Steel-Manning
Adversarial verification of convergence decisions. This campaign subjects winners, criteria, and rejected alternatives to rigorous challenge — ensuring that final decisions survive the strongest possible counter-arguments rather than merely the weakest objections.
The campaign deploys four complementary attack vectors: resurrecting rejected candidates to test whether elimination was justified, stress-testing winners to find hidden weaknesses, interrogating the criteria framework itself, and simulating stakeholder objections for political feasibility.
Strategy Routing
| Signal |
Strategy |
| resurrect rejected candidates / give losers a fair hearing |
resurrection-advocacy |
| stress-test the winner / find weaknesses in top pick |
winner-stress-testing |
| challenge the criteria themselves / meta-level questioning |
criteria-interrogation |
| simulate stakeholder objections / political feasibility |
stakeholder-objection-simulation |
| construct strongest counter-argument / dialectical challenge |
counter-thesis-construction |
Manifest
Strategies
| Strategy |
Method Lineage |
| resurrection-advocacy |
Devil's Advocacy, Dialectical Inquiry, Adversarial Collaboration (Kahneman) |
| winner-stress-testing |
Pre-mortem (Klein), Red Teaming, Failure Mode Analysis |
| criteria-interrogation |
Assumption-based Planning, Critical Systems Heuristics, Boundary Critique |
| stakeholder-objection-simulation |
Role-play, Stakeholder Analysis, Political Feasibility |
| counter-thesis-construction |
Dialectical Inquiry, Thesis-Antithesis-Synthesis, Adversarial Debate |
Tactics
| Tactic |
SOPs Used |
| adversarial-debate-protocol |
advocate-construction, critic-attack, judge-verdict |
| assumption-excavation |
assumption-extraction, assumption-challenge, conclusion-sensitivity |
| multi-perspective-attack |
perspective-assignment, perspective-attack, steel-manning-synthesis |
SOPs
| SOP |
Input |
Output |
Shareable |
| advocate-construction |
rejected_candidate, context |
strongest_case_for_resurrection |
validation |
| critic-attack |
winner, advocate_case |
attack_arguments[], severity_ratings |
validation |
| judge-verdict |
advocate_case, critic_attacks |
verdict, reasoning, conditions |
validation |
| assumption-extraction |
decision, evidence |
assumptions[], confidence_levels |
— |
| assumption-challenge |
assumption |
challenge_argument, alternative, impact_if_wrong |
— |
| conclusion-sensitivity |
assumptions[], challenges[] |
sensitivity_map, critical_assumptions[] |
— |
| perspective-assignment |
decision, stakeholders |
perspective_briefs[] |
— |
| perspective-attack |
decision, perspective_brief |
attacks[], constructive_alternatives[] |
— |
| steel-manning-synthesis |
all_attacks, all_verdicts |
final_verdict, surviving_concerns, modifications |
— |
Budget Table (M Tier)
| Metric |
Minimum |
| Attack perspectives |
>= 3 distinct angles |
| Debate rounds |
>= 2 per contested decision |
| Assumptions challenged |
>= 5 per winner |
| Final verdict |
Explicit ACCEPT / REJECT / REVISE with evidence |
MCP Tools
mcp__wiki-vault__vault_search — retrieve prior decisions and contextmcp__wiki-vault__vault_query_graph — trace dependency chains for impact analysismcp__wiki-vault__vault_add_edge — record challenge relationships
Context Management
The campaign maintains a Challenge Ledger tracking:
- Which decisions have been challenged
- Attack vectors applied to each
- Verdicts rendered and their conditions
- Surviving concerns requiring monitoring
State is passed between strategies via the ledger. Each strategy updates it upon completion.
Available Strategies
Optional, no fixed order; the final leaf is always a sop.
| Strategy |
When to use |
| counter-thesis-construction |
Construct the strongest possible counter-argument to the convergence decision using Dialectical Inquiry and Thesis-Antithesis-Synthesis methods. |
| criteria-interrogation |
Challenge the evaluation criteria themselves using Assumption-based Planning, Critical Systems Heuristics, and Boundary Critique to ensure the framework is sound. |
| resurrection-advocacy |
Argue for rejected candidates using Devil's Advocacy, Dialectical Inquiry, and Adversarial Collaboration to ensure elimination was justified. |
| stakeholder-objection-simulation |
Simulate stakeholder objections through role-play and political feasibility analysis to test whether the decision survives real-world opposition. |
| winner-stress-testing |
Stress-test the winning candidate using Pre-mortem, Red Teaming, and Failure Mode Analysis to expose hidden weaknesses before commitment. |
Available SOPs
Optional, no fixed order; the final leaf is always a sop.
