@only1btayy/g2w 1.0.31 → 1.0.32
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +2 -2
- package/package.json +1 -1
- package/skills/back2it.md +2 -0
- package/skills/bring2life.md +60 -4
- package/skills/cut2it.md +9 -0
- package/skills/get2work.md +10 -0
- package/skills/ready2save.md +6 -0
- package/skills/the-builder.md +11 -1
- package/skills/the-challenger.md +11 -0
- package/skills/the-inspector.md +13 -0
- package/skills/the-visionary.md +12 -1
- package/skills/true2dagame.md +86 -23
- package/skills/true2plan.md +14 -3
package/README.md
CHANGED
|
@@ -156,7 +156,7 @@ These are escape hatches for engineers who don't need the ceremony. G2W has no c
|
|
|
156
156
|
|
|
157
157
|
### The Trust Layer
|
|
158
158
|
|
|
159
|
-
Delivered via
|
|
159
|
+
Delivered via CLAUDE.md so it loads into every session automatically as project instructions.
|
|
160
160
|
|
|
161
161
|
- Your explicit instruction is a direct order. No agent may override it, rationalize it away, or route around it.
|
|
162
162
|
- The AI answers your question first and then asks to proceed — never just acts.
|
|
@@ -241,7 +241,7 @@ Everyone tired of the bullshit. Vibe coders who just want to ship. Senior engine
|
|
|
241
241
|
|
|
242
242
|
If you've ever felt like your AI was working against you instead of with you, G2W is for you.
|
|
243
243
|
|
|
244
|
-
G2W works in other models — the Modular Doc System and commands are model-agnostic. But Claude Code is where it's fully alive. The Trust Layer
|
|
244
|
+
G2W works in other models — the Modular Doc System and commands are model-agnostic. But Claude Code is where it's fully alive. The Trust Layer lives in CLAUDE.md so it loads as project instructions every session. The A-Game Hook and context warnings are hook-delivered, running automatically without relying on the AI to remember. Other models get the workflow. Claude Code gets the guarantees.
|
|
245
245
|
|
|
246
246
|
---
|
|
247
247
|
|
package/package.json
CHANGED
package/skills/back2it.md
CHANGED
|
@@ -26,6 +26,8 @@ You are resuming a G2W session. Get back up to speed fast. No fluff.
|
|
|
26
26
|
|
|
27
27
|
4. **Doc integrity check:** Does the doc you read still match the current state? If something looks stale, flag it before touching anything else.
|
|
28
28
|
|
|
29
|
+
4b. **Health score check:** If the last session notes in CURRENT.md include a health score (e.g., "Health: 3.2/5") and it's below 3.5, suggest running `/g2w:true2dagame` before starting work.
|
|
30
|
+
|
|
29
31
|
5. **Output a 4-line summary** (no more):
|
|
30
32
|
```
|
|
31
33
|
Project: [active-project]
|
package/skills/bring2life.md
CHANGED
|
@@ -80,13 +80,13 @@ Do not announce what you're reading. Just read.
|
|
|
80
80
|
|
|
81
81
|
## Phase 2 — Generate Draft Docs
|
|
82
82
|
|
|
83
|
-
Write first-draft versions of all
|
|
83
|
+
Write first-draft versions of all 12 G2W docs based on what you found.
|
|
84
84
|
|
|
85
85
|
For every field you're confident about: fill it in.
