This diagram maps the complete ConceptOps run from a rough request to a verified, usable next step. It follows the behavior defined in conceptops/SKILL.md and its reference documents.
flowchart TD
A([Request]) --> A1{Does it need discovery, evidence,<br/>validation, a decision, or artifact selection?}
A1 -->|No| A2["Follow the host and project instructions<br/>Execute the clear, well-scoped task directly"]
A2 --> A3([No ConceptOps run files])
A1 -->|Yes| B{Work language explicit?}
B -->|No| B1[Ask only for the work language]
B1 --> B2([Wait for answer])
B2 --> C
B -->|Yes| C["Set the run-wide language contract<br/>Chat · questions · subagents · artifacts · visuals"]
C --> C1[Explain ConceptOps in two plain sentences]
C1 --> D{Task supplied?}
D -->|No| D1["Ask for the task<br/>A short or messy fragment is enough"]
D1 --> D2([Wait for task])
D2 --> E
D -->|Yes| E[Preserve intent and normalize the request]
subgraph CONTEXT["1 · BUILD THE CONTEXT"]
E --> F[Classify the task mode]
F --> F1["New idea"]
F --> F1B["Partial or rough idea"]
F --> F2["Existing-project change"]
F --> F3["Replacement or migration"]
F --> F4["Research or technology decision"]
F --> F5["Communication, brand, or launch"]
F1 --> G
F1B --> G
F3 --> G
F4 --> G
F5 --> G
F2 --> P1[Resolve the exact repository and applicable instructions]
P1 --> P2["Trace the affected user, UI, service, data,<br/>authorization, and integration flow"]
P2 --> P3["Find reusable helpers, patterns, components,<br/>dependencies, assets, and tests"]
P3 --> G
G["Build a compact context snapshot<br/>Use: user input → project evidence → current sources → labeled assumptions"]
G --> H{Material ambiguity or trust boundary?}
H -->|Yes| H1["Ask the smallest blocking question<br/>No more than three before a useful result<br/>Money · medical/legal · minors · sensitive data<br/>physical action · production · public deployment"]
H1 --> H2([Wait for answer])
H2 --> I
H -->|No| I
I["Separate facts, estimates, hypotheses,<br/>assumptions, opinions, and unknowns"]
I --> J[Draft the task brief]
end
subgraph OUTPUTS["2 · SELECT OUTPUTS AND PREPARE THE RUN"]
J --> K[Recommend the smallest useful component set]
K --> L{Language, task, and outputs already explicit?}
L -->|Yes| L1[Echo the interpreted configuration and continue]
L -->|No| L2["Offer: recommended set · changed set · full set"]
L2 --> L3([Wait for selection])
L3 --> M
L1 --> M
M["Resolve AND / OR rules<br/>Always include summary · exclude unselected components"]
M --> M1["Selected component contracts<br/>brief-report · presentation · website · documentation · brand-book<br/>unit-economics · research-pack · validation-kit · prototype<br/>go-to-market · solution-blueprint"]
M1 --> N["Prepare the output location before writing<br/>Honor an explicit location, otherwise use .ideas/<unique-slug>/<br/>If needed, ignore it in .git/info/exclude · never overwrite an earlier run<br/>Do not change tracked .gitignore without an explicit request"]
N --> N1["Write run-manifest.md and task-brief.md<br/>Record language · mode · assumptions · components · planned files · status"]
end
subgraph ROUTING["3 · ROUTE EFFORT AND SUBAGENTS"]
N1 --> O["Route each job independently<br/>fast/low · balanced/medium · strong/high<br/>Escalate only disputed or high-risk work<br/>Disclose capability substitutions"]
O --> Q{Independent parallel work or specialist needed?}
Q -->|No| R
Q -->|Yes| Q1["Main agent defines independent, bounded perspectives<br/>Share the brief and source list, not full transcripts<br/>Prefix every task with the language contract"]
Q1 --> Q2["Run within the host concurrency limit"]
Q2 --> Q3["Each subagent returns findings · evidence · assumptions<br/>largest risk · confidence · next experiment"]
Q3 --> Q4["Main agent resolves conflicts and owns<br/>the brief · decisions · synthesis · final artifacts"]
Q4 --> Q5[Keep raw agent chatter out of deliverables]
Q5 --> R
end
subgraph RESEARCH["4 · RESEARCH, COMPARE, AND VALIDATE"]
R[Gather only the project and external context required by selected outputs]
R --> S{Could an important claim have changed?}
S -->|Yes| S1["Use current primary sources<br/>Record retrieval dates · cite claims near the claim"]
S -->|No| S2[Use stable references and verified project evidence]
S1 --> T
S2 --> T
T{Does the task require a solution choice?}
T -->|No| U
T -->|Yes| T1["Climb the solution ladder<br/>1. No new feature<br/>2. Existing capability<br/>3. Native platform feature<br/>4. Installed dependency<br/>5. Mature external option<br/>6. Minimal custom solution"]
