augment

name	augment
description	Writes experimental implementation of a module change from an approved Blueprint. For AUGMENT type: completes a modified copy of the existing module. For NEW type: completes Factory stubs. All output goes to .wabblespec/state/experiments/augments/. Never touches production module space.

Augment

You write the experiment. Not the production module — the experiment.

What this skill does

Augment reads an approved blueprint.json and produces a working experimental module implementation in .wabblespec/state/experiments/augments/{blueprint-id}/. For AUGMENT type blueprints, it copies the current module files and applies the after_behavior change described in the blueprint. For NEW type blueprints, it completes the Factory stubs. The result must be a complete, runnable module — not a partial draft.

Augment does not write to modules/. It does not modify existing receipts. It is the only step between Blueprint and Benchmark.

When to use

Augment activates when:

1. blueprint.json exists with status: approved and Blueprint receipt confirms Attestation
2. For AUGMENT type: current module files at modules/{layer}/{module}/ are readable
3. For NEW type: Factory stubs exist at .wabblespec/state/experiments/augments/{blueprint-id}/
4. Human explicitly invokes Augment with the blueprint ID

Inputs

Input	Path	Required
Approved blueprint	.wabblespec/state/experiments/blueprints/{id}.blueprint.json	Yes
Blueprint receipt	.wabblespec/state/receipts/blueprint-*.json	Yes
Current module (AUGMENT)	modules/{layer}/{module}/	Yes for AUGMENT type
Factory stubs (NEW)	.wabblespec/state/experiments/augments/{blueprint-id}/	Yes for NEW type

Workflow

For AUGMENT type:

1. Copy all files from modules/{layer}/{module}/ to .wabblespec/state/experiments/augments/{blueprint-id}/
2. Apply the after_behavior change from blueprint.json to the copied SKILL.md
3. Update skill-rules.json as needed for the behavior change
4. Update receipt schema if new fields are required
5. Update tests/acceptance.md to include criteria for the new behavior
6. Do not copy receipts/ directory from the source module

For NEW type:

1. Complete each [FILL] stub in the Factory scaffold
2. Write full SKILL.md — all required sections (purpose, activation, inputs, output contract, failure modes, not-tested)
3. Populate skill-rules.json activators, anti_activators, preconditions
4. Extend receipt schema with module-specific required fields
5. Write concrete GWT acceptance scenarios

Reference Routing

Situation	Reference
Augment receipt write	engine/shared/references/script-delegation-contract.md → receipt-writer.py --type augment

Output contract

Directory: .wabblespec/state/experiments/augments/{blueprint-id}/

Complete module files — no [FILL] markers, no placeholders. Every section populated. Receipt schema contains at least one module-specific required field beyond the base.

The experimental module must pass the same structural checks as a production module: SKILL.md has purpose statement, activation, input/output contracts, failure modes, not-tested. skill-rules.json has layer, activators, authority, verification_mode. Receipt schema is valid JSON Schema extending the base.

Failure modes

Incomplete stub — Augment writes a file with [FILL] markers remaining. Fix: all placeholders must be resolved before receipt is written. Not-tested section must name what Augment could not complete.

Scope drift — Augment implements more than what blueprint.json specifies. Fix: stay within before_behavior → after_behavior scope. Additional improvements go through a new Synth cycle.

Production contamination — Augment modifies files in modules/. Fix: only .wabblespec/state/experiments/augments/ is writable. Any write to modules/ is a hard violation of I11.

Not tested

Augment cannot verify that the experimental module will pass Benchmark. Augment verifies structural completeness only. Benchmark validates behavioral quality against fixtures.