problem-solver

Name: problem-solver
Author: Rune-kit

80stars21forks

How to install problem-solver?

rune-kit/rune/problem-solver

$curl -o .claude/agents/problem-solver.md https://raw.githubusercontent.com/Rune-kit/rune/HEAD/agents/problem-solver.md

Installs into the current project.

Summary

Structured reasoning — applies McKinsey-grade frameworks (5 Whys, Fishbone, First Principles, SCAMPER) with mandatory bias detection. Use for complex multi-factor problems, not simple linear analysis.

Files · 1

View on GitHub

agents/problem-solver.md

1	# problem-solver
2
3	## Purpose
4
5	Structured reasoning utility for problems that resist straightforward analysis. Receives a problem statement, detects cognitive biases, selects the appropriate analytical framework, applies it step-by-step with evidence, and returns ranked solutions with a communication structure. Stateless — no memory between calls.
6
7	Inspired by McKinsey problem-solving methodology and cognitive science research on decision-making errors.
8
9	## Calls (outbound)
10
11	- `council` (L3): Step 6.5 — decorrelated judgment on whether a high-stakes framework conclusion holds, narrow trigger only (documented L3→L3 coordination — see Step 6.5)
12
13	## Called By (inbound)
14
15	- `debug` (L2): complex bugs that resist standard debugging
16	- `brainstorm` (L2): structured frameworks for creative exploration
17	- `plan` (L2): complex architecture decisions with many trade-offs
18	- `ba` (L2): requirement analysis when scope is ambiguous
19
20	## Execution
21
22	### Input
23
24	```
25	problem: string — clear statement of the problem to analyze
26	context: string — (optional) relevant background, constraints, symptoms observed
27	goal: string — (optional) desired outcome or success criteria
28	mode: string — (optional) "analyze" \| "decide" \| "decompose" \| "communicate"
29	```
30
31	### Step 1 — Receive and Classify
32
33	Read the `problem` and `context` inputs. Restate the problem in one sentence to confirm understanding.
34
35	Classify the problem type:
36
37	\| Type \| Signal Words \| Primary Approach \|
38	\|------\|-------------\|-----------------\|
39	\| Root cause / diagnostic \| "why", "broken", "failing", "declining" \| 5 Whys, Fishbone, Root Cause \|
40	\| Decision / choice \| "should I", "choose", "compare", "vs" \| Decision Frameworks (Step 3b) \|
41	\| Decomposition \| "break down", "understand", "structure" \| Decomposition Methods (Step 3c) \|
42	\| Creative / stuck \| "stuck", "no ideas", "exhausted options" \| SCAMPER, Collision-Zone, Inversion \|
43	\| Architecture / scale \| "design", "architecture", "will it scale" \| First Principles, Scale Game \|
44
45	### Step 1.5 — Domain Classification (Cynefin)
46
47	Before selecting a framework, classify the problem's complexity domain. This determines HOW MUCH analysis is warranted and WHICH class of frameworks applies.
48
49	\| Domain \| Signal \| Framework Class \| Analysis Depth \|
50	\|--------\|--------\|----------------\|----------------\|
51	\| Clear (obvious) \| Best practice exists, cause-effect obvious, "just do X" \| Direct action — no framework needed \| Minimal — act immediately \|
52	\| Complicated (expert analysis) \| Cause-effect discoverable through analysis, multiple right answers exist \| Analytical frameworks (5 Whys, Fishbone, SWOT, Weighted Matrix) \| Moderate — structured analysis \|
53	\| Complex (emergent) \| Cause-effect only visible in retrospect, no right answer — only better probes \| Probe-sense-respond (Pre-Mortem, Systems Map, Sensitivity Analysis, PESTLE) \| Deep — experiment and iterate \|
54	\| Chaotic (crisis) \| No cause-effect, need to stabilize first \| Act-sense-respond — triage, then analyze \| Immediate — stabilize before analyzing \|
55	\| Confused (don't know which domain) \| Can't classify → decompose until sub-problems land in a known domain \| Decomposition first (Issue Tree, MECE) → re-classify each branch \| Meta — decompose then classify \|
56
57	Output: State the domain and justify in one sentence. If Confused, decompose before proceeding.
58
59	Why this matters: Applying Complicated-domain tools (deep analysis) to a Clear problem wastes effort. Applying Clear-domain tools ("just do X") to a Complex problem creates false confidence. Match the tool to the terrain.
60
61	### Step 2 — Bias Check (ALWAYS RUN)
62
63	<HARD-GATE>
64	NEVER skip bias detection. Every problem has biases — explicitly address them.
65	This is the #1 value-add from structured reasoning. Without it, solutions are just dressed-up gut feelings.
66	</HARD-GATE>
67
68	Two catalogs run here, not one: the biases of the people and organisation in the problem, and the failure modes of the model doing the analysis. Skipping the second is how a rigorous-looking analysis reaches a confidently wrong conclusion.
69
70	#### Human and organisational biases
71
72	Scan the problem statement and context for bias indicators. Check the top 6 most dangerous biases:
73
74	\| Bias \| Detection Question \| Debiasing Strategy \|
75	\|------\|-------------------\|-------------------\|
76	\| Confirmation Bias \| Have we actively sought evidence AGAINST our preferred option? Are we explaining away contradictory data? \| Assign devil's advocate. Explicitly seek disconfirming evidenc

People who liked this also viewed

6 picks

Type

problem-solver

80stars21forks

View on GitHub

How to install problem-solver?

rune-kit/rune/problem-solver

$curl -o .claude/agents/problem-solver.md https://raw.githubusercontent.com/Rune-kit/rune/HEAD/agents/problem-solver.md

Installs into the current project.

