Kaizen Guided Execution Pipeline

The Guided Execution pipeline is a sequence of Kaizen blocks that turn raw todos into normalized, contextualized, time-bounded tasks with prompts, monitoring, and validation. Use it when you want structured task execution with intent inference, time windows, and completion validation.

Pipeline Order

Run blocks in this order. Each block consumes output from the previous step:

1. Kaizen Todos – Fetch todos (or use another source).
2. Todo Normalizer (kaizen_todo_normalizer) – Normalize text, tags, intent, effort, urgency.
3. Context Extractor (kaizen_context_extractor) – Add prerequisites, dependencies, blockers.
4. Intent Derivation (kaizen_intent_derivation) – Define outcomes and success criteria.
5. Time Window (kaizen_time_window) – Create execution windows and check-ins.
6. Prompt Generator (kaizen_prompt_generator) – Generate start, nudge, and completion prompts.
7. Execution Monitor (kaizen_execution_monitor) – Track progress and detect anomalies.
8. Completion Validator (kaizen_completion_validator) – Validate against success criteria.
9. Reflection (kaizen_reflection) – Adapt effort and prompts from execution data.
10. Streak Validator (kaizen_streak_validator) – Optional; validate completions against streak rules (e.g. after activity trigger).

Block Types

Referencing Outputs

Use block output references to connect the pipeline. Standard syntax is <block_id.outputName>.

Example:

• Normalizer output: <kaizen_todo_normalizer_1.normalizedTodos> or <kaizen_todo_normalizer_1.normalizedTodoIds>
• Context Extractor input: reference normalizer output
• Intent Derivation input: reference context extractor output
• Time Window input: reference intent derivation output
• Prompt Generator input: reference time window and intent outputs
• Execution Monitor input: reference prompt generator and time window outputs
• Completion Validator input: reference execution monitor and intent outputs
• Reflection input: reference completion validator and execution monitor outputs

See Reference syntax for trigger vs block output syntax.

Common Use Cases

• Structured task execution: Normalize todos → set time windows → generate prompts → monitor → validate.
• Habit/streak workflows: After todo completion (activity trigger), run Streak Validator to check streak rules, then conditionally unlock badge or notify.
• Learning loop: Use Reflection to adapt effort estimates and prompt style for future runs.

Optional Steps

You can shorten the pipeline depending on needs:

• Minimal: Todos → Normalizer → Context Extractor → Intent Derivation (no time window or prompts).
• With validation: Add Completion Validator after Execution Monitor; optionally add Reflection.
• With streaks: Add Streak Validator after an activity trigger (todo completed) or after Completion Validator.