Where do you stop trusting an AI agent with write access to Salesforce, SAP and Jira?
ken_kauneki10 · reddit · 2026-09-16
The author raises a pressing question as agents move from read-only analysis to actually performing actions: where do you stop trusting an agent with direct API access? When one agent can write to Salesforce, SAP, Slack and Jira, the options include granting it its own permissions, inheriting the user's permissions, adding human approval before destructive actions, or hard-restricting tools. The thread asks practitioners which API actions should never run without human sign-off.
More from coding & agent
- GitHub Actions' ubuntu-slim runner is lighter than ubuntu-latest, but caps jobs at 15 minutes — DanielLockyer · 2026-11-02
- GitHub Actions has a lesser-known ubuntu-slim runner: 1 CPU, 5GB RAM, cheaper CI — DanielLockyer · 2026-11-02
- Before automating a workflow with AI, ask if anyone understands it end to end — djgoel123 · 2026-11-02
- One stale date silently broke a WhatsApp reply agent: 19 messages unanswered — petrusenko_max · 2026-11-02
- The 10X engineer is now just a SKILLS.md: 6 workflow rules for AI coding agents — mdancho84 · 2026-11-02
- Auditing my MCP credential broker: the worst hole was the config file — Relevant_Mousse_9117 · 2026-11-02