The latest articles on DEV Community by Alex Rivera (@briefstack_alex). https://dev.to/briefstack_alex https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4152157%2Fa079f35a-d415-4af8-83fe-304ba5f491a5.png https://dev.to/briefstack_alex en Alex Rivera Wed, 30 Sep 2026 11:32:44 +0000 https://dev.to/briefstack_alex/agent-task-briefing-checklist-10-anti-patterns-free-20nl https://dev.to/briefstack_alex/agent-task-briefing-checklist-10-anti-patterns-free-20nl <p>If you hand work to an AI agent (Cursor, Claude Code, Codex, custom agents), most failures trace back to the brief, not the model. Here are 10 anti-patterns I keep seeing, each with a one-line fix. Copy freely.</p> <p>Smell these in your briefs. Each wastes tokens, time, or trust.</p> <h2> 1. The vibes brief </h2> <p>“Make it better.” / “Clean this up.” / “Explore options.”<br><br> <strong>Fix:</strong> Name the artifact and the acceptance test.</p> <h2> 2. Hidden prod access </h2> <p>Tools say “repo write: Y” with no path limits.<br><br> <strong>Fix:</strong> Allowlist directories; forbid <code>.env</code>, secrets, infra by default.</p> <h2> 3. Success = activity </h2> <p>“Research the market” with no table, memo, or decision.<br><br> <strong>Fix:</strong> Definition of Done = files a human will open in a meeting.</p> <h2> 4. Constraint collision </h2> <p>“Be exhaustive” + “max 10 minutes” + “cite everything.”<br><br> <strong>Fix:</strong> Rank constraints; mark Hard vs Soft.</p> <h2> 5. Secret smuggling </h2> <p>API keys in the brief “for convenience.”<br><br> <strong>Fix:</strong> Point to a secret manager / env name; never paste values.</p> <h2> 6. Infinite autonomy </h2> <p>“Whatever you need to get it done.”<br><br> <strong>Fix:</strong> Escalation table; stop conditions.</p> <h2> 7. Unowned brief </h2> <p>No human owner → agent invents product policy.<br><br> <strong>Fix:</strong> Owner field required; acceptance is a human step.</p> <h2> 8. Example-free taste tasks </h2> <p>“Write in our voice” with no voice doc or samples.<br><br> <strong>Fix:</strong> Link 2 good examples or a 10-line voice sheet.</p> <h2> 9. Fake precision </h2> <p>“Confidence must be 95%.”<br><br> <strong>Fix:</strong> Ask for source grades and <code>assumption</code> tags instead.</p> <h2> 10. Scope amnesia </h2> <p>Out of scope empty → agent expands forever.<br><br> <strong>Fix:</strong> At least two explicit non-goals.</p> <h2> Quick pre-flight checklist </h2> <ul> <li>[ ] Named artifact + acceptance test</li> <li>[ ] Tool/path allowlist; secrets referenced by env name only</li> <li>[ ] Hard vs Soft constraints ranked</li> <li>[ ] Stop conditions + escalation owner</li> <li>[ ] At least two explicit non-goals</li> <li>[ ] 1–2 examples for any "taste" task</li> </ul> <p><em>This is a free excerpt from a small set of briefing templates I sell. If you want the full fill-in-the-blank packs (briefing, eval, digest, research, incident), they're at <a href="https://briefstack.surge.sh/" rel="noopener noreferrer">https://briefstack.surge.sh/</a> — the list above is useful on its own either way.</em></p> ai productivity agents llm