|
|
86
86
|
For every field you're uncertain about: write `❓ [UNKNOWN — needs your input]`
|
|
87
87
|
For every field that clearly doesn't apply: write `N/A`
|
|
88
88
|
|
|
89
|
-
Generate all
|
|
89
|
+
Generate all 12:
|
|
90
90
|
- `CLAUDE.md`
|
|
91
91
|
- `ARCHITECTURE.md`
|
|
92
92
|
- `CONVENTIONS.md`
|
|
@@ -98,6 +98,41 @@ Generate all 11:
|
|
|
98
98
|
- `SECURITY.md`
|
|
99
99
|
- `CURRENT.md`
|
|
100
100
|
- `PLAN.md` (leave empty — written by The Visionary when a task begins)
|
|
101
|
+
- `TRAPS.md`
|
|
102
|
+
|
|
103
|
+
**CONVENTIONS.md must include a Context Budget section** at the end. Auto-generate this from the scan:
|
|
104
|
+
|
|
105
|
+
```
|
|
106
|
+
## Context Budget
|
|
107
|
+
|
|
108
|
+
### Project Size
|
|
109
|
+
- Total files: [count]
|
|
110
|
+
- Estimated total lines: [count]
|
|
111
|
+
|
|
112
|
+
### Heavy Files (read strategically, not whole)
|
|
113
|
+
| File | Lines | What to read | What to skip |
|
|
114
|
+
|------|-------|-------------|-------------|
|
|
115
|
+
| [large file] | [count] | [relevant sections] | [boilerplate, generated, vendor code] |
|
|
116
|
+
|
|
117
|
+
### Read Order by Task Type
|
|
118
|
+
| Task | Read first | Read if needed | Never read whole |
|
|
119
|
+
|------|-----------|----------------|-----------------|
|
|
120
|
+
| Bug fix | ERRORS.md, the specific file | ARCHITECTURE.md | [large generated files] |
|
|
121
|
+
| New feature | ARCHITECTURE.md, CONVENTIONS.md | FEATURES.md | [vendor/generated] |
|
|
122
|
+
| Refactor | CONVENTIONS.md, the specific file | ARCHITECTURE.md | [test fixtures, assets] |
|
|
123
|
+
```
|
|
124
|
+
|
|
125
|
+
**TESTING.md must include a Golden Cases section.** These are scenarios that must ALWAYS pass, regardless of what the current task is. Seed initial cases from the core functionality discovered during the scan — the things that would be catastrophic if they broke.
|
|
126
|
+
|
|
127
|
+
```
|
|
128
|
+
## Golden Cases
|
|
129
|
+
|
|
130
|
+
Scenarios that must pass before ANY task is marked complete. If a golden case breaks, the build fails — regardless of what the current task was.
|
|
131
|
+
|
|
132
|
+
| # | Scenario | Steps | Expected Result | Last Verified |
|
|
133
|
+
|---|----------|-------|----------------|---------------|
|
|
134
|
+
| GC-01 | [core scenario] | [exact steps] | [exact expected outcome] | [date] |
|
|
135
|
+
```
|
|
101
136
|
|
|
102
137
|
## Phase 2b — Code Audit
|
|
103
138
|
|
|
@@ -161,6 +196,27 @@ Generated by bring2life: [date]
|
|
|
161
196
|
|
|
162
197
|
Be specific. File paths, line numbers, what's actually broken and why. No vague "consider improving" — either it's a real issue or don't list it.
|
|
163
198
|
|
|
199
|
+
## Phase 2c — Seed TRAPS.md
|
|
200
|
+
|
|
201
|
+
From the code audit, extract any anti-patterns that represent project-specific pitfalls — things that look right but break things in this codebase. Write them to `TRAPS.md`:
|
|
202
|
+
|
|
203
|
+
```
|
|
204
|
+
# TRAPS.md — [project-name]
|
|
205
|
+
|
|
206
|
+
Project-specific pitfalls. Things that look right but break things.
|
|
207
|
+
Different from CONVENTIONS.md (what TO do) — this is what NOT to do and WHY.
|
|
208
|
+
|
|
209
|
+
---
|
|
210
|
+
|
|
211
|
+
## TRAP-01 — [short title]
|
|
212
|
+
**Pattern:** [what the bad code/approach looks like]
|
|
213
|
+
**Why it breaks:** [specific failure mode — not "might cause issues"]
|
|
214
|
+
**What to do instead:** [the correct approach]
|
|
215
|
+
**Learned:** [date]
|
|
216
|
+
```
|
|
217
|
+
|
|
218
|
+
If the code audit found no anti-patterns, create TRAPS.md with the header and no entries — it will grow over time as sessions discover new pitfalls.
|
|
219
|
+
|
|
164
220
|
## Phase 3 — Gap Interview
|
|
165
221
|
|
|
166
222
|
After generating drafts, ask the questions you couldn't answer from the code alone. Ask them conversationally — no more than 3 at a time. Cover:
|
|
@@ -189,7 +245,7 @@ Fill in the `❓ [UNKNOWN]` gaps based on their answers.
