soloscrum-define-story-points
SP definition and estimation criteria.
Two-Level SP Structure
soloscrum uses SP estimation at two levels: Issue level (PO layer) and subtask level (Dev layer).
Why Two Levels?
Tasks can be decomposed infinitely. The scope of a root task is entirely flexible — "implement the app", "build the auth feature", "build the login screen" are all valid roots. Because of this, a layer is needed before breakdown to ask: "is this Issue actually a manageable size?" Entering breakdown with an oversized Issue causes subtask explosion and makes tracking unmanageable.
The lightweight estimate in the PO layer serves as this entry gate.
Issue SP (PO Layer)
- Purpose: Determine whether the Issue is a manageable size
- Owner:
soloscrum-po (during /soloscrum:refine)
- Precision: Rough is fine. Estimate scope and uncertainty from intuition without enumerating every file
- Threshold: Trigger
suggest_split when SP > 5
- Registered in tracker: No — this is a size-check value only, never written to any tracker storage
When Issue SP exceeds the threshold, treat as a mis-scope smell — the Issue likely bundles multiple intents — and return to /soloscrum:refine for Issue split per soloscrum-define-issue-size, then re-estimate each split Issue. (A coherent single intent that genuinely needs > 5 SP worth of work is delivered through /soloscrum:breakdown Subtask slices, not by Issue split — see soloscrum-define-issue-size for the distinction.)
Subtask SP (Dev Layer)
- Purpose: The actual value registered in the tracker, used for planning and progress tracking
- Owner:
soloscrum-dev (during /soloscrum:breakdown)
- Precision: Calculate after carefully reviewing the Subtask's slice scope (the "what" sentence and any checklist items per
soloscrum-define-issue-format's Subtask Body section; Subtasks do not carry their own AC), affected files, and novelty
- Registered in tracker: Yes — storage location depends on the active profile (per
soloscrum-define-tracker-profile):
github-only → GH Projects v2 SP Number field
linear+github → Linear subtask estimate field
- Set via
soloscrum-tracker-{github|linear}-set-sp (typically inlined at subtask creation)
SP Table
SP is a size class anchored on scope x uncertainty — not a time unit. The "Calibration" column lists observed agent runtimes / token budgets so the scale can be sanity-checked, but those numbers are not the unit and must not be used as the primary input.
| SP |
Scope |
Uncertainty |
Calibration (observed, not the unit) |
| 1 |
1 file / 1 concern |
All decisions known (AC fully prescribes the change) |
~30K-100K tokens, agent ~5-10 min |
| 2 |
2-3 files / single skill area |
1 minor decision |
~100K-200K tokens, agent ~10-20 min |
| 3 |
Single subsystem cross-cut |
1-2 design decisions |
~200K-500K tokens, agent ~20-45 min |
| 5 |
Multi-subsystem cross-cut |
Multiple design decisions |
~500K-1M tokens, agent ~45 min-2h |
| >5 |
(mis-scope signal) |
(mis-scope signal) |
Do not register — treat as "this Issue likely bundles multiple intents," return to /soloscrum:refine and split per soloscrum-define-issue-size, re-estimate each split Issue. A coherent single intent that genuinely needs >5 SP worth of work is delivered through /soloscrum:breakdown Subtask slices rather than by Issue split. |
The SP table is shared by both Issue SP and subtask SP.
Estimation Procedure
Issue SP (PO Layer)
- Read the Goal and AC to identify scope: how many subsystems does the change touch, and how many concerns are bundled into the Issue?
- Identify uncertainty: how many open design decisions remain after AC? Is any part novel (no precedent in this repo)?
- Map (scope, uncertainty) to the SP table — both axes must fit; pick the higher row when in doubt
- If SP > 5, treat as a mis-scope smell — return to
/soloscrum:refine and split per soloscrum-define-issue-size. A coherent single intent that genuinely needs > 5 SP worth of work goes to /soloscrum:breakdown (delivery slicing) instead, not Issue split.
Subtask SP (Dev Layer)
- Count affected files and concerns from the Subtask's slice scope (its "what" and checklist per
soloscrum-define-issue-format's Subtask Body — Subtasks do not have their own AC; refer to the parent Issue's AC for the broader contract); locate the corresponding Scope column
- Count remaining design decisions (anything the parent's AC and the Subtask's "what" do not prescribe verbatim) and judge novelty against this repo's existing patterns; locate the corresponding Uncertainty column
- Pick the higher row if scope and uncertainty disagree
- If estimate exceeds SP 5, a missed Issue split is likely — confirm with user
Notes
- SP measures scope and uncertainty, not time. This is the primary basis for the scale; the Calibration column is observation only, used to sanity-check the chosen row.
- Time anchors were intentionally removed: model speed shifts release-to-release (today's "2 hours" becomes next month's "20 minutes"), agents run in parallel which warps wall-clock comparisons, and human time vs AI agent time do not map linearly — pinning SP to a clock unit made the scale unstable.
- Past Issues with established SP can be used to triangulate: if the current Issue is "obviously bigger than #X (SP 2) and obviously smaller than #Y (SP 5)", SP 3 is the answer regardless of any timing intuition.
- Solo development — do not factor in team velocity.
