Recommended Free Tools
A project constitution for AI agents is meant to guide persistent behavior—not document what the codebase happens to look like today. In Derek Wang’s proposed AI harness methodology, it establishes a project’s purpose, boundaries, and working rules; individual prompts then assign specific tasks within that framework. It is a practical approach, not a proven technical standard: Wang’s essay offers examples and prescriptions, but no controlled comparison showing that a constitution improves outcomes.
What a project constitution is—and what it is not
Wang frames a constitution as a durable, project-level behavior contract for the people and AI agents working on a project. A task prompt says what to do now; the constitution says how the agent should approach work across tasks. Code conventions, by contrast, chiefly constrain the form of the artifact the agent produces. They can complement a constitution, but they answer a different question.
The distinction also separates governance from documentation. A file that records the current architecture describes what exists. A constitution sets expectations for behavior or, in a related practitioner framing, states what the system should become after an intervention.
In the related Generative Specification practitioner protocol, an architectural constitution guides how a system is built, while a technical specification describes what it is and must do. That terminology offers a useful comparison, but it comes from its own practitioner context; it does not independently validate Wang’s method.
Quick wins for a faster PC:
Repair Windows errors before they cause bigger problemsFix Now →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →Clear out junk files and repair common Windows errorsFree Scan →#1 Best Overall
What should a constitution answer?
Wang’s three framing questions offer a compact way to decide what belongs in the document:
- Who are we? State the project’s essence and purpose, so an agent can orient decisions to the project rather than merely to the latest request.
- Where are the boundaries? Specify what is outside scope or unacceptable. Boundaries help distinguish a useful solution from one that violates project constraints.
- How do we work together? Define the working relationship between humans and AI, including the process the agent is expected to follow.
These questions are Wang’s proposed framework, not a standardized schema. The right amount of detail depends on the project and how much collaboration it needs to govern.
Rank #2
- 40 Color-coded tabs: Highlight the most important sections with over 40 colored tabs for the Applied Behavior Analysis Cooper 3rd edition. The colors match the part for easy reference.
- Find Sections Easily and Efficiently: Our color-coded tabs have large font and are printed on both sides so you can easily navigate the Applied Behavior Analysis Cooper 3rd edition.
- Includes Alignment Card for Perfectly Aligned Tabs: Our tabs are easy to install in a perfect alignment using our tabs alignment system. Each tab includes the location and page number for super easy installation.
- Repositionable: If you misalign the tab no problem! The tabs are repositionable but also once they are folded, stick securely so navigating the ABA is easy and efficient.
- Blank Tabs Included: Additionally we include blank tabs so you can highlight anything specific to your needs.
Behavior rules versus code conventions
Wang gives ten example behavior rules from his methodology project. They illustrate the kind of guidance a constitution might hold; they are not a universal checklist validated across teams or tools.
- Read relevant files before writing.
- State assumptions rather than silently treating them as facts.
- Check downstream effects of a change.
- Make the smallest useful edit.
- Verify a fix instead of assuming it worked.
- Keep each function focused on one responsibility.
- Trust current files over chat memory when they conflict.
- Fail safely when a situation is uncertain or an operation cannot be completed.
- Record lessons that should inform later work.
- Replace a tool after repeated failures of the same kind.
Several rules touch implementation—such as keeping a function focused—but their role in this list is behavioral: they tell the agent how to work. A code-style guide may specify formatting or naming; the constitution can establish broader expectations such as reading before editing or checking the consequences of a change.
Rank #3
- Keep track of everything from attendance to test scores
- Spiral bound
- Measures 8-1/2" x 11"
How to make the rules more than aspirations
A document cannot ensure that an agent follows it simply by existing. Wang’s central implementation point is that constitutional rules need a path into the layers beneath them: the agent must load the document before beginning work, and the surrounding process must make important requirements checkable. Without that connection, the file may remain an aspiration rather than an operating rule. This is an argument about implementation, not a demonstrated guarantee of compliance.
Wang describes two structures: a five-part constitution set in a methodology framework and a single CONSTITUTION.md in a trading system. His suggested principle is to match the structure to the complexity of the collaboration, rather than treating one file layout as mandatory.
Rank #4
He also argues that the constitution should take precedence over an agent’s own AGENTS.md and skill documentation. Teams adopting that hierarchy need to make it explicit in their setup and loading process; a priority claim written in one file does not, on its own, establish how an agent resolves conflicting instructions.
To keep the document current, Wang recommends making amendments visible through repository changes, review, and recorded decisions. That gives collaborators a way to see what changed and why. It is a maintenance proposal, not evidence that review alone prevents drift or guarantees adherence.
Best Value
What the evidence does—and does not—show
Wang’s September 19, 2026 DEV Community essay, “A constitution is a behavior contract, not a description”, is the primary source for this proposed methodology. It supplies examples and implementation advice, but does not report a controlled comparison, measured error reduction, or validated confidence calibration. The approach may help a team make expectations explicit; the essay does not establish how much it improves agent performance.
The essay also says that, in Wang’s methodology data contract, a [HIGH] label maps to at least 80% confidence. That is a project-specific threshold, not a general standard or a measure of calibrated accuracy. It should not be carried over to another confidence system without separate evidence.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