|
|
|
189
245
|
|
|
190
246
|
**Determine the project name** from the folder name or package.json name field.
|
|
191
247
|
|
|
192
|
-
**Write all
|
|
248
|
+
**Write all 12 docs to `~/.g2w/projects/[project-name]/`** — NOT inside the project codebase. Zero footprint on the actual project.
|
|
193
249
|
|
|
194
250
|
**Update `~/.g2w/CURRENT.md`** to set this project as active:
|
|
195
251
|
```
|
|
@@ -206,7 +262,7 @@ bring2life complete.
|
|
|
206
262
|
Project: [project-name]
|
|
207
263
|
Docs written to: ~/.g2w/projects/[project-name]/
|
|
208
264
|
|
|
209
|
-
Docs generated:
|
|
265
|
+
Docs generated: 12
|
|
210
266
|
Auto-filled from code: [X fields]
|
|
211
267
|
Filled from your answers: [X fields]
|
|
212
268
|
Still needs input: [list any remaining ❓ items]
|
package/skills/cut2it.md
CHANGED
|
@@ -5,6 +5,15 @@ description: Fast mode — for clearly scoped, low-risk changes. Skip planning c
|
|
|
5
5
|
|
|
6
6
|
# /g2w:cut2it
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `~/.g2w/projects/[active-project]/` exists (if not → tell user to run `/g2w:bring2life` first)
|
|
11
|
+
- No locked plan currently active (if a locked plan exists, use `/g2w:get2work` instead — cut2it skips planning)
|
|
12
|
+
|
|
13
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
14
|
+
|
|
15
|
+
---
|
|
16
|
+
|
|
8
17
|
No ceremony. You know what you're doing, it's small, and it's low-risk. Get it done.
|
|
9
18
|
|
|
10
19
|
## When to Use This
|
package/skills/get2work.md
CHANGED
|
@@ -5,6 +5,16 @@ description: Execute the locked plan — declare scope, build exactly what was p
|
|
|
5
5
|
|
|
6
6
|
# /g2w:get2work
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `~/.g2w/projects/[active-project]/` exists (if not → tell user to run `/g2w:bring2life` first)
|
|
11
|
+
- `PLAN.md` exists and is locked, OR `PLAN.md` is empty (triggers existing codebase flow)
|
|
12
|
+
- `CONVENTIONS.md` exists (cannot build without knowing the rules)
|
|
13
|
+
|
|
14
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
15
|
+
|
|
16
|
+
---
|
|
17
|
+
|
|
8
18
|
You are executing a G2W task. Read `~/.g2w/CURRENT.md` first to confirm the active project. All docs are at `~/.g2w/projects/[active-project]/`.
|
|
9
19
|
|
|
10
20
|
## Detect: New Project or Existing Codebase?
|
package/skills/ready2save.md
CHANGED
|
@@ -16,6 +16,12 @@ You are closing out this session. Leave the project in a state where any future
|
|
|
16
16
|
|
|
17
17
|
Do NOT ask the user what was done, what to save, or what decisions were made. You witnessed the entire session — extract it yourself and present it. The user confirms or corrects your summary. Never ask them to recall.
|
|
18
18
|
|
|
19
|
+
1b. **Harvest traps** — If any "don't do this again" lessons were learned this session, write them as new TRAP entries in `~/.g2w/projects/[active-project]/TRAPS.md`. Use the next available TRAP number. These are patterns that looked right but broke things — they survive context clears so the same mistake never happens twice.
|
|
20
|
+
|
|
21
|
+
1c. **Update golden cases** — If any golden cases in `TESTING.md` were verified this session (by Inspector or true2plan), update their "Last Verified" column with today's date.
|
|
22
|
+
|
|
23
|
+
1d. **Record health score** — If `/g2w:true2dagame` was run this session, include the overall score (e.g., "Health: 4.2/5") in the session notes so `back2it` can reference it next session.
|
|
24
|
+
|
|
19
25
|
2. **Identify the active project** — Read `~/.g2w/CURRENT.md` (Windows: `C:/Users/[username]/.g2w/CURRENT.md`). If it has an `active:` line, that's the project. Use it. Do NOT ask.
|
|
20
26
|
- Only if the file doesn't exist or has no `active:` line: list folders in `~/.g2w/projects/` and ask which one. This is the ONLY scenario where you ask.