1---2name: soloscrum-define-story-points3description: "Reference: story point scale and estimation criteria. SP is anchored on scope x uncertainty (size class, not a time unit); SP 5 is the upper bound and Issues estimated above it are mis-scope smells signalling likely intent bundling — return to /soloscrum:refine for Issue split per soloscrum-define-issue-size."4user-invocable: false5metadata:6 source: https://skillmd.com/skills/mew-ton/soloscrum-define-story-points7---89# soloscrum-define-story-points1011SP definition and estimation criteria.1213## Two-Level SP Structure1415soloscrum uses SP estimation at two levels: **Issue level (PO layer)** and **subtask level (Dev layer)**.1617### Why Two Levels?1819Tasks can be decomposed infinitely. The scope of a root task is entirely flexible — "implement the app", "build the auth feature", "build the login screen" are all valid roots. Because of this, a layer is needed before breakdown to ask: "is this Issue actually a manageable size?" Entering breakdown with an oversized Issue causes subtask explosion and makes tracking unmanageable.2021The lightweight estimate in the PO layer serves as this **entry gate**.2223### Issue SP (PO Layer)2425- **Purpose**: Determine whether the Issue is a manageable size26- **Owner**: `soloscrum-po` (during `/soloscrum:refine`)27- **Precision**: Rough is fine. Estimate scope and uncertainty from intuition without enumerating every file28- **Threshold**: Trigger `suggest_split` when SP > 529- **Registered in tracker**: No — this is a size-check value only, never written to any tracker storage3031When Issue SP exceeds the threshold, treat as a mis-scope smell — the Issue likely bundles multiple intents — and return to `/soloscrum:refine` for Issue split per `soloscrum-define-issue-size`, then re-estimate each split Issue. (A coherent single intent that genuinely needs > 5 SP worth of work is delivered through `/soloscrum:breakdown` Subtask slices, not by Issue split — see `soloscrum-define-issue-size` for the distinction.)3233### Subtask SP (Dev Layer)3435- **Purpose**: The actual value registered in the tracker, used for planning and progress tracking36- **Owner**: `soloscrum-dev` (during `/soloscrum:breakdown`)37- **Precision**: Calculate after carefully reviewing the Subtask's slice scope (the "what" sentence and any checklist items per `soloscrum-define-issue-format`'s Subtask Body section; Subtasks do not carry their own AC), affected files, and novelty38- **Registered in tracker**: Yes — storage location depends on the active profile (per `soloscrum-define-tracker-profile`):39 - `github-only` → GH Projects v2 `SP` Number field40 - `linear+github` → Linear subtask `estimate` field41- Set via `soloscrum-tracker-{github|linear}-set-sp` (typically inlined at subtask creation)4243---4445## SP Table4647SP is a **size class anchored on scope x uncertainty** — not a time unit. The "Calibration" column lists observed agent runtimes / token budgets so the scale can be sanity-checked, but those numbers are not the unit and must not be used as the primary input.4849| SP | Scope | Uncertainty | Calibration (observed, not the unit) |50|----|-------|-------------|---|51| 1 | 1 file / 1 concern | All decisions known (AC fully prescribes the change) | ~30K-100K tokens, agent ~5-10 min |52| 2 | 2-3 files / single skill area | 1 minor decision | ~100K-200K tokens, agent ~10-20 min |53| 3 | Single subsystem cross-cut | 1-2 design decisions | ~200K-500K tokens, agent ~20-45 min |54| 5 | Multi-subsystem cross-cut | Multiple design decisions | ~500K-1M tokens, agent ~45 min-2h |55| >5 | (mis-scope signal) | (mis-scope signal) | Do not register — treat as "this Issue likely bundles multiple intents," return to `/soloscrum:refine` and split per `soloscrum-define-issue-size`, re-estimate each split Issue. A coherent single intent that genuinely needs >5 SP worth of work is delivered through `/soloscrum:breakdown` Subtask slices rather than by Issue split. |5657The SP table is shared by both Issue SP and subtask SP.5859## Estimation Procedure6061### Issue SP (PO Layer)62631. Read the Goal and AC to identify scope: how many subsystems does the change touch, and how many concerns are bundled into the Issue?642. Identify uncertainty: how many open design decisions remain after AC? Is any part novel (no precedent in this repo)?653. Map (scope, uncertainty) to the SP table — both axes must fit; pick the higher row when in doubt664. If SP > 5, treat as a mis-scope smell — return to `/soloscrum:refine` and split per `soloscrum-define-issue-size`. A coherent single intent that genuinely needs > 5 SP worth of work goes to `/soloscrum:breakdown` (delivery slicing) instead, not Issue split.6768### Subtask SP (Dev Layer)69701. Count affected files and concerns from the Subtask's slice scope (its "what" and checklist per `soloscrum-define-issue-format`'s Subtask Body — Subtasks do not have their own AC; refer to the parent Issue's AC for the broader contract); locate the corresponding Scope column712. Count remaining design decisions (anything the parent's AC and the Subtask's "what" do not prescribe verbatim) and judge novelty against this repo's existing patterns; locate the corresponding Uncertainty column723. Pick the higher row if scope and uncertainty disagree734. If estimate exceeds SP 5, a missed Issue split is likely — confirm with user7475## Notes7677- **SP measures scope and uncertainty, not time.** This is the primary basis for the scale; the Calibration column is observation only, used to sanity-check the chosen row.78- Time anchors were intentionally removed: model speed shifts release-to-release (today's "2 hours" becomes next month's "20 minutes"), agents run in parallel which warps wall-clock comparisons, and human time vs AI agent time do not map linearly — pinning SP to a clock unit made the scale unstable.79- Past Issues with established SP can be used to triangulate: if the current Issue is "obviously bigger than #X (SP 2) and obviously smaller than #Y (SP 5)", SP 3 is the answer regardless of any timing intuition.80- Solo development — do not factor in team velocity.