DAPLab observes incomplete-task-execution pattern (DAP-01)
Agents mark a task complete, produce a celebratory summary, and claim success while one or more named subtasks remain unimplemented. Observed across all five agents evaluated by DAPLab.
Description
The DAPLab study's DAP-01 pattern documents coding agents that claim completion while key subtasks remain undone — a skipped migration, an unwritten test, a TODO comment left in the output, a feature flag defaulted wrong. The paper's contribution is that every evaluated agent exhibited the pattern, making it population-scale rather than per-vendor. SAICA-KG records the pattern as the empirical anchor for the `incomplete_execution` FailureMode.
Exhibited failure modes
- Incomplete execution — Agent claims completion while subtasks are skipped, stubbed, or silently dropped.
Affected systems
- Claude Code
- Cline
- Cursor
- v0
- Replit Agent
Mitigated by
No mitigating tools indexed for this incident yet.
Sources
Notes
Mitigation: plan-mode checklists that the agent must explicitly mark done. The DAPLab finding suggests such checklists alone are insufficient (the agents self-report completion); pairing with test-based completion gates is part of Recipe 10.