|
|
21
27
|
|
package/skills/the-builder.md
CHANGED
|
@@ -5,6 +5,16 @@ description: Use when executing a G2W locked plan. Triggered by The Leader at Ph
|
|
|
5
5
|
|
|
6
6
|
# /g2w:the-builder
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `PLAN.md` exists and is locked (look for "Challenger approved" marker — if missing, Challenger hasn't signed off yet)
|
|
11
|
+
- `CONVENTIONS.md` exists (cannot build without knowing the rules)
|
|
12
|
+
- Scope declaration approved by user (do not build without it)
|
|
13
|
+
|
|
14
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
15
|
+
|
|
16
|
+
---
|
|
17
|
+
|
|
8
18
|
You are The Builder. You build exactly what the locked plan says. Nothing more. Nothing less.
|
|
9
19
|
|
|
10
20
|
## Prime Directive
|
|
@@ -46,7 +56,7 @@ Do not improvise. Do not make a judgment call and note it as "should be fine." S
|
|
|
46
56
|
|
|
47
57
|
1. **Read the locked `PLAN.md`** — this is your only instruction set
|
|
48
58
|
|
|
49
|
-
2. **Read `CONVENTIONS.md`** — follow existing code patterns
|
|
59
|
+
2. **Read `CONVENTIONS.md`** — follow existing code patterns. Check the Context Budget section: for any file listed as "heavy," read only the sections the plan specifies. Do not read whole files unless the plan says to.
|
|
50
60
|
|
|
51
61
|
3. **Confirm scope with The Leader before writing a single line**:
|
|
52
62
|
```
|
package/skills/the-challenger.md
CHANGED
|
@@ -5,6 +5,15 @@ description: Use when a G2W plan needs adversarial review before Builder touches
|
|
|
5
5
|
|
|
6
6
|
# /g2w:the-challenger
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `PLAN.md` exists in `~/.g2w/projects/[active-project]/` and contains no TBD, TODO, or ❓ markers (Visionary must finish first)
|
|
11
|
+
- `ERRORS.md` exists (code audit must have run — if missing, note it as a finding)
|
|
12
|
+
|
|
13
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
14
|
+
|
|
15
|
+
---
|
|
16
|
+
|
|
8
17
|
You are The Challenger. Your job is not to help. Your job is to break the plan before the code does.
|
|
9
18
|
|
|
10
19
|
## Prime Directive
|
|
@@ -38,6 +47,8 @@ Attack every dimension:
|
|
|
38
47
|
|
|
39
48
|
**Edge cases in the test matrix** — Are the manual tests exhaustive, or do they only cover the happy path?
|
|
40
49
|
|
|
50
|
+
**Known traps** — Read `TRAPS.md` from `~/.g2w/projects/[active-project]/`. Does the plan use any pattern that has already been identified as a project-specific pitfall? If TRAPS.md doesn't exist, note it as a finding.
|
|
51
|
+
|
|
41
52
|
## Process
|
|
42
53
|
|
|
43
54
|
1. **Read `PLAN.md`** from `~/.g2w/projects/[active-project]/`
|
package/skills/the-inspector.md
CHANGED
|
@@ -5,6 +5,16 @@ description: Use when verifying a G2W build against the locked plan. Triggered b
|
|
|
5
5
|
|
|
6
6
|
# /g2w:the-inspector
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- Build is complete (Builder has handed off)
|
|
11
|
+
- `PLAN.md` exists and is locked in `~/.g2w/projects/[active-project]/`
|
|
12
|
+
- `TESTING.md` exists (cannot verify without test criteria)
|
|
13
|
+
|
|
14
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
15
|
+
|
|
16
|
+
---
|
|
17
|
+
|
|
8
18
|
You are The Inspector. You verify that what was built matches what was planned — completely, not approximately.
|
|
9
19
|
|
|
10
20
|
## Prime Directive
|
|
@@ -37,6 +47,8 @@ Not done means:
|
|
|
37
47
|
|
|
38
48
|
4. **Error handling** — every error case specified in the plan must be verified. "It probably handles it" is not verification.
|
|
39
49
|
|
|
50
|
+
4b. **Golden Cases** — read the Golden Cases section of `TESTING.md`. Run every golden case, regardless of whether the current task touched that area. A golden case failure is an automatic FAIL — the build broke something fundamental.
|
|
51
|
+
|
|
40
52
|
5. **Conventions** — read `CONVENTIONS.md`. Does the code follow existing patterns? If not, flag it.
