Skip to content

Commit a90c417

Browse files
bniladridasbniladridas
committed
fix: replace em dashes in UX docs
1 parent 6adbc68 commit a90c417

2 files changed

Lines changed: 78 additions & 78 deletions

File tree

‎docs/ux/query_journey_audit.md‎

Lines changed: 35 additions & 35 deletions
Original file line numberDiff line numberDiff line change
@@ -29,14 +29,14 @@ The most fragmented intent in the system.
2929

3030
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
3131
|---|---|---|---|---|---|---|---|---|
32-
| 01 | Show modified files | `"show modified files"` | CommitHistory | 1 (`git status`) | ~100ms | `Investigating... → ✓ complete → answer` | OK | — |
33-
| 02 | Check changed files | `"check changed files"` | CommitHistory | 1 (`git status`) | ~100ms | Same as above | OK | — |
34-
| 03 | What files changed | `"what files changed"` | CommitHistory | 1 (`git status`) | ~100ms | Same as above | OK | — |
32+
| 01 | Show modified files | `"show modified files"` | CommitHistory | 1 (`git status`) | ~100ms | `Investigating... → ✓ complete → answer` | OK | -- |
33+
| 02 | Check changed files | `"check changed files"` | CommitHistory | 1 (`git status`) | ~100ms | Same as above | OK | -- |
34+
| 03 | What files changed | `"what files changed"` | CommitHistory | 1 (`git status`) | ~100ms | Same as above | OK | -- |
3535
| 04 | **Show uncommitted changes** | `"show me uncommitted changes"` | **GeneralChat** | **0** | **0ms** | `cursor\n\n[AI answers from knowledge, no tools run]` | **Wrong answer. No git status run. AI hallucinates.** | **Critical** |
3636
| 05 | **What files are modified** | `"what files are modified"` | **CodebaseQuery** | 2-3 (find → grep → read) | ~2s | `Investigating... → find/grep/read → "Insufficient evidence"` | **Wrong investigation path. Runs code search on a status query.** | **Critical** |
3737
| 06 | **Did I edit anything** | `"did I edit anything"` | **GeneralChat** | **0** | **0ms** | `cursor\n\n[AI general answer]` | **No tools run. AI guesses.** | **Critical** |
38-
| 07 | Current status | `"git status"` | CommitHistory | 1 (`git status`) | ~100ms | Status shown | OK | — |
39-
| 08 | What branch | `"what branch am I on"` | CommitHistory | 1 (`git status`) | ~100ms | Branch shown | OK | — |
38+
| 07 | Current status | `"git status"` | CommitHistory | 1 (`git status`) | ~100ms | Status shown | OK | -- |
39+
| 08 | What branch | `"what branch am I on"` | CommitHistory | 1 (`git status`) | ~100ms | Branch shown | OK | -- |
4040

4141
**Pattern:** Three phrasings of the same intent route to three different GoalTypes, two of which produce wrong results. The keyword list for status queries in CommitHistory is long but has gaps. "Uncommitted changes" and "are modified" fall through.
4242

@@ -46,54 +46,54 @@ The most fragmented intent in the system.
4646

4747
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
4848
|---|---|---|---|---|---|---|---|---|
49-
| 09 | Last commit | `"what is the last commit"` | CommitHistory | 2 (`git log -10` + `git log -1`) | ~200ms | `Investigating... → ✓ complete → answer` | OK | — |
50-
| 10 | Recent commits | `"show recent commits"` | CommitHistory | 2 | ~200ms | Same | OK | — |
49+
| 09 | Last commit | `"what is the last commit"` | CommitHistory | 2 (`git log -10` + `git log -1`) | ~200ms | `Investigating... → ✓ complete → answer` | OK | -- |
50+
| 10 | Recent commits | `"show recent commits"` | CommitHistory | 2 | ~200ms | Same | OK | -- |
5151
| 11 | What changed | `"what changed"` | CommitHistory | 1 (`git status`) | ~100ms | Status shown (not log!) | **Minor: "what changed" does `git status`, not `git log`. Shows working tree diff, not commit history.** | Low |
5252
| 12 | What changed last week | `"what changed last week"` | CommitHistory | 1 (`git status`) | ~100ms | Status shown (ignores "last week") | **Ignored time qualifier. `git status` has no date concept.** | Medium |
53-
| 13 | Commit history | `"commit history"` | CommitHistory | 2 (`git log -10` + `git log -1`) | ~200ms | Same | OK | — |
54-
| 14 | Check the files we changed | `"check the files we changed"` | CommitHistory | 1 (`git status`) | ~100ms | Same | OK | — |
53+
| 13 | Commit history | `"commit history"` | CommitHistory | 2 (`git log -10` + `git log -1`) | ~200ms | Same | OK | -- |
54+
| 14 | Check the files we changed | `"check the files we changed"` | CommitHistory | 1 (`git status`) | ~100ms | Same | OK | -- |
5555

