rohitg00 plan-slicing
Lifecycle stage: PLAN
Trigger
Use after a brief/design is accepted and before build.
When not to use
Do not use when this trigger is absent; choose the command or skill that owns the requested state, artifact, and verification gate.
Inputs
- User request or current artifact.
- Known constraints and context.
- Relevant evidence or source links, if available.
- Current Agent Brain state.
Procedure
- State the current state and target artifact.
- Identify missing blockers and ask at most three blocking questions.
- Separate facts, assumptions, hypotheses, and open questions.
- Split broad work into vertical slices small enough to verify independently.
- For each slice, name the user-visible outcome, files likely touched, acceptance checks, and verification command.
- Apply the anti-rationalization table below.
- Produce the required artifact: Implementation Plan.
- Add evidence, risks, decision, and next state.
Anti-Rationalization
| Shortcut | Rebuttal |
|---|---|
| "This is obvious." | Write the assumption and evidence. If you cannot, it is not obvious. |
| "We can do this later." | If the missing step changes the decision, do it now or state the risk. |
| "The user wants speed." | Reduce scope; do not skip the quality bar. |
| "This does not need verification." | Every important claim or behavior needs proof. |
Verification
- Required artifact exists and is named.
- Every slice has acceptance checks and a verification command.
- Facts, assumptions, and open questions are separated.
- Evidence or evidence gaps are explicit.
- Next state is stated.
- Stop conditions are honored.
Output Artifact
Implementation Plan
Use templates/implementation-plan.md. The artifact should be concise, auditable, and include evidence, blockers, and next action so another agent can resume. Each slice should be independently runnable by another agent without private context.
Failure Modes
- Producing advice instead of an artifact.
- Accepting user assumptions without challenge.
- Skipping evidence because the task feels simple.
- Recommending an agent when a simpler system is enough.
- Hiding risks or open questions.
- Listing tasks without acceptance checks or a verification command.
Example
Trigger: a broad plan needs a safe first slice. Action: split work into ordered slices that start with a validator, eval, or failing test where behavior changes. Output artifact: templates/implementation-plan.md with blockers and next action. Verification: each slice names acceptance checks, verification command, rollback, and artifact changed.
A brief requires a new adapter. Plan slices should start with a failing validator or smoke fixture for the adapter contract, then a docs/template update, then local validation and real-runtime smoke evidence. Each slice names acceptance checks, a verification command, rollback, and the artifact it updates.