|
|
41
53
|
|
|
42
54
|
## Hard Stop Conditions
|
|
@@ -44,6 +56,7 @@ Not done means:
|
|
|
44
56
|
Do not pass anything to The Leader until:
|
|
45
57
|
- All plan items are verified
|
|
46
58
|
- All edge cases from the test matrix pass
|
|
59
|
+
- All golden cases pass (not just the ones "related" to this change)
|
|
47
60
|
- The user has confirmed behavior in their actual environment
|
|
48
61
|
|
|
49
62
|
If the user says "I trust it" — that's not verification. Ask them to run the specific scenarios.
|
package/skills/the-visionary.md
CHANGED
|
@@ -5,6 +5,16 @@ description: Use when a G2W task needs a complete, locked implementation plan wr
|
|
|
5
5
|
|
|
6
6
|
# /g2w:the-visionary
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `ARCHITECTURE.md` exists in `~/.g2w/projects/[active-project]/` (cannot plan without knowing the system)
|
|
11
|
+
- `CONVENTIONS.md` exists (cannot plan without knowing the rules)
|
|
12
|
+
- Task defined in one sentence by user or Leader
|
|
13
|
+
|
|
14
|
+
**Check these first. If any prerequisite is not met, stop and tell the user what's missing and which skill to run first. Do not proceed.**
|
|
15
|
+
|
|
16
|
+
---
|
|
17
|
+
|
|
8
18
|
You are The Visionary. Your only job: write a complete plan. Not a starting point. Not a skeleton. A contract.
|
|
9
19
|
|
|
10
20
|
## Prime Directive
|
|
@@ -33,7 +43,7 @@ Every decision made. Nothing left to the Builder's judgment:
|
|
|
33
43
|
|
|
34
44
|
## Process
|
|
35
45
|
|
|
36
|
-
1. **Read the project docs first** — `ARCHITECTURE.md`, `CONVENTIONS.md`, `FEATURES.md` from `~/.g2w/projects/[active-project]/`. Know the codebase before you write a single line of the plan.
|
|
46
|
+
1. **Read the project docs first** — `ARCHITECTURE.md`, `CONVENTIONS.md`, `FEATURES.md`, and `TRAPS.md` from `~/.g2w/projects/[active-project]/`. Know the codebase — and what NOT to do — before you write a single line of the plan.
|
|
37
47
|
|
|
38
48
|
2. **Resolve every ambiguity upfront** — if you can't determine something from the docs alone, ask the user NOW. Do not leave it as a question in the plan.
|
|
39
49
|
|
|
@@ -44,6 +54,7 @@ Every decision made. Nothing left to the Builder's judgment:
|
|
|
44
54
|
- Files to create (with full paths)
|
|
45
55
|
- Files to modify (with full paths)
|
|
46
56
|
- For each file: exact changes, exact function signatures, exact logic
|
|
57
|
+
- For heavy files (check Context Budget in CONVENTIONS.md): specify which sections the Builder should read — not "read PluginProcessor.cpp" but "read lines 800-900 of PluginProcessor.cpp (the processBlock grain loop)"
|
|
47
58
|
- Test matrix: every scenario the Builder must verify manually
|
|
48
59
|
|
|
49
60
|
4. **Self-audit before handing off:**
|
package/skills/true2dagame.md
CHANGED
|
@@ -5,6 +5,14 @@ description: Full G2W system health check — are docs in sync, is the plan lock
|
|
|
5
5
|
|
|
6
6
|
# /g2w:true2dagame
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `~/.g2w/projects/[active-project]/` exists (at least `/g2w:bring2life` has been run)
|
|
11
|
+
|
|
12
|
+
**Check this first. If the project docs don't exist, tell the user to run `/g2w:bring2life` first. Do not proceed.**
|
|
13
|
+
|
|
14
|
+
---
|
|
15
|
+
|
|
8
16
|
You are auditing the project's G2W health. No rubber-stamping. Find what's broken before it breaks you.
|
|
9
17
|
|
|
10
18
|
## Checks
|
|
@@ -58,44 +66,99 @@ Flag any doc that looks stale or inconsistent with what you can observe.
|
|
|
58
66
|
|
|
59
67
|
---
|
|
60
68
|
|
|
69
|
+
## Scoring Rubric
|
|
70
|
+
|
|
71
|
+
After running each check, score it 1-5. Use the criteria below — no vibes, no rounding up.
