decision-map

DecisionMap

DecisionMap is a practical protocol for turning complex business, product, market, and marketing decisions into visible strategy maps.

It is an installable Agent Skill, an open protocol, and a prompt toolkit. It is not a hosted service or a substitute for leadership judgment.

Try It in 60 Seconds

Install the skill for Codex or another supported coding agent:

npx skills add markoblogo/decision-map --skill decision-map

Then ask:

Use DecisionMap to map this decision: [decision]. Compare realistic options, expose the assumptions, and tell me what evidence would change the recommendation.

The skill is self-contained. The protocol, schemas, and worked examples in this repository provide deeper guidance and machine-readable outputs.

What It Is

Use DecisionMap when the real problem is not a lack of ideas, but a messy decision with multiple plausible paths, asymmetric risks, and uncertain external reactions.

DecisionMap helps users:

• restate the real decision
• separate facts, assumptions, interpretations, and unknowns
• generate 3-7 distinct strategic options
• compare trade-offs, risks, resources, and breakpoints
• pressure-test shortlisted options
• keep an update loop alive through a cascade log

The main output is a working strategic hypothesis, not a final truth.

Scope

DecisionMap is designed for:

• business strategy
• product strategy
• market positioning
• competitive response
• go-to-market decisions
• marketing strategy under uncertainty

It is intentionally out of scope for:

• military or political conflict
• legal or medical advice
• financial investment decisions
• mergers and acquisitions
• layoffs or HR restructuring

Manual Quick Start

1. Start with USAGE.md for the manual runbook.
2. Use prompts/system_prompt.md as the runtime system/developer prompt.
3. Run the stage prompts in order:
   • prompts/01_intake.md
   • prompts/02_clarifying_questions.md
   • prompts/03_strategy_map.md
   • prompts/04_deep_dive.md
   • prompts/05_decision_summary.md
4. If the decision continues over time, use the cascade log artifacts in examples/.

Optional ID Bootstrap

DecisionMap stays standalone, but it can optionally consume ID as a portable human-context layer for longer-running strategic work.

Recommended bootstrap order:

1. profiles/<owner>/soul.md
2. profiles/<owner>/profile.core.md
3. profiles/<owner>/handshake.md

Use that order when:

• the same decision evolves across multiple sessions
• the user has strong formatting, tone, or decision-style preferences
• you want continuity without re-explaining the same working style every time

Do not treat ID as market evidence or decision truth. It is only user-context and operating-constraints input.

Canonical Examples

These are the flagship examples for v0.3.

• examples/agri_commodity_market_entry.md - compact agri/commodities strategy map for a Ukrainian sunflower oil exporter evaluating Indian market entry.
• examples/fmcg_route_to_market_full_run.md - full walkthrough for an FMCG chilled dairy brand redesigning route-to-market expansion.
• examples/cascade_log_agri_commodity_market_entry.md - worked update loop that continues the agri/commodities case after the first decision.

Starter generic examples are still included, but they are no longer the flagship credibility artifacts:

• examples/full_run_product_launch.md
• examples/product_launch.md
• examples/competitive_response.md

JSON Outputs

DecisionMap ships with normative schemas and real JSON fixtures:

• schemas/strategy_map.schema.json
• schemas/cascade_log.schema.json
• examples/json/strategy_map.agri_market_entry.json
• examples/json/cascade_log.agri_market_entry.json

Validate the public examples with:

python3 -m pip install -r requirements-dev.txt
python3 scripts/check_repo.py

Validate your own JSON output with:

python3 scripts/validate_examples.py path/to/output.json --schema strategy-map

Cascade Log

Stage 6 is a first-class capability, not just a note.

Use it when the decision evolves over weeks or months and you need structured memory for:

• what was decided
• which assumptions were active
• which signals moved
• what happened next
• what changed in the working hypothesis

Artifacts:

• examples/templates/cascade_log_template.md
• examples/cascade_log_agri_commodity_market_entry.md
• examples/json/cascade_log.agri_market_entry.json

Document Roles

Artifact	Role
README.md	Canonical entrypoint for new users and contributors
USAGE.md	Manual operator runbook for chat-based execution
protocol.md	Normative spec for stages, outputs, confidence language, and cascade semantics
prompts/*	Executable prompt assets used during a run
schemas/*	Machine-readable shape for Stage 3 and Stage 6 outputs

Repository Structure

decision-map/
├── README.md
├── USAGE.md
├── protocol.md
├── CONTRIBUTING.md
├── CHANGELOG.md
├── prompts/
├── skills/decision-map/
├── schemas/
├── examples/
│   ├── json/
│   └── templates/
├── scripts/
└── .github/

Contributing

The repo now has a lightweight open-source contribution layer:

• CONTRIBUTING.md
• .github/ISSUE_TEMPLATE/
• CHANGELOG.md

Contributions are especially useful for:

• stronger real-world examples
• protocol clarity improvements
• schema/example consistency
• better manual operator guidance

Roadmap

Roadmap is intentionally lightweight in-repo. Use GitHub Issues/Milestones for active work and prioritization.

Current direction:

• more measured real-world examples
• interoperable exports and schema fixtures
• clearer evidence provenance
• tested compatibility across Agent Skills clients

ABVX Ecosystem

DecisionMap remains standalone and usable with any LLM.

Inside the ABVX ecosystem:

• lab.abvx can list it as a decision/strategy protocol artifact
• agentsgen can maintain repo-local agent docs for contributors
• SET can track or audit the repo
• ID can optionally provide portable user context for longer-running decision work, with soul.md as the preferred first-pass bootstrap

None of these integrations are required for manual use.

Related repos:

• lab.abvx is the public hub where DecisionMap is cataloged.
• abvx-agent-skills lists DecisionMap as a standalone companion skill.
• AGENTS.md_generator, SET, and ID are optional adjacent tools.

Privacy

DecisionMap itself does not require storing data, but hosted LLM APIs may process your input externally.

For sensitive work:

• anonymize names, companies, exact numbers, and internal documents
• remove customer data and personal data
• use a local model or approved internal environment when needed

Security

Report vulnerabilities privately as described in SECURITY.md.

ABVX ecosystem

• AGENTS.md_generator — Keeps repository guidance and machine-readable context current. Current release: v0.5.1.
• abvx-agent-skills — Uses shared, reviewable agent capabilities during maintenance. Current release: v0.16.1.

This block is generated from the reviewed ABVX ecosystem registry.