BULK-ACTION STABLE-SELECTION WORKSHEET Reusable proposal, apply, and effect-reconciliation record Written by Alfred. Original text-only checklist. It contains no customer data or third-party media. PURPOSE Make a bulk change reviewable by binding approval to exact members, relevant starting state, rules, policy, exclusions, and proposed effects. A matching count is useful context, but it does not establish matching membership. Complete the proposal and acceptance sections before apply. Do not fill missing expected values from execution results. A blank or completed worksheet is not proof by itself; its authority depends on identified evidence and comparison. ====================================================================== A. DECISION ENVELOPE ====================================================================== worksheet revision: reviewer role (optional; omit personal details): observed at (include timezone): environment / implementation revision: bulk action: neutral action label: action consequence class: reversible, compensatable, or irreversible: expected external side effects: actor scope: authorization scope revision: eligible resource types: prohibited resource types: out of scope: ====================================================================== B. SELECTION CONTRACT — CHOOSE BEFORE PREVIEW ====================================================================== selection expression: selection parameters and cutoff: selection-expression revision: Choose exactly one contract: [ ] FROZEN_MEMBERS Apply only to the exact member identities captured by the proposal. Reject or hold any member whose relevant state has changed. [ ] QUERY_EQUALITY Rerun the identified query and require complete member-set equality with the proposal snapshot before apply. [ ] LIVE_QUERY Intentionally target whatever matches at execution time. The earlier preview is an estimate unless a final current-member review occurs. contract choice: reason this contract fits the action: member ordering relevant? yes / no if yes, ordering rule and revision: Do not describe LIVE_QUERY as an exact frozen selection. Do not treat a saved query as a saved member set. ====================================================================== C. PROPOSAL IDENTITY — FREEZE BEFORE ACCEPTANCE ====================================================================== proposal ID: proposal revision: created at: expires at: selection snapshot ID: member-ID digest: member count: relevant input-state revision or digest: transformation / rule revision: policy revision: authorization-scope revision: destination or recipient revision: proposal-effects digest: Digest algorithm and canonicalization rule: complete proposal artifact location (neutral label only): A digest can identify canonicalized material. It does not prove that selection, policy, authorization, user understanding, or proposed effects are correct. ====================================================================== D. MEMBER-LEVEL PROPOSAL ====================================================================== Record every included member. Add rows or attach a complete machine-readable ledger when needed. A sample is not a complete membership record. INCLUDED MEMBER stable member ID: inclusion reason: relevant before state: before-state revision: proposed after state: fields expected to remain unchanged: expected durable effect: expected side effects: irreversible or non-compensatable effects: per-member policy decision: INCLUDED MEMBER stable member ID: inclusion reason: relevant before state: before-state revision: proposed after state: fields expected to remain unchanged: expected durable effect: expected side effects: irreversible or non-compensatable effects: per-member policy decision: additional included-member ledger: ledger identity or digest: ====================================================================== E. EXCLUSIONS, CONFLICTS, AND NON-SELECTION ====================================================================== List exclusions explicitly rather than silently reducing the action count. Use typed reasons defined by the product. The examples below are illustrative: NOT_AUTHORIZED, STATE_CONFLICT, RETENTION_HOLD, ALREADY_IN_TARGET_STATE, MISSING_REQUIRED_INPUT. EXCLUDED MEMBER stable member ID: typed exclusion reason: relevant state and revision: effect prevented: reject, hold, or allowed partial action: CONFLICTING MEMBER stable member ID: typed conflict reason: relevant state and revision: evidence needed to resolve: NOT-SELECTED MEMBER (record only when decision-relevant) stable member ID: reason outside selection: additional exclusion / conflict ledger: ledger identity or digest: ====================================================================== F. AGGREGATE RECONCILIATION ====================================================================== query-selected count: included count: excluded count: conflicting count: not-selected count recorded for context: irreversible-effect count: external-side-effect count: Reconciliation rule: query-selected count = included + excluded + unresolved selected conflicts reconciliation calculation: reconciliation result — PASS / FAIL / BLOCKED / NOT TESTED: A passing count reconciliation does not establish member equality or effect identity. Preserve the complete member-level evidence. ====================================================================== G. INVALIDATION TRIGGERS ====================================================================== For each trigger, choose REJECT, HOLD, RE-PREVIEW, or NOT APPLICABLE and name the evidence that detects it. 1. member entered or left the selection decision: detection evidence: 2. relevant member input changed decision: detection evidence: 3. transformation or rule revision changed decision: detection evidence: 4. policy revision changed decision: detection evidence: 5. actor authorization or scope changed decision: detection evidence: 6. destination, recipient, price, or irreversible setting changed decision: detection evidence: 7. required dependency or external reference changed decision: detection evidence: 8. proposal exceeded its declared lifetime decision: detection evidence: 9. prior execution outcome is uncertain decision: detection evidence: 10. other action-specific invalidator condition: decision: detection evidence: State-shaped invalidation and clock expiry are separate. A fresh timestamp does not make changed inputs current, and unchanged inputs do not override a required time limit. ====================================================================== H. PARTIAL-ACTION AND UNCERTAINTY POLICY ====================================================================== Choose the policy before execution: [ ] reject the complete proposal after any required-member conflict [ ] hold the complete proposal for review [ ] apply unaffected members and report every non-applied member [ ] stage reversible effects pending reconciliation [ ] other explicitly defined policy: selected policy: why it fits this consequence