|
|
72
|
+
|
|
73
|
+
### Docs (1-5)
|
|
74
|
+
- **5** — All docs match code, updated this session or last
|
|
75
|
+
- **4** — All docs match code, updated within last 3 sessions
|
|
76
|
+
- **3** — 1 doc stale or missing a section
|
|
77
|
+
- **2** — 2+ docs stale or contradicting the code
|
|
78
|
+
- **1** — Docs are fiction — major drift from reality
|
|
79
|
+
|
|
80
|
+
### Plan (1-5)
|
|
81
|
+
- **5** — Locked plan exists, all files accurate, clear definition of done
|
|
82
|
+
- **4** — Locked plan exists, minor file renames since lock
|
|
83
|
+
- **3** — Plan exists but is a draft (not locked)
|
|
84
|
+
- **2** — Plan exists but has TBD/TODO markers
|
|
85
|
+
- **1** — No plan, or plan references files/functions that don't exist
|
|
86
|
+
|
|
87
|
+
### Errors (1-5)
|
|
88
|
+
- **5** — No unresolved critical/high bugs, golden cases all passing
|
|
89
|
+
- **4** — Unresolved bugs exist but all are medium/low
|
|
90
|
+
- **3** — 1 high-severity bug unresolved
|
|
91
|
+
- **2** — 1 critical bug unresolved
|
|
92
|
+
- **1** — Multiple critical bugs unresolved or ERRORS.md doesn't exist
|
|
93
|
+
|
|
94
|
+
### Freshness (1-5)
|
|
95
|
+
- **5** — CURRENT.md updated this session, all sections accurate
|
|
96
|
+
- **4** — Updated last session, still accurate
|
|
97
|
+
- **3** — "In Progress" is stale but "Next" is correct
|
|
98
|
+
- **2** — Both "In Progress" and "Next" are wrong
|
|
99
|
+
- **1** — CURRENT.md is from a different phase of the project entirely
|
|
100
|
+
|
|
101
|
+
### Git (1-5)
|
|
102
|
+
- **5** — Clean tree, last commit follows G2W format
|
|
103
|
+
- **4** — Clean tree, last commit message is informal
|
|
104
|
+
- **3** — Uncommitted changes exist but they're small
|
|
105
|
+
- **2** — Large uncommitted changes or untracked files
|
|
106
|
+
- **1** — Uncommitted verified work sitting for 2+ sessions
|
|
107
|
+
|
|
108
|
+
---
|
|
109
|
+
|
|
110
|
+
## Action Map
|
|
111
|
+
|
|
112
|
+
| Score | Action |
|
|
113
|
+
|-------|--------|
|
|
114
|
+
| 1 in any dimension | **HARD STOP** — fix this before any other work |
|
|
115
|
+
| 2 in Docs | Run through each stale doc, update from code |
|
|
116
|
+
| 2 in Plan | Send plan back to Visionary for completion |
|
|
117
|
+
| 2 in Errors | Triage with user — which critical bugs to fix first |
|
|
118
|
+
| 2 in Freshness | Rewrite CURRENT.md from scratch based on actual state |
|
|
119
|
+
| 2 in Git | Commit or stash — decide now |
|
|
120
|
+
| 3 in any dimension | **WARNING** — surface to user, ask if they want to address now |
|
|
121
|
+
| 4-5 in all dimensions | "System is healthy. Ready to build." |
|
|
122
|
+
|
|
123
|
+
---
|
|
124
|
+
|
|
61
125
|
## Output Format
|
|
62
126
|
|
|
63
127
|
```
|
|
64
128
|
G2W Health Check — [date]
|
|
65
129
|
|
|
66
|
-
Docs:
|
|
67
|
-
|
|
68
|
-
|
|
69
|
-
|
|
70
|
-
|
|
130
|
+
Docs: ████████████████░░░░ 4/5
|
|
131
|
+
Plan: ████████████████████ 5/5
|
|
132
|
+
Errors: ████████████░░░░░░░░ 3/5
|
|
133
|
+
Freshness: ████████████████████ 5/5
|
|
134
|
+
Git: ████████████████░░░░ 4/5
|
|
71
135
|
|
|
72
|
-
|
|
73
|
-
|
|
74
|
-
✅ Definition of done present
|
|
75
|
-
⚠️ 2 declared files may have been renamed since lock
|
|
76
|
-
|
|
77
|
-
Errors:
|
|
78
|
-
⚠️ ERR-01 — CRITICAL, unresolved
|
|
79
|
-
✅ No stale "in progress" errors
|
|
136
|
+
Overall: 4.2/5
|
|
137
|
+
```
|
|
80
138
|
|
|
81
|
-
|
|
82
|
-
✅ Looks fresh
|
|
139
|
+
Then below the scores, show the detail for anything scoring 3 or below:
|
|
83
140
|
|
|
84
|
-
|
|
85
|
-
|
|
86
|
-
|
|
141
|
+
```
|
|
142
|
+
⚠️ Errors: 3/5 — ERR-01 marked HIGH, unresolved, no activity in 3 sessions
|
|
143
|
+
→ Action: Triage with user — fix or downgrade severity
|
|
87
144
|
|
|
88
|
-
|
|
145
|
+
⚠️ Docs: ... (if applicable)
|
|
89
146
|
```
|
|
90
147
|
|
|
148
|
+
If overall is 4.0+ with no dimension below 3: "System is healthy. Ready to build."
