| name | stcc - Short-Term Cluster Completion |
| description | Identify partially-built clusters of work and determine what's needed to finish them. Prioritize by effort-to-complete. |
| output | {"format":"prose"} |
Short-Term Cluster Completion
Input: $ARGUMENTS
Step 1: Map Related Items
Identify clusters — groups of related work that belong together logically.
CLUSTER MAP:
1. [Cluster name]: [Brief description of what this group of work does together]
- [Item A]: [exists / partial / missing]
- [Item B]: [exists / partial / missing]
- [Item C]: [exists / partial / missing]
2. [Cluster name]: [Brief description]
- [Item A]: [exists / partial / missing]
...
Rules:
- A cluster is a set of things that only deliver full value when all are present
- Look for natural groupings: features, workflows, documentation sets, API surfaces
- Don't force unrelated items into clusters
Step 2: Assess Completion Percentage
For each cluster, estimate how done it is.
COMPLETION ASSESSMENT:
1. [Cluster name]: [X]% complete
- Done: [what's finished]
- Partial: [what's started but incomplete]
- Missing: [what hasn't been started]
2. [Cluster name]: [X]% complete
...
Rules:
- Be honest about percentages — 80% done often means 50% of the value is missing
- "Partial" means some work exists but it doesn't function independently
- A cluster at 95% is more valuable to finish than one at 30%
Step 3: Identify Completion Requirements
For each cluster, specify exactly what's needed to reach 100%.
COMPLETION REQUIREMENTS:
1. [Cluster name]:
- [ ] [Specific task 1] — effort: [hours/days]
- [ ] [Specific task 2] — effort: [hours/days]
- [ ] [Specific task 3] — effort: [hours/days]
Total remaining effort: [estimate]
2. [Cluster name]:
...
Rules:
- Tasks must be specific and actionable, not vague
- "Polish the UI" is not a task; "Add error states to the login form" is
- Include only what's needed for completion, not enhancement
Step 4: Prioritize by Effort-to-Complete
Rank clusters by the ratio of value delivered to effort remaining.
PRIORITY RANKING:
| Rank | Cluster | % Done | Effort Left | Value When Complete | Priority Score |
|------|---------|--------|-------------|---------------------|----------------|
| 1 | [name] | [X]% | [time] | [HIGH/MED/LOW] | [score] |
| 2 | [name] | [X]% | [time] | [HIGH/MED/LOW] | [score] |
...
Rules:
- Priority score = (completion % * value) / effort remaining
- High-completion + high-value + low-effort = finish first
- Low-completion + low-value + high-effort = finish last or abandon
Step 5: Recommend Finishing Order
Produce a concrete plan.
RECOMMENDED ORDER:
1. FINISH NOW: [Cluster] — [why: nearly done, high value]
2. FINISH NEXT: [Cluster] — [why]
3. FINISH LATER: [Cluster] — [why]
4. CONSIDER ABANDONING: [Cluster] — [why: too much effort for too little value]
Rules:
- "Finish now" should be completable within days
- "Consider abandoning" is legitimate — sunk cost is not a reason to continue
- If two clusters share dependencies, note that and sequence accordingly
Integration
Use with:
/immg -> Fix urgent gaps before completing clusters/de -> Turn finishing plan into a structured delivery/cba -> If unsure whether to finish or abandon, do a cost-benefit analysis/mtnw -> After completing clusters, look for new domains to expand into