Summary

Files · 1

View on GitHub

agents/problem-solver.md

1	# problem-solver
2
3	## Purpose
4
5	Structured reasoning utility for problems that resist straightforward analysis. Receives a problem statement, detects cognitive biases, selects the appropriate analytical framework, applies it step-by-step with evidence, and returns ranked solutions with a communication structure. Stateless — no memory between calls.
6
7	Inspired by McKinsey problem-solving methodology and cognitive science research on decision-making errors.
8
9	## Calls (outbound)
10
11	- `council` (L3): Step 6.5 — decorrelated judgment on whether a high-stakes framework conclusion holds, narrow trigger only (documented L3→L3 coordination — see Step 6.5)
12
13	## Called By (inbound)
14
15	- `debug` (L2): complex bugs that resist standard debugging
16	- `brainstorm` (L2): structured frameworks for creative exploration
17	- `plan` (L2): complex architecture decisions with many trade-offs
18	- `ba` (L2): requirement analysis when scope is ambiguous
19
20	## Execution
21
22	### Input
23
24	```
25	problem: string — clear statement of the problem to analyze
26	context: string — (optional) relevant background, constraints, symptoms observed
27	goal: string — (optional) desired outcome or success criteria
28	mode: string — (optional) "analyze" \| "decide" \| "decompose" \| "communicate"
29	```
30
31	### Step 1 — Receive and Classify
32
33	Read the `problem` and `context` inputs. Restate the problem in one sentence to confirm understanding.
34
35	Classify the problem type:
36
37	\| Type \| Signal Words \| Primary Approach \|
38	\|------\|-------------\|-----------------\|
39	\| Root cause / diagnostic \| "why", "broken", "failing", "declining" \| 5 Whys, Fishbone, Root Cause \|
40	\| Decision / choice \| "should I", "choose", "compare", "vs" \| Decision Frameworks (Step 3b) \|
41	\| Decomposition \| "break down", "understand", "structure" \| Decomposition Methods (Step 3c) \|
42	\| Creative / stuck \| "stuck", "no ideas", "exhausted options" \| SCAMPER, Collision-Zone, Inversion \|
43	\| Architecture / scale \| "design", "architecture", "will it scale" \| First Principles, Scale Game \|
44
45	### Step 1.5 — Domain Classification (Cynefin)
46
47	Before selecting a framework, classify the problem's complexity domain. This determines HOW MUCH analysis is warranted and WHICH class of frameworks applies.
48
49	\| Domain \| Signal \| Framework Class \| Analysis Depth \|
50	\|--------\|--------\|----------------\|----------------\|
51	\| Clear (obvious) \| Best practice exists, cause-effect obvious, "just do X" \| Direct action — no framework needed \| Minimal — act immediately \|
52	\| Complicated (expert analysis) \| Cause-effect discoverable through analysis, multiple right answers exist \| Analytical frameworks (5 Whys, Fishbone, SWOT, Weighted Matrix) \| Moderate — structured analysis \|
53	\| Complex (emergent) \| Cause-effect only visible in retrospect, no right answer — only better probes \| Probe-sense-respond (Pre-Mortem, Systems Map, Sensitivity Analysis, PESTLE) \| Deep — experiment and iterate \|
54	\| Chaotic (crisis) \| No cause-effect, need to stabilize first \| Act-sense-respond — triage, then analyze \| Immediate — stabilize before analyzing \|
55	\| Confused (don't know which domain) \| Can't classify → decompose until sub-problems land in a known domain \| Decomposition first (Issue Tree, MECE) → re-classify each branch \| Meta — decompose then classify \|
56
57	Output: State the domain and justify in one sentence. If Confused, decompose before proceeding.
58
59	Why this matters: Applying Complicated-domain tools (deep analysis) to a Clear problem wastes effort. Applying Clear-domain tools ("just do X") to a Complex problem creates false confidence. Match the tool to the terrain.
60
61	### Step 2 — Bias Check (ALWAYS RUN)
62
63	<HARD-GATE>
64	NEVER skip bias detection. Every problem has biases — explicitly address them.
65	This is the #1 value-add from structured reasoning. Without it, solutions are just dressed-up gut feelings.
66	</HARD-GATE>
67
68	Two catalogs run here, not one: the biases of the people and organisation in the problem, and the failure modes of the model doing the analysis. Skipping the second is how a rigorous-looking analysis reaches a confidently wrong conclusion.
69
70	#### Human and organisational biases
71
72	Scan the problem statement and context for bias indicators. Check the top 6 most dangerous biases:
73
74	\| Bias \| Detection Question \| Debiasing Strategy \|
75	\|------\|-------------------\|-------------------\|
76	\| Confirmation Bias \| Have we actively sought evidence AGAINST our preferred option? Are we explaining away contradictory data? \| Assign devil's advocate. Explicitly seek disconfirming evidenc

People who liked this also viewed

6 picks

Type