Files
claw/docs/superpowers/plans/2026-04-20-scene-skill-102-natural-language-parameter-readiness-plan.md

105 lines
3.7 KiB
Markdown

# Scene Skill 102 Natural-Language Parameter Readiness Plan
> Date: 2026-04-20
> Design: `2026-04-20-scene-skill-102-natural-language-parameter-readiness-design.md`
## Plan Intent
Build a 102-scene natural-language invocation parameter readiness view before pseudo-production testing.
This plan answers which skills should be invoked with query conditions such as organization and period, which skills currently only support scene-keyword deterministic selection, and which required-param skills have resolver gaps.
## Fixed Inputs
1. `examples/scene_skill_102_final_materialization_2026-04-19/skills`
2. `tests/fixtures/generated_scene/scene_skill_102_deterministic_invocation_readiness_after_keyword_refinement_2026-04-20.json`
3. `tests/fixtures/generated_scene/scene_skill_102_full_direct_mock_execution_2026-04-20.json`
## Allowed Outputs
1. `tests/fixtures/generated_scene/scene_skill_102_natural_language_parameter_readiness_2026-04-20.json`
2. `tests/fixtures/generated_scene/scene_skill_102_natural_language_invocation_samples_2026-04-20.json`
3. `docs/superpowers/reports/2026-04-20-scene-skill-102-natural-language-parameter-readiness-report.md`
## Forbidden Files
1. `src/compat/scene_platform/dispatch.rs`
2. `src/compat/scene_platform/resolvers.rs`
3. `src/generated_scene/analyzer.rs`
4. `src/generated_scene/generator.rs`
5. `examples/scene_skill_102_final_materialization_2026-04-19/skills/*`
6. `tests/fixtures/generated_scene/scene_execution_board_2026-04-18.json`
## Phase 0: Freeze Boundary
### Tasks
1. Confirm final skill count is `102`.
2. Confirm this plan is analysis-only.
3. Confirm no browser, network, host bridge, or production execution is performed.
### Acceptance Criteria
1. No generated skill files are modified.
2. No runtime source files are modified.
## Phase 1: Parameter Manifest Scan
### Tasks
1. Read each `SKILL.toml` for scene name and archetype.
2. Read each `scene.toml` for deterministic suffix, params, and resolver declarations.
3. Record required params and resolver types.
4. Check resolver resources such as dictionary files.
### Acceptance Criteria
1. All `102` scenes have one parameter scan record.
2. Required param scenes are explicitly identified.
## Phase 2: Readiness Classification
### Tasks
1. Mark scenes with supported, populated resolver resources as `parameter-ready`.
2. Mark scenes with empty or missing resolver resources as `parameter-gap`.
3. Mark no-param scenes as `parameter-not-required`.
4. Mark no-param scenes with likely filter words as `parameter-implicit-risk`.
### Acceptance Criteria
1. Every scene has exactly one primary readiness class.
2. Resolver gaps list concrete file or config reasons.
## Phase 3: Invocation Sample Generation
### Tasks
1. Generate minimal invocation samples for every scene.
2. Generate parameterized samples for scenes with required params.
3. Generate cautionary samples for implicit-risk scenes.
4. Make clear when organization or period wording is not currently parsed.
### Acceptance Criteria
1. Sample JSON covers all `102` scenes.
2. Parameterized samples are not generated as if resolver gaps are resolved.
## Phase 4: Publish Report
### Tasks
1. Write readiness JSON.
2. Write invocation sample JSON.
3. Write superpowers report with counts, gaps, and next route.
### Acceptance Criteria
1. Report explains why `场景名。。。` is insufficient for parameterized scenes.
2. Report states whether pseudo-production batch input should be regenerated.
3. Stop after report; do not start implementation.
## Stop Statement
Stop after readiness assets and report are published. Do not edit runtime, generated skills, board assets, or pseudo-production execution records under this plan.