# Bottleneck Locator

> Find the true bottleneck in a task, system, plan, or workflow. Use when progress is slower than expected, effort is scattered, or teams are optimizing around the real constraint instead of through it.

- Skill: `alubiama/bottleneck-locator` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add alubiama/bottleneck-locator`
- Raw SKILL.md: https://api.skillmd.com/api/skills/alubiama/bottleneck-locator/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: Alubiama (https://skillmd.com/u/alubiama)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/alubiama/bottleneck-locator

---


# Bottleneck Locator

## Goal

Identify the single constraint that most limits progress right now.

This skill does not optimize everything. It finds what matters most so effort stops scattering.

## When To Use

- progress stalls despite effort
- queues, delays, or coordination drag appear
- many issues compete for attention
- local fixes are not improving throughput

## Scope Boundaries

In scope:

- identify the active bottleneck
- distinguish root constraint from symptoms
- recommend the next move against the bottleneck

Out of scope by default:

- full systems maps
- broad redesigns before the constraint is named

## Escalation Conditions

Pause and keep the diagnosis provisional when:

- the blocked outcome is still vague
- the visible delay may be downstream of a deeper unknown
- two constraints appear co-equal and the evidence is weak
- the proposed fix is much broader than the named constraint

If the diagnosis cannot be tightened after one pass, recommend `systems-thinking`, `evidence-gap-detector`, or a smaller validation step.

## Workflow

1. State the blocked outcome.
2. Name the current system boundary being analyzed.
3. List the main stages or dependencies.
4. Find where work, attention, or decisions accumulate.
5. Check whether the apparent bottleneck is primary or downstream.
6. Check whether the wrong system boundary is hiding the real constraint.
7. Name the active constraint.
8. Recommend the smallest intervention that relieves it.

## Output Contract

Always return:

1. `blocked_outcome`
2. `system_boundary`
3. `candidate_constraints`
4. `active_bottleneck`
5. `why_this_is_primary`
6. `misleading_symptoms`
7. `relief_action`
8. `risk_of_misdiagnosis`
9. `recommended_followup`
10. `next_step`

## Guardrails

- Do not name multiple primary bottlenecks unless they truly co-bind.
- Do not optimize around the bottleneck instead of through it.
- If evidence is weak, say the diagnosis is provisional.
- Prefer the smallest bottleneck-relief action that can prove the diagnosis.
- Check whether the apparent bottleneck disappears when the system boundary is redrawn.

## Anti-Patterns

- every problem is the bottleneck
- symptom chasing
- broad optimization with no named constraint