a9a7a328e6
Mechanical rename produced by scripts/msd-rename.cjs: gsd/Gsd/GSD -> msd/Msd/MSD across contents and paths, upstream package/repo coordinates -> @golem15/msd-core and golem15com/msd-core. Deep links into upstream history, sibling upstream packages, the GSD-2 import feature, CHANGELOG.md and .changeset/ are kept as-is. Hand edits on top: MSD block-letter banner and logos, LICENSE copyright line, package/plugin identity, regenerated lockfile, install-tree fixtures, derived registries and benchmark baseline; migration checksum baseline re-locked (MSD keeps its own install state, so no install had applied the old sums); sort-order and regex-escaped expectations in tests adjusted.
97 lines
3.0 KiB
Markdown
97 lines
3.0 KiB
Markdown
# Thinking Partner Integration
|
|
|
|
Conditional extended thinking at workflow decision points. Activates when `features.thinking_partner: true` in `.planning/config.json` (default: false).
|
|
|
|
---
|
|
|
|
## Tradeoff Detection Signals
|
|
|
|
The thinking partner activates when developer responses contain specific signals indicating competing priorities:
|
|
|
|
**Keyword signals:**
|
|
- "or" / "versus" / "vs" connecting two approaches
|
|
- "tradeoff" / "trade-off" / "tradeoffs"
|
|
- "on one hand" / "on the other hand"
|
|
- "pros and cons"
|
|
- "not sure between" / "torn between"
|
|
|
|
**Structural signals:**
|
|
- Developer lists 2+ competing options
|
|
- Developer asks "which is better" or "what would you recommend"
|
|
- Developer reverses a previous decision ("actually, maybe we should...")
|
|
|
|
**When NOT to activate:**
|
|
- Developer has already made a clear choice
|
|
- The "or" is rhetorical or trivial (e.g., "tabs or spaces" — use project convention)
|
|
- Simple yes/no questions
|
|
- Developer explicitly asks to move on
|
|
|
|
---
|
|
|
|
## Integration Points
|
|
|
|
### 1. Discuss Phase — Tradeoff Deep-Dive
|
|
|
|
**When:** During `discuss_areas` step, after a developer answer reveals competing priorities.
|
|
|
|
**What:** Pause the normal question flow and offer a brief structured analysis:
|
|
```
|
|
I notice competing priorities here — {X} optimizes for {A} while {Y} optimizes for {B}.
|
|
|
|
Want me to think through the tradeoffs before we decide?
|
|
[Yes, analyze tradeoffs] / [No, I've decided]
|
|
```
|
|
|
|
If yes, provide a brief (3-5 bullet) analysis covering:
|
|
- What each approach optimizes for
|
|
- What each approach sacrifices
|
|
- Which aligns better with the project's stated goals (from PROJECT.md)
|
|
- A recommendation with reasoning
|
|
|
|
Then return to the normal discussion flow.
|
|
|
|
### 2. Plan Phase — Architectural Decision Analysis
|
|
|
|
**When:** During step 11 (Handle Checker Return), when the plan-checker flags issues containing architectural tradeoff keywords.
|
|
|
|
**What:** Before sending to the revision loop, analyze the architectural decision:
|
|
```
|
|
The plan-checker flagged an architectural tradeoff: {issue description}
|
|
|
|
Brief analysis:
|
|
- Option A: {approach} — {pros/cons}
|
|
- Option B: {approach} — {pros/cons}
|
|
- Recommendation: {choice} because {reasoning aligned with phase goals}
|
|
|
|
Apply this recommendation to the revision? [Yes] / [No, let me decide]
|
|
```
|
|
|
|
### 3. Explore — Approach Comparison (requires #1729)
|
|
|
|
**When:** During Socratic conversation, when multiple viable approaches emerge.
|
|
**Note:** This integration point will be added when /msd:explore (#1729) lands.
|
|
|
|
---
|
|
|
|
## Configuration
|
|
|
|
```json
|
|
{
|
|
"features": {
|
|
"thinking_partner": true
|
|
}
|
|
}
|
|
```
|
|
|
|
Default: `false`. The thinking partner is opt-in because it adds latency to interactive workflows.
|
|
|
|
---
|
|
|
|
## Design Principles
|
|
|
|
1. **Lightweight** — inline analysis, not a separate interactive session
|
|
2. **Opt-in** — must be explicitly enabled, never activates by default
|
|
3. **Skippable** — always offer "No, I've decided" to bypass
|
|
4. **Brief** — 3-5 bullets max, not a full research report
|
|
5. **Aligned** — recommendations reference PROJECT.md goals when available
|