AI agent
Plan mode
For complex tasks: agent drafts a plan → you approve → execution begins.
What plan mode is for
Default mode: prompt in, work starts immediately — fine for small tasks. Plan mode forces the agent to pause and write a structured plan first: goal, steps, tools per step, likely failure points. Execution starts only after you Approve.
For irreversible work (deleting files, hitting APIs, writing production databases) this is the key guardrail. Thirty seconds of review prevents thirty minutes of rollback.
How to trigger
- Say "Use plan mode" or "plan first" in your prompt
- Or click the plan toggle next to the input
- Use the
/planslash command - The agent outputs the plan → you Approve / Edit / Reject
What a plan looks like
Plan: migrate notes/ from old format to v2
Steps:
1. grep notes/**/*.md for "::deprecated::" markers
tools: grep
2. For each match, read the file and parse front-matter
tools: Read (parallel)
3. Rewrite front-matter into v2 schema, keep body
tools: Edit (parallel, batched 10)
4. Re-validate by running scripts/validate-v2.sh
tools: Bash
5. Print a summary table of migrated / skipped / failed
Risk:
- Step 3 rewrites files. I will skip files modified in the
last hour to avoid clobbering active edits.
- Step 4 requires Bash permission for scripts/.
Estimated tool calls: ~120
Estimated cost: $0.18 with sonnetWhen it pays off
- Cross-doc / cross-table edits
- External API calls involved
- Irreversible ops (rm / write APIs)
- Over ~20 tool calls expected — review saves cost
- Team vaults — others can audit the decision trail
Plan vs Goal mode
Plan = "write the whole plan once, then execute" — fits well-scoped work. Goal mode is the opposite: you state a long-horizon objective and the agent autonomously decomposes, executes, reflects, and decomposes again until it lands or you stop it — fits open-ended jobs like "research three competitors and produce a comparison".
They stack: run Goal mode for the long arc, but require Plan review at each sub-phase.