Skip to main content
Technology & EngineeringEnso Method94 lines

Enso Method Initiative and Phase Planning

Turn an initiative into bounded requirements documents and implementation phases with stable acceptance criteria, dependency contracts, and reviewable handoffs.

Quick Summary3 lines
This SkillDB entry is a self-contained adaptation of the Enso Method by EnsoDynamics.
skilldb get enso-method-skills/initiative-and-phase-planningFull skill: 94 lines
Paste into your CLAUDE.md or agent config

Enso Method Initiative and Phase Planning

This SkillDB entry is a self-contained adaptation of the Enso Method by EnsoDynamics.

Use this workflow when an initiative spans several product outcomes or a single requirements document exceeds one implementation session. Start from the product brief, prior decisions, requirements and design revisions, plus the current implementation. Preserve the outcomes and boundaries already agreed. Produce a plan that lets someone pick up a phase, tell when it is done, and identify consequences if a prior interface changes.

Decide whether to split the initiative or the work

Read the source material and write the initiative's outcome in terms the owner can check. Inventory the named requirements, operating limits, systems, current delivery state, and decisions that still need owners. Give each independent product outcome its own requirements document when it can be accepted using distinct evidence and has a durable boundary with the other outcomes. Keep a single requirements document for one outcome that takes several sessions to build. A frontend/backend division alone usually describes implementation order within one outcome.

Record why each proposed split is useful and who can use the output at its boundary. Test the boundary by having one person describe the inputs and result without reading both implementation proposals. Combine scopes that require repeated private agreements over the same lifecycle or cannot be accepted separately. Put an explicit interface contract between scopes that can proceed in parallel after the contract stabilizes.

Use requirements as the source of truth

Create an initiative brief for the objective, audiences, declared exclusions, dependencies, and acceptance of the overall outcome. Maintain a contract for each requirements document. Use stable IDs for requirements, acceptance criteria, assumptions, and phases. Keep the acceptance language in requirements.md and reference IDs in implementation records. Put concrete choices about state, interfaces, and evidence collection in technical-design.md. Keep provisional decisions, impact, and the confirmation trigger in assumptions.md.

Build a coverage matrix when the brief names several distinct asks. Give each ask one owning requirements document or mark it deferred, excluded, or unresolved with a reason. Identify shared permissions, stores and messages from the system itself and compare that list with the dependencies in the plan. Give each shared capability a producer and name each consumer that requires it. If a consumer can go first, document its interim behavior and who owns switching to the permanent contract.

Use this contract for every requirements document in the initiative.

FieldEntry
Scope ID and accountable roleStable identifier and person or role that can accept the scope
User outcomeActor, initiating event, result, and observable boundary
Requirements and exclusionsRequirement IDs, linked criteria, explicitly excluded work
Inputs and outputsInterface, ownership, data semantics, and failure states
DependenciesPredecessor ID, required artifact or behavior, and verification
AssumptionsIDs of open or resolved decisions and affected criteria
Completion evidenceExpected environments, artifacts, and accountable reviewer

Write a criterion as preconditions, action, and observable result. Add a negative case for an unauthorized actor, invalid state, unavailable dependency, or duplicate action where it matters. State what must remain unchanged after a rejected attempt. A provisional assumption may support authorized, reversible work if the affected choices and recovery cost are clear. Record the evidence behind the assumption, who can confirm it, and the event that forces a decision. Treat product policy, external effects, access, and irreversible migration as approval boundaries. Carry those boundaries into the plan and continue independent preparation where feasible.

Plan and size executable phases

Map each criterion to a phase and identify the check it needs to produce. Start with a thin path that proves the highest-risk seam in an authorized environment. Build any shared foundation inside the first behavior that can exercise it. Size each phase for the full cycle of implementation, evidence, review, and handoff. Use comparable changes in the repository to estimate whether the session can include regression runs and cleanup. Start a new plan with a flat, execution-ordered list. If a phase already referenced elsewhere turns out too large, keep its stable parent ID as a routing record and point downstream dependencies at the child that actually produces the needed artifact.

A phase record includes a stable ID, intent, baseline requirements and design revisions, acceptance IDs, prerequisites and their verification method, exclusions, expected files or artifacts, evidence plan, and next owner. Mark each criterion required for phase exit or contributes evidence only. An early phase can validate a local policy rule without claiming that the operator-visible flow is complete. Put a later check at the real boundary. Reconcile the initiative overview, requirement coverage, and the dependency graph after a split. Resolve cycles by identifying the shared contract or coherent unit instead of declaring dependent work parallel.

Project Returns example

The owner wants authorized support reps to approve full refunds to the original payment for eligible orders. The refund workflow owns the eligibility decision, provider operation, and refund receipt. A separate reporting document would be justified if finance can accept it against a stable refund event contract and reporting has its own delivery outcome. Keep reporting out of the first refund slice when it is outside the approved scope.

RequirementObservable acceptance
R-01An authorized rep approving an eligible paid order receives a receipt with the original payment reference and full refundable amount.
R-02Retrying the same operation ID and request returns the same operation and produces one provider effect. Changed payload with the same operation ID is rejected.
R-03A denied or ineligible request leaves the payment unchanged with no provider effect or committed refund, and returns an appropriate safe result.
R-04A timeout with an uncertain provider outcome leaves a visible pending receipt until authoritative reconciliation.

Record A-01 as a proposed eligibility window of 30 elapsed 24-hour days from the recorded payment timestamp through the recorded request timestamp in UTC, inclusive. Note the origin of the proposal, its owner, eligibility and cutoff fixtures it changes, and the deadline for confirming policy before release. Local design and tests can use a configurable synthetic cutoff. When the owner confirms 45 days and supplies its effective date, update requirements.md, resolve A-01, update affected design and fixtures, and direct any reopened phase to evidence for the new baseline. Keep prior phase receipts labeled with the revision they checked.

PhaseCoverage and dependencyExit evidence
PH-01 Read-only eligibility and permissionsR-03 policy contribution, no provider dependencyAuthorized rep can review a synthetic order and eligibility result at fixed clock boundaries, denied actor receives a safe response, all payment state stays unchanged
PH-02 Authorized refund pathR-01 and R-02, requires PH-01 and approved adapter contractSynthetic or approved sandbox successful refund, duplicate operation, changed-payload rejection, persisted references
PH-03 Timeout and recoveryR-04 and end-to-end R-03, requires observed PH-02 operation stateTimeout fixture, provider inquiry result, stable receipt, exactly one modeled effect, negative checks at the entry point

Hand off with a start condition

Give the implementer the repository or artifact location, exact baseline revision, phase ID, prerequisite evidence, remaining assumption IDs, and the first safe action. For Project Returns, send PH-02 with the passing PH-01 policy report, a synthetic adapter fixture, and the required receipt and duplicate-effect checks. State whether the approved environment can run those checks and which provider interactions require separate authorization.

On phase completion, compare the stated exit with the observed evidence and record any criterion still unverified. After the implementation phases and after any completion work, rerun a whole-requirements delivery audit across all required variants before auditing test strength. A successful phase check can establish its local exit while leaving the overall requirement open. If the owner approves deferring a missing outcome, point to its new phase and preserve the open coverage status. A handoff for blocked work identifies the missing decision, accountable person, and safe tasks that remain available. The plan gives the next owner enough context to continue without reconstructing decisions from chat.

Upstream license

MIT License

Copyright (c) 2026 Doug Kerwin

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OF OR OTHER DEALINGS IN THE SOFTWARE.

Install this skill directly: skilldb add enso-method-skills

Get CLI access →