Verify claims before accepting them
- 1
Ask for evidence, not assurance.
“Tests pass” and “done” are claims. Ask for the command, the output and the exit code, for every claim.
- 2
Ask “what did you not check?”
Agents report what they did, not what they skipped. If an agent finds no PR comments through the API, ask which sources it queried.
- 3
Demand a negative-result report.
“Found zero X” must come with the endpoints or paths searched. Make “zero results, here is where I looked” mandatory.
- 4
Test the fix, not the agent’s story about it.
Re-run the failing case yourself, or ask for a failing test written first. A diagnosed fix that was never committed or verified is not a fix.
- 5
Check the diff, not the summary.
Read
git diff --statbefore accepting. A summary says what the agent intended; the diff shows what it did.
Control scope
- 6
Ban blanket staging.
git add -Acan sweep agent worktrees into a PR. Require explicit paths and enforce it with a PreToolUse hook, not a polite request. - 7
State the write boundary up front.
“Only touch files under X” prevents stray planning files. Agents expand scope unless fenced.
- 8
Ask for the plan before any edit on more than about 3 files.
Review it for what is missing (migrations, callers, docs, tests), not just for what is present.
- 9
Challenge the size.
If the diff exceeds your PR cap, ask for a split proposal before work continues. Oversized PRs happen because nobody challenged size early.
- 10
Stop scope creep with a “why” test.
For each changed file ask “which requirement needs this?” Revert what cannot answer.
Question the decisions
- 11
Ask for alternatives and when each wins.
“What are two other approaches and when would you pick them?” Agents default to the first plausible one.
- 12
Ask what breaks.
“How does this fail in production?” surfaces thresholds, race conditions and rollback gaps the agent glossed over.
- 13
Challenge unexplained numbers.
Rollout steps, error thresholds and timeouts with no rationale are guesses. Insist on a source, a derivation, or the label “starting point”.
- 14
Push on tool defaults.
Agents bake in popular tools as if they were the principle. Ask “what is the requirement, and is this tool one example?”
- 15
Check for hallucinated specifics.
Verify that versions, flags, API fields and file paths exist, with
--help, a doc lookup orls. Fluent wrongness is the agent’s signature error.
Manage the process
- 16
Cut loops at two repeats.
If the agent says the same thing twice, it is stuck. Interrupt and change the instruction, the hook or the state; do not wait it out.
- 17
Fix the system, not the output.
When a correction repeats (BSD
sed,git add -A, stray plan files), put it in CLAUDE.md or a hook. Correcting the same mistake three times in chat is a process failure. - 18
Separate permission from capability.
When a classifier blocks a destructive action, ask for the exact command block to run yourself. Do not let the agent route around the block.
- 19
Give intent with terse commands.
A bare “plan” produces nothing. Add scope, inputs and done criteria: what, where, and how you will know it is finished.
- 20
Make the agent state assumptions and confidence.
Ask “what did you assume, and what is least certain?” before big actions. Compacted sessions lose context, so verify state (
git status, branch, open PRs) instead of trusting the summary.
What people typically miss
- Accepting fluent output as verified output. Confidence in tone says nothing about correctness.
- Reviewing the answer but not the omissions. What the agent did not mention is where the bugs are.
- Pushing back on code but not on scope or process. Most damage is a wrong approach, size or boundary, not a wrong line.
- Treating one-off corrections as sufficient. If you correct it twice, encode it.
- Skipping the independent check. Run the test, read the diff, query the API yourself. The agent cannot grade its own work.
Interview framing
Describe pushback as a loop: bound the task, demand evidence, check the diff, probe alternatives and failure modes, then encode the lesson in a hook or standard. Anchor each step in a real example: a missing-comments case, a blanket git add -A incident, or a CI check that replaced a judgment call with an automated rule.