AIPM AI skills
Eleven reusable skills turn AIPM into repeatable AI assistance across project, program, and portfolio management. A skill is a packaged set of instructions for a defined task: when to use it, what evidence it needs, how to reason, and what useful result to produce. Skills share the same operating rules; they are not separate autonomous agents.
Each skill is self-contained. Install only the one you need, or install all of them with the aipm umbrella skill, which routes a request to the right one.
| Skill | Central question | Produces |
|---|---|---|
aipm (umbrella) | Which AIPM help does this need? | Routing, end-to-end cycles, framework explanation |
aipm-true-north | What are we trying to accomplish, for whom, and why now? | Outcome agreement |
aipm-portfolio-choices | What should we start, sequence, defer, or stop? | Feasible scenario comparison |
aipm-evidence-brief | What do we know, what changed, what is disputed? | Traceable evidence brief |
aipm-stay-on-target | Are we actually getting closer? | Alignment and drift review |
aipm-dependency-readiness | What could prevent the outcome, and when do options narrow? | Exposure and escalation brief |
aipm-decision-brief | Who must choose what, by when, with which tradeoffs? | Decision brief |
aipm-break-the-mold | What one uncomfortable change is worth testing? | Experiment card |
aipm-benefits-adoption | Did delivery create the intended value, and will it last? | Benefits assessment |
aipm-applied-learning | What did we learn, and where will it change the next decision? | Applied lesson |
aipm-leadership-dialogue | What might we be overlooking? | Facilitation guide |
Where to start
- True North → Evidence Brief → Decision Brief on one existing review.
- Add Stay on Target → Break the Mold → Applied Learning for a full adaptive cycle.
- Add dependency, benefits, portfolio, and dialogue skills when the work requires them.
Install
Claude Code (all projects):
git clone https://github.com/JeffGeissler/Adaptive-Intelligence-PPM.git
mkdir -p ~/.claude/skills
cp -R Adaptive-Intelligence-PPM/skills/* ~/.claude/skills/
To install one skill, copy only its folder, for example skills/aipm-decision-brief.
Claude Code (this repository only): open the cloned folder as your project. The skills are discovered through .claude/skills/.
Claude.ai: zip a skill folder (for example aipm-true-north/) and upload it as a custom skill in your Claude settings.
Other assistants: paste the contents of a skill’s SKILL.md as instructions. The skills are plain Markdown with no code or vendor dependencies.
Shared operating rules
Every skill:
- Anchors analysis to an outcome and a pending decision or useful action.
- Uses the minimum relevant, permitted evidence, and keeps references, dates, scope, and limitations.
- Distinguishes fact, interpretation, assumption, recommendation, and human approval.
- Surfaces uncertainty and dissent, and never fills missing management agreements with invented facts.
- Identifies the responsible human and the next review point, labeling unconfirmed owners as proposed.
- Reuses existing records and ceremonies, scaling detail to consequence.
- Stays advisory: no external communications, system-of-record changes, funding commitments, or other execution.
Common response shape: Outcome → Evidence → Interpretation and uncertainty → Options or recommendation → Owner and next review.
The skills are proposed specifications. They have not yet been evaluated in field use, and no effectiveness claims are made.