Free workflow planner / Agent to Outcome

Make the work
clear enough to delegate.

Map the goal, context, tools and human decisions. Take the resulting brief to the agent you choose.

This planner creates a brief; it does not run an agent or connect accounts. Entries stay in this page unless you choose to save them in this browser or download them. Use sanitized examples, never credentials.

1. Define the outcomeWhat should be finished?Name one deliverable, its audience and what it helps someone do.

Who owns the result?What makes the result acceptable?

Which kind of work is this?

2. Provide context and boundariesWhat does the agent need to know?Audience, definitions, examples, preferred format and the reason this task matters.

Which sources are approved?Name files or systems, freshness requirements and which source wins if they conflict.What must stay out?Private records, unneeded folders, credentials, unsupported claims or actions.

3. Choose tools and authority

Which tools or connections are needed?Start with read access. List each app, file area or MCP connection and its permitted action.Which approved product and model will you test?Use one you have access to. Record its name/version when the run happens.

What may it do without asking?Where must a person approve?Specify the action, approver and exact information shown before approval. Configure gates in the tool or workflow too.

4. Test, stop and improve

Which tests must pass?Normal example, missing data, conflicting sources, tool failure, hostile source instruction, duplicate trigger.When should it stop or escalate?Missing access, unclear commitment, exceeded budget, repeated failure or a high-impact decision.

Cost and time limitsSet per-run and monthly limits, retries and maximum steps. Enforce limits in platform settings where supported.What will you measure?Quality, preparation, waiting, review, rework, actual usage and accepted outputs.

Your portable agent brief

Review the brief, then paste it into an approved agent. Connect tools and enforce permissions separately. Keep a copy outside the chat so you can reuse and revise it.

Complete the planner and choose “Create my agent brief.”

Optional: estimate token cost for one run

This is arithmetic using the rates you enter, not a vendor price quote or spending cap. Include every agent call and retry in your token totals. Provider cache-write, tool, hosting and other charges may apply. Enter every value; use 0 only when that cost or usage is deliberately zero. If a rate is unknown, check it before calculating.

The initial numbers are fictional: $0.172 model cost and $10 of human effort, totaling $10.172 before additional costs. Subscriptions have their own limits and billing; this API-style calculation does not replace them.

Make the workclear enough to delegate.

Your portable agent brief

Make the work
clear enough to delegate.