crossing with the handoff. An approval step is an additional control, not removal of a leg.

- Store the audit in a restricted repository, without credentials or sensitive payloads. Configure review rules if approval must be mandatory, and test attempted disallowed actions after each permission change.

This is a useful threat-model checklist, not a complete security assessment. Missing one leg does not rule out destructive writes, credential theft or other prompt-injection harms.

### Related agent skills

Two reviewed skills can support this check. Sentry’s [Skill Scanner](https://ionbin.com/skills/sentry-skill-scanner/) looks for skills that change agent settings or reach out to unknown sites. OpenAI’s [Security Threat Model](https://ionbin.com/skills/openai-security-threat-model/) maps trust boundaries and abuse paths in a codebase. Neither is a security guarantee, and IonBin reviewed both from source only.

### Sources

- [The lethal trifecta for AI agents](https://simonwillison.net/2025/jun/16/the-lethal-trifecta/) · Simon Willison · practitioner · accessed 2026-10-18

- [Pull requests documentation](https://docs.github.com/en/pull-requests) · [GitHub](https://ionbin.com/tools/github/) · official · accessed 2026-10-18

### Related in the wiki

[The lethal trifecta (agent security)](https://ionbin.com/wiki/concepts/lethal-trifecta/)[I need an AI agent to take actions without risking our systems](https://ionbin.com/wiki/needs/let-an-agent-act-safely/)

### Related workflows

[Turn sprint issues into booked time](https://ionbin.com/workflows/sprint-planning-to-calendar/)[Package your best prompts as reusable agent skills](https://ionbin.com/workflows/reusable-agent-skills/)[Make your offer page readable by AI agents](https://ionbin.com/workflows/agent-readable-offer-page/)[Get a weekly repo report from a scheduled agent](https://ionbin.com/workflows/scheduled-weekly-repo-report/)

No affiliate links on this page. No paid placements in workflows. [How IonBin makes money](https://ionbin.com/disclosure/).