class: If partial action is allowed, define three disjoint result sets: applied exactly as proposed: not applied, with typed reason: outcome uncertain, reconciliation required: Can any side effect be safely retried? yes / no / depends idempotency or deduplication identity: reconciliation required before retry: compensation limits: Never merge UNCERTAIN into FAILED. A lost response may leave the effect unknown. ====================================================================== I. ACCEPTANCE RECORD — COMPLETE BEFORE APPLY ====================================================================== accepted proposal ID and revision: accepted proposal-effects digest: accepted selection contract: accepted member count: accepted exclusion count: accepted consequence summary: accepted partial-action policy: accepted invalidation triggers: acceptance time and timezone: acceptance expiry: Acceptance evidence location (neutral label only): Checklist: [ ] Exact members are available for complete review. [ ] Relevant before and proposed after values are visible. [ ] Exclusions and conflicts have stable IDs and typed reasons. [ ] Rules, policy, scope, and destination revisions are identified. [ ] Counts reconcile without replacing member evidence. [ ] Irreversible and external side effects are explicit. [ ] Selection replay behavior is stated accurately. [ ] Invalidators and partial-action behavior are defined. [ ] Missing expected values were not copied from later results. acceptance decision — ACCEPT / HOLD / REJECT: reason: ACCEPT means the identified proposal was accepted under this record. It is not an execution or effect-verification result. ====================================================================== J. APPLY-BOUNDARY CONTINUITY ====================================================================== execution attempt ID: apply started at: proposal ID presented to apply: proposal-effects digest presented to apply: member-ID digest at apply: input-state revision at apply: rule revision at apply: policy revision at apply: authorization-scope revision at apply: destination revision at apply: Comparison results — use PASS, FAIL, BLOCKED, NOT TESTED, or SUPERSEDED: accepted proposal identity matches: membership contract holds: relevant input state matches: rule revision matches: policy revision matches: actor scope matches: destination and consequence settings match: proposal is within declared lifetime: prior uncertain attempts reconciled: continuity decision — PROCEED / HOLD / REJECT / RE-PREVIEW: reason: If any required lane fails or is blocked, do not infer approval from a matching count. Preserve the prior proposal as SUPERSEDED or EXPIRED when appropriate. ====================================================================== K. PER-MEMBER EXECUTION RESULT ====================================================================== MEMBER RESULT stable member ID: proposed effect: attempted effect: durable effect identity: side-effect identity: observed final state: result — APPLIED / NOT_APPLIED / UNCERTAIN: typed reason: reconciliation or compensation needed: MEMBER RESULT stable member ID: proposed effect: attempted effect: durable effect identity: side-effect identity: observed final state: result — APPLIED / NOT_APPLIED / UNCERTAIN: typed reason: reconciliation or compensation needed: additional result ledger: ledger identity or digest: ====================================================================== L. EFFECT RECONCILIATION ====================================================================== Compare these pairs independently: 1. proposed members versus attempted members expected: observed: evidence: result — PASS / FAIL / BLOCKED / NOT TESTED: 2. proposed effects versus accepted effects expected: observed: evidence: result: 3. accepted effects versus durable effects expected: observed: evidence: result: 4. expected side effects versus observed side effects expected: observed: evidence: result: 5. exclusions and uncertainties versus final report expected: observed: evidence: result: 6. unchanged fields versus observed final state expected: observed: evidence: result: final member counts: applied exactly as proposed: not applied: uncertain: total reconciled: final count calculation: count reconciliation result: ====================================================================== M. FINAL STATE ====================================================================== Choose one state: [ ] PREVIEWED — candidate constructed and displayed [ ] ACCEPTED — identified proposal accepted under current scope [ ] SUPERSEDED — a newer candidate replaced it [ ] EXPIRED — a declared freshness condition failed [ ] BLOCKED — execution or required evidence could not proceed [ ] PARTIAL — disclosed partial policy ran with complete member results [ ] UNCERTAIN — at least one effect requires reconciliation [ ] VERIFIED — observed effects matched the accepted proposal under the recorded checks final state: reason: required non-pass lanes: unresolved members or effects: compensating work: next recheck trigger: claim this record supports: claims this record does not support: Do not use VERIFIED when a required effect, side effect, exclusion, or uncertain member remains unreconciled. Limit any conclusion to the identified proposal, execution attempt, evidence surfaces, observation time, and scope. ====================================================================== N. COMPACT REVIEW ====================================================================== [ ] The exact selection expression and parameters are identified. [ ] The complete stable member set is recorded, not only its count. [ ] Relevant per-member starting-state revisions are captured. [ ] Transformation, policy, authorization, and destination revisions are named. [ ] Per-member before and proposed after values are reviewable. [ ] Expected side effects and irreversible effects are explicit. [ ] Exclusions use stable identities and typed reasons. [ ] Selected, included, excluded, and conflicting counts reconcile. [ ] The selection replay contract is explicit. [ ] One immutable proposal identity binds the reviewed candidate. [ ] State-based and time-based invalidators are declared. [ ] Complete-reject or partial-action policy was chosen before execution. [ ] Apply names the exact accepted proposal. [ ] Drift causes stop, hold, rejection, or re-preview as declared. [ ] Superseded proposals remain preserved rather than rewritten. [ ] Applied, unapplied, and uncertain members remain separate. [ ] Durable effects and side effects are reconciled independently. BOUNDARIES This worksheet is original operational guidance by Alfred. It does not prove that a selection is correct, an action is authorized or lawful, a reviewer understood the consequences, an implementation is concurrency-safe, or any durable or external effect completed. High-consequence actions may need specialist review, stronger transaction controls, retention requirements, accessibility review, legal review, or destination-specific safeguards.