The AI Taxonomy.
Terms  /  Execution Quality  /  Plan Rot
04 · Execution Quality

Plan Rot

Decay of truth in planning artifacts — task state in markdown files goes stale.

Plan Rot happens when the planning document and reality diverge. A task marked "in progress" was actually abandoned two days ago. A task marked "done" was partially completed but never verified. A task not yet started has had its requirements change since the plan was written. Plan Rot is especially common in agentic workflows where plans live in markdown files or TODO lists that aren't automatically synchronized with actual work. The plan becomes a historical document rather than a living guide, and decisions made based on it are based on fiction.

Example
A team uses a markdown plan to track an agent's progress on a feature. The plan shows 8 of 12 tasks complete. But when a human reviews the actual code: task 3 was done incorrectly and needs rework, task 6 was marked done but the agent only created a stub, and task 8's requirements changed after it was "completed." The plan says the feature is 67% done. Reality says it's closer to 40%. The plan has rotted — it no longer reflects the truth.

More in Execution Quality

Execution Fidelity

How closely a system follows and completes its intended plan.

Plan Drift

Deviation from the original plan over time.

Execution Hallucination

Claiming work is done when it is not.

Execution Integrity

Consistent, verifiable completion of planned tasks with evidence.

Plan Resilience

A plan's ability to remain accurate and actionable over time despite changing conditions. The opposite of Plan Rot.

Task Accountability

Every task has a verifiable record of who did what, when, and whether it actually completed.

Agentic Scope Creep

An agent silently expanding beyond what was delegated, usually from lack of a clear plan or direction. Adds breadth the user didn't ask for.

Depth Fixation

The agent keeps drilling deeper into implementation detail, generating new sub-tasks and refinements rather than recognizing the work is done and returning control to the user. Expands depth rather than breadth.