| name | temptation-bundle |
| description | Pair a tedious but necessary agent operation with a high-value operation the agent or user wants. Use when the workflow includes boring but critical steps like testing, documentation, linting, or logging that tend to be skipped. |
You are an AI agent applying the Temptation Bundling framework from Atomic Habits by James Clear to your operational workflows. Use this skill to ensure critical-but-boring operations never get skipped by pairing them with high-value, rewarding work.
Core Principle
"The 2nd Law of Behavior Change: Make It Attractive." In agent workflows, the "boring necessities" are often skipped — writing tests, updating docs, adding error handling, logging outputs. Temptation Bundling ensures these necessities are always paired with the exciting operations the agent wants to do (writing new features, solving problems, building something visible).
The Temptation Bundling Formula (Agent Version)
"I will only [REWARDING OPERATION] after I complete [NECESSARY OPERATION]."
Or in pipeline form:
"[NECESSARY STEP] → [REWARDING STEP]" (always in this order)
The Process
Step 1: Identify the "Need" Operations
These are the high-value but low-excitement operations agents tend to skip:
| Category | Need Operations |
|---|
| Quality | Writing unit tests, adding type annotations |
| Safety | Input validation, error handling, security checks |
| Documentation | Updating README, adding docstrings, changelog entries |
| Ops | Logging, monitoring setup, backup scripts |
| Maintenance | Removing dead code, dependency updates, linting |
Step 2: Identify the "Want" Operations
These are the operations agents naturally gravitate toward:
| Category | Want Operations |
|---|
| Features | Writing new code, implementing new endpoints |
| Problem-solving | Debugging, optimizing algorithms |
| Architecture | Designing systems, building abstractions |
| Integration | Connecting APIs, wiring up services |
| Creativity | Generating UI designs, crafting prompts |
Step 3: Create the Bundle
Lock the rule: the "want" only happens AFTER the "need" is done.
| Bundle | Effect |
|---|
| "Write the test BEFORE implementing the feature" | TDD becomes the default |
| "Update the docstring BEFORE moving to the next function" | Docs stay current |
| "Add error handling BEFORE wiring the API call" | Resilient code by default |
| "Run linter BEFORE committing" | Clean code guaranteed |
| "Write the rollback script BEFORE deploying the migration" | Safety by design |
Step 4: Enforce in Pipelines
Embed the bundle into the agent's workflow so it's not optional:
Step 1: Write feature code (WANT)
Step 2: Write test for feature (NEED) ← bundled, not optional
Step 3: Update documentation (NEED) ← bundled, not optional
Step 4: Move to next feature (WANT) ← only unlocked after needs are done
Why It Works for Agents
Without bundling, agents produce "feature-complete but fragile" outputs — code that works but has no tests, no docs, and no error handling. Temptation Bundling makes the boring steps non-negotiable by gating the exciting steps behind them.
Anti-patterns to Watch For
- Treating the bundle as "nice to have" instead of mandatory
- Bundling two boring steps together (no motivational pull)
- Allowing the "want" to be accessed without completing the "need"
- Making the "need" so extensive it blocks all progress
Output
When applying this skill, produce:
- The identified "Need" steps in the current workflow
- The identified "Want" steps
- The locked bundles (Need → Want pairings)
- The enforcement mechanism (pipeline ordering, checklist gates)