T1 --> T2["Stop at the first sufficient rung<br/>Compare correctness · safety · fit · reuse · cost<br/>accessibility · maintenance · verification · rollback"]
T2 --> T3["For external options: at most three serious candidates<br/>Verify current facts and state decisive rejection reasons"]
T3 --> U
U{Does a selected artifact need visuals?}
U -->|Yes| U1["Choose what the visual must prove<br/>real screenshot · faithful mockup · wireframe · creative concept<br/>Use authorized project visuals · label the mode"]
U -->|No| V
U1 --> V
V{Is unit economics selected?}
V -->|Yes| V1["Use editable assumptions and formulas<br/>Conservative · base · upside scenarios<br/>Show units · rationale · confidence · sensitivity"]
V -->|No| W
V1 --> W
W["Find contrary evidence and the riskiest decisive assumption"]
W --> X["Define the cheapest decisive experiment<br/>hypothesis · target · procedure · signal · pass/fail thresholds<br/>cost · duration · decision after each outcome"]
end
subgraph DECISION["5 · DECIDE"]
X --> Y{What does the evidence justify?}
Y -->|Narrow next step supported| Y1["GO"]
Y -->|Cheap experiment can resolve the unknown| Y2["CONDITIONAL"]
Y -->|Further investment is not justified| Y3["FREEZE"]
end
subgraph BUILD["6 · BUILD ONLY WHAT THE DECISION NEEDS"]
Y1 --> Z[Build the selected artifacts]
Y2 --> Z1["Build the selected artifacts<br/>Include the decisive experiment and thresholds"]
Y3 --> Z2["Build decision and validation artifacts<br/>State what evidence could reopen the idea<br/>Stop costly promotional work unless explicitly requested"]
Z --> AA
Z1 --> AA
Z2 --> AA
AA["Build shared evidence, messages, calculations,<br/>tokens, charts, and images once"]
AA --> AB["Finish independent components when another is blocked<br/>Mark blocked or skipped work instead of fabricating it"]
AB --> AC{Would this change application code?}
AC -->|No| AD
AC -->|Yes| AC1{Did the user request implementation?}
AC1 -->|No| AC2[Stop at the decision or blueprint]
AC1 -->|Yes| AD["Preserve the current architecture<br/>Change the fewest cohesive files<br/>Leave one proportionate regression check"]
AC2 --> AE
AD --> AE
AE{External write, purchase, production access, deployment, or public release?}
AE -->|No| AF[Keep the result local]
AE -->|Yes| AE1{Exact action already authorized?}
AE1 -->|No| AE2([Ask before acting])
AE2 --> AE3([Wait for authorization])
AE3 -->|Approved| AF1[Perform only the authorized action]
AE3 -->|Declined| AF
AE1 -->|Yes| AF1
end
subgraph DELIVERY["7 · VERIFY, PERSIST, AND HAND OFF"]
AF --> AG
AF1 --> AG
AG["Verify selected outputs · sources · assumptions · arithmetic<br/>editable files · rendered visuals · terminology · verdict<br/>language consistency · usability · absence of placeholders"]
AG --> AH{All checks pass?}
AH -->|No| AH1[Return to the relevant research, build, or verification step]
AH1 --> AG
AH -->|Yes| AI["Complete sources.md and run-manifest.md<br/>Keep evidence, calculations, research notes, assets,<br/>statuses, and substitutions traceable"]
AI --> AJ["Write summary.md<br/>What · why · verdict · assumptions · usage · links<br/>limits · unresolved risks · next action"]
AJ --> AK[Return a concise chat summary in the selected language]
AK --> AL([Usable next step])
end
classDef start fill:#0B1220,stroke:#D7FF64,color:#F4F1E8,stroke-width:2px;
classDef gate fill:#F4F1E8,stroke:#FF9D66,color:#0B1220,stroke-width:2px;
classDef action fill:#16233B,stroke:#7AB8FF,color:#F4F1E8;
classDef wait fill:#E8E3D5,stroke:#64748B,color:#0B1220,stroke-dasharray:4 3;
classDef go fill:#D7FF64,stroke:#0B1220,color:#0B1220,stroke-width:2px;
classDef conditional fill:#7AB8FF,stroke:#0B1220,color:#0B1220,stroke-width:2px;
classDef freeze fill:#FF9D66,stroke:#0B1220,color:#0B1220,stroke-width:2px;
class A,A3,AL start;
class A1,B,D,H,L,Q,S,T,U,V,Y,AC,AC1,AE,AE1,AH gate;
class B2,D2,H2,L3,AE2,AE3 wait;
class Y1 go;
class Y2 conditional;
class Y3 freeze;
class A2,C,C1,D1,E,F,F1,F1B,F2,F3,F4,F5,G,H1,I,J,K,L1,L2,M,M1,N,N1,O,P1,P2,P3,Q1,Q2,Q3,Q4,Q5,R,S1,S2,T1,T2,T3,U1,V1,W,X,Z,Z1,Z2,AA,AB,AC2,AD,AF,AF1,AG,AH1,AI,AJ,AK action;
- Scope gate, language contract, operating rules, effort routing, subagents, decision loop, persistence, verification, and delivery:
conceptops/SKILL.md - Context modes, evidence order, task brief, and question policy:
context-and-brief.md - Component selection, artifact contracts, and branching behavior:
output-components.md - Repository discovery, solution ladder, visual modes, and implementation boundary:
project-decisions.md - Research integrity, decision gate, experiments, and economics:
validation-and-economics.md
The diagram summarizes these documents. The linked files remain the normative instructions.