|
|
149
|
+
|
|
150
|
+
---
|
|
151
|
+
|
|
91
152
|
## After the Report
|
|
92
153
|
|
|
93
|
-
- If
|
|
94
|
-
- If
|
|
95
|
-
- If
|
|
154
|
+
- If any dimension scores **1** → hard stop, fix before any other work
|
|
155
|
+
- If any dimension scores **2** → fix now, use the action map
|
|
156
|
+
- If any dimension scores **3** → surface to user, ask if they want to address now or later
|
|
157
|
+
- If all dimensions score **4-5** → "System is healthy. Ready to build."
|
|
96
158
|
|
|
97
159
|
## Rules
|
|
98
160
|
|
|
99
161
|
- Don't just check if files exist — read enough to assess if they're accurate
|
|
100
162
|
- A doc that exists but is 6 months stale is worse than no doc (false confidence)
|
|
101
|
-
- If you can't check something (no access, build required), say so — don't guess
|
|
163
|
+
- If you can't check something (no access, build required), say so — don't guess a 5
|
|
164
|
+
- Score honestly. A 4 that should be a 2 is worse than the bug it hides.
|
package/skills/true2plan.md
CHANGED
|
@@ -5,6 +5,14 @@ description: Verify that what was built matches what the plan declared — PASS/
|
|
|
5
5
|
|
|
6
6
|
# /g2w:true2plan
|
|
7
7
|
|
|
8
|
+
## Prerequisites
|
|
9
|
+
|
|
10
|
+
- `PLAN.md` exists and is locked in `~/.g2w/projects/[active-project]/` (cannot verify against a plan that doesn't exist)
|
|
11
|
+
|
|
12
|
+
**Check this first. If no locked plan exists, tell the user there's nothing to verify against. Do not proceed.**
|
|
13
|
+
|
|
14
|
+
---
|
|
15
|
+
|
|
8
16
|
You are the verifier. Your job is not to say "looks good" — it is to confirm that every specific thing the plan declared was actually done, correctly.
|
|
9
17
|
|
|
10
18
|
## Steps
|
|
@@ -26,10 +34,13 @@ You are the verifier. Your job is not to say "looks good" — it is to confirm t
|
|
|
26
34
|
|
|
27
35
|
Run each verification step from the plan. Record actual result vs expected.
|
|
28
36
|
|
|
29
|
-
5. **Regression check:**
|
|
37
|
+
5. **Regression check (Golden Cases):**
|
|
30
38
|
|
|
31
|
-
-
|
|
32
|
-
-
|
|
39
|
+
- Read the Golden Cases section of `TESTING.md`
|
|
40
|
+
- Run every golden case — not just the ones that seem related to this change
|
|
41
|
+
- Any golden case failure = automatic FAIL, even if all plan items pass
|
|
42
|
+
- Update "Last Verified" column for each case that passes
|
|
43
|
+
- Also check: what was working before this change? Is that list still intact? → PASS / FAIL
|
|
33
44
|
|
|
34
45
|
6. **Output the report:**
|
|
35
46
|
```
|