Actions
Safety guide for the Amazon Ads operations Titan AI Connect can take on your behalf. Read this before enabling actions in any host.
⚠️ These tools spend real money
Action tools call the Amazon Ads API. A single misread of a sales report could spawn 10 campaigns at $500/day each. There is no automatic rollback. Treat every operation as if you were typing into the Amazon Ads UI yourself.
What are actions?
Actions are operations that change something in your Amazon Advertising account — pausing a campaign, adjusting a bid, adding a negative keyword, launching a new structure. Titan AI Connect proposes them to you; you approve or deny per call. They cover Sponsored Products, Sponsored Brands, and Sponsored Display surfaces.
Real-world danger scenarios
- Duplicate campaigns: agent misreads a "do this for each ASIN" instruction and creates 10 copies of the same campaign at full daily budget each.
- Wrong-marketplace target: a US-targeted keyword gets attached to a UK ad group because the agent didn't re-validate the active marketplace before the call.
- Accidental archive: an "update state" call on a winning campaign pauses it instead of the intended low-performer.
- Always-Allow runaway: the user toggled "Always Allow" once for a single low-risk negative keyword, then a follow-up turn spawns 10 high-risk campaigns without further prompts.
Dry-run by default
Titan AI Connect ships in dry-run mode by default — every action is validated against Amazon's API but is NOT executed. The result tells you exactly what would have happened. Your operator team flips to live mode only after the runbook is signed off.
⚠️ Do not enable "Always Allow"
Most MCP hosts (Claude.ai, Claude Desktop, Claude Code) let you toggle "Always Allow" per tool — once on, the model can call that tool with no further prompt. Once you toggle Always Allow for an action tool, the human-in-the-loop is gone. We strongly recommend reviewing every action call. The tools are designed assuming you'll see and read each prompt.
Approval surfaces by host
| Host | Approval surface |
|---|---|
| Claude.ai web | Modal dialog with tool name + arguments |
| Claude Desktop | Similar permission dialog |
| Claude Code plugin | Permission prompt unless pre-approved in Claude Code settings |
| Cowork | Per-tool prompt |
| OpenClaw | NO PROMPT.OpenClaw runs unattended — the skill's narration is the only safeguard. |
Rollback recipes
| Action | How to undo |
|---|---|
| Created campaign | Re-pause via the same operation, setting status to Archived. |
| Created keyword | Update the keyword's status to Archived. |
| Updated budget | Re-run the campaign update with the prior budget value. |
| Added neg keyword | Update its state to Archived via the matching propose_update_*_neg_keyword tool. |
| Added neg target | Update its state to Archived via the matching propose_update_*_neg_target tool. |
Each rollback call also requires user approval (or, in OpenClaw, explicit narration before the call).
Recommended account hygiene
- Set account-wide budget capsin Titan Tools or directly in Amazon Ads Console before enabling action tools. Even a runaway agent can't create campaigns above the cap.
- Start with low-risk actions only(negative keywords, campaign pauses). Don't enable high-risk tools until you've watched a full week of low-risk usage.
- Audit your Amazon Ads change history weekly for the first month. Spot anything you didn't intend and reach out before it compounds.
- Use the same Titan AI Connect account for all action calls — don't share access across team members. Audit-trail integrity depends on a 1:1 user mapping.
Reporting unexpected behavior
When reporting an issue, include the timestamp and the action description; we'll trace it through our logs.
Per-interface gotchas live on the host docs: Claude.ai, Claude Code, Cowork, OpenClaw, Other clients.