| SOP |
When to use |
| context-checkpoint |
Append research process and results to the current Phase's context file. Covers both process and results with genuine substance. Use this skill at plan-designated checkpoint points — typically after each strategy completes or at key decision nodes within a research Phase. |
| context-init |
Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed. |
| convergence-multi-stakeholder-simulation |
Simulates diverse stakeholder perspectives and their strongest objections/support arguments. Shared across steel-manning and consensus campaigns. |
| convergence-saturation-detection |
Determines when to stop iterating — coverage threshold met or marginal returns diminishing. Shared across all campaigns. |
| convergence-sensitivity-analysis |
Tests conclusion robustness by perturbing parameters and observing rank changes. Shared across scoring, portfolio, and steel-manning campaigns. |
1---2name: steel-manning3description: Steel-Manning Campaign — adversarial verification of convergence decisions through resurrection advocacy, winner stress-testing, criteria interrogation, and multi-perspective attack using Devil's Advocacy, Pre-mortem, Red Teaming, Dialectical Inquiry methods.4---56# Steel-Manning78Adversarial verification of convergence decisions. This campaign subjects winners, criteria, and rejected alternatives to rigorous challenge — ensuring that final decisions survive the strongest possible counter-arguments rather than merely the weakest objections.910The campaign deploys four complementary attack vectors: resurrecting rejected candidates to test whether elimination was justified, stress-testing winners to find hidden weaknesses, interrogating the criteria framework itself, and simulating stakeholder objections for political feasibility.1112## Strategy Routing1314| Signal | Strategy |15|--------|----------|16| resurrect rejected candidates / give losers a fair hearing | resurrection-advocacy |17| stress-test the winner / find weaknesses in top pick | winner-stress-testing |18| challenge the criteria themselves / meta-level questioning | criteria-interrogation |19| simulate stakeholder objections / political feasibility | stakeholder-objection-simulation |20| construct strongest counter-argument / dialectical challenge | counter-thesis-construction |2122## Manifest2324### Strategies2526| Strategy | Method Lineage |27|----------|---------------|28| resurrection-advocacy | Devil's Advocacy, Dialectical Inquiry, Adversarial Collaboration (Kahneman) |29| winner-stress-testing | Pre-mortem (Klein), Red Teaming, Failure Mode Analysis |30| criteria-interrogation | Assumption-based Planning, Critical Systems Heuristics, Boundary Critique |31| stakeholder-objection-simulation | Role-play, Stakeholder Analysis, Political Feasibility |32| counter-thesis-construction | Dialectical Inquiry, Thesis-Antithesis-Synthesis, Adversarial Debate |3334### Tactics3536| Tactic | SOPs Used |37|--------|-----------|38| adversarial-debate-protocol | advocate-construction, critic-attack, judge-verdict |39| assumption-excavation | assumption-extraction, assumption-challenge, conclusion-sensitivity |40| multi-perspective-attack | perspective-assignment, perspective-attack, steel-manning-synthesis |4142### SOPs4344| SOP | Input | Output | Shareable |45|-----|-------|--------|-----------|46| advocate-construction | rejected_candidate, context | strongest_case_for_resurrection | validation |47| critic-attack | winner, advocate_case | attack_arguments[], severity_ratings | validation |48| judge-verdict | advocate_case, critic_attacks | verdict, reasoning, conditions | validation |49| assumption-extraction | decision, evidence | assumptions[], confidence_levels | — |50| assumption-challenge | assumption | challenge_argument, alternative, impact_if_wrong | — |51| conclusion-sensitivity | assumptions[], challenges[] | sensitivity_map, critical_assumptions[] | — |52| perspective-assignment | decision, stakeholders | perspective_briefs[] | — |53| perspective-attack | decision, perspective_brief | attacks[], constructive_alternatives[] | — |54| steel-manning-synthesis | all_attacks, all_verdicts | final_verdict, surviving_concerns, modifications | — |5556## Budget Table (M Tier)5758| Metric | Minimum |59|--------|---------|60| Attack perspectives | >= 3 distinct angles |61| Debate rounds | >= 2 per contested decision |62| Assumptions challenged | >= 5 per winner |63| Final verdict | Explicit ACCEPT / REJECT / REVISE with evidence |6465## MCP Tools6667- `mcp__wiki-vault__vault_search` — retrieve prior decisions and context68- `mcp__wiki-vault__vault_query_graph` — trace dependency chains for impact analysis69- `mcp__wiki-vault__vault_add_edge` — record challenge relationships7071## Context Management7273The campaign maintains a **Challenge Ledger** tracking:74- Which decisions have been challenged75- Attack vectors applied to each76- Verdicts rendered and their conditions77- Surviving concerns requiring monitoring7879State is passed between strategies via the ledger. Each strategy updates it upon completion.8081<!-- BEGIN available-tables (generated) -->8283## Available Strategies8485Optional, no fixed order; the final leaf is always a sop.8687| Strategy | When to use |88| --- | --- |89| counter-thesis-construction | Construct the strongest possible counter-argument to the convergence decision using Dialectical Inquiry and Thesis-Antithesis-Synthesis methods. |90| criteria-interrogation | Challenge the evaluation criteria themselves using Assumption-based Planning, Critical Systems Heuristics, and Boundary Critique to ensure the framework is sound. |91| resurrection-advocacy | Argue for rejected candidates using Devil's Advocacy, Dialectical Inquiry, and Adversarial Collaboration to ensure elimination was justified. |92| stakeholder-objection-simulation | Simulate stakeholder objections through role-play and political feasibility analysis to test whether the decision survives real-world opposition. |93| winner-stress-testing | Stress-test the winning candidate using Pre-mortem, Red Teaming, and Failure Mode Analysis to expose hidden weaknesses before commitment. |9495## Available SOPs9697Optional, no fixed order; the final leaf is always a sop.9899| SOP | When to use |100| --- | --- |101| context-checkpoint | Append research process and results to the current Phase's context file. Covers both process and results with genuine substance. Use this skill at plan-designated checkpoint points — typically after each strategy completes or at key decision nodes within a research Phase. |102| context-init | Create a new context file for a research Phase. Called once at Phase start to initialize the file that subsequent context-checkpoint calls will append to. Use this skill whenever a new research Phase begins and a fresh context file is needed. |103| convergence-multi-stakeholder-simulation | Simulates diverse stakeholder perspectives and their strongest objections/support arguments. Shared across steel-manning and consensus campaigns. |104| convergence-saturation-detection | Determines when to stop iterating — coverage threshold met or marginal returns diminishing. Shared across all campaigns. |105| convergence-sensitivity-analysis | Tests conclusion robustness by perturbing parameters and observing rank changes. Shared across scoring, portfolio, and steel-manning campaigns. |106107<!-- END available-tables (generated) -->
yogsoth-ai