56-
**Pattern:** Lines 11-12 show the limit of keyword matching — "what changed" always means `git status`, not `git log --since=last.week`. The planner has no concept of time qualifiers.
56+
**Pattern:** Lines 11-12 show the limit of keyword matching -- "what changed" always means `git status`, not `git log --since=last.week`. The planner has no concept of time qualifiers.
5757

5858
---
5959

6060
## 3. ARCHITECTURE & DESIGN QUESTIONS
6161

6262
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
6363
|---|---|---|---|---|---|---|---|---|
64-
| 15 | How is this designed | `"how is this agent designed"` | CodebaseOverview | 2 (discovery + read README) | ~3s | `Investigating... → ✓ complete → answer` | OK | — |
65-
| 16 | Explain architecture | `"explain the architecture"` | CodebaseOverview | 2 | ~3s | Same | OK | — |
64+
| 15 | How is this designed | `"how is this agent designed"` | CodebaseOverview | 2 (discovery + read README) | ~3s | `Investigating... → ✓ complete → answer` | OK | -- |
65+
| 16 | Explain architecture | `"explain the architecture"` | CodebaseOverview | 2 | ~3s | Same | OK | -- |
6666
| 17 | **Tell me how repo investigation works** | `"tell me how repository investigation works"` | **CodebaseQuery** | 2-3 (find/grep + read) | ~2s | Runs code search instead of overview | **Wrong path. Should be CodebaseOverview (high-level). Runs file search on implementation.** | **High** |
6767
| 18 | **How does the build system work** | `"how does the build system work"` | **CodebaseQuery** | 2-3 (find/grep + read about "build") | ~2s | File search on "build" instead of project overview | **Wrong path. Should describe build system, not grep for "build".** | High |
6868
| 19 | Review architecture | `"review the architecture"` | ArchitectureReview | **11** (full audit) | ~15-30s | `Investigating...` (stalls for 15-30s with no progress change) | **No progress updates during 11-tool audit. User sees frozen terminal.** | Medium |
6969
| 20 | Review codebase | `"review codebase"` | ArchitectureReview | 11 | ~15-30s | Same as above | **Same stall problem** | Medium |
7070
| 21 | Explain this codebase | `"explain this codebase"` | **CodebaseQuery** | 2-3 | ~2s | Code search instead of overview | **Wrong path. Should be CodebaseOverview.** | High |
7171
| 22 | What isthis repository | `"what is this repository"` | **CodebaseQuery** | 2-3 | ~2s | Code search instead of overview | **Wrong path.** | High |
7272

73-
**Pattern:** Queries starting with "tell me how", "how does the", "explain this" — ambiguous between architecture explanation and code search. The current classifier sends them to CodebaseQuery. A human would recognize these as overview questions.
73+
**Pattern:** Queries starting with "tell me how", "how does the", "explain this" -- ambiguous between architecture explanation and code search. The current classifier sends them to CodebaseQuery. A human would recognize these as overview questions.
7474

7575
---
7676

7777
## 4. CODE SEARCH (Working Well)
7878

7979
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
8080
|---|---|---|---|---|---|---|---|---|
81-
| 23 | Find CommandRouter | `"find CommandRouter"` | CodebaseQuery | 2 (find + read) | ~1s | `Investigating... → ✓ complete → answer` | OK | — |
82-
| 24 | Grep for Agent | `"grep Agent"` | CodebaseQuery | 2-3 (find+grep+read) | ~2s | Same | OK | — |
83-
| 25 | Where is ReplayService used | `"where is ReplayService used"` | CodebaseQuery | 2 (find + read) | ~1s | Same | OK | — |
84-
| 26 | How does auth work (code-level) | `"how does auth work in this project"` | CodebaseQuery | 2-3 | ~2s | Same | OK | — |
85-
| 27 | **Find binary** | `"find the cursor binary"` | CodebaseQuery | 2-3 (find fails, grep fallback) | ~2s | Now works (was a retrieval bug, fixed) | Resolved | — |
86-
| 28 | Read file | `"read file src/main.cpp"` | CodebaseQuery | 1 (`read src/main.cpp`) | ~100ms | OK | OK | — |
81+
| 23 | Find CommandRouter | `"find CommandRouter"` | CodebaseQuery | 2 (find + read) | ~1s | `Investigating... → ✓ complete → answer` | OK | -- |
82+
| 24 | Grep for Agent | `"grep Agent"` | CodebaseQuery | 2-3 (find+grep+read) | ~2s | Same | OK | -- |
83+
| 25 | Where is ReplayService used | `"where is ReplayService used"` | CodebaseQuery | 2 (find + read) | ~1s | Same | OK | -- |
84+
| 26 | How does auth work (code-level) | `"how does auth work in this project"` | CodebaseQuery | 2-3 | ~2s | Same | OK | -- |
85+
| 27 | **Find binary** | `"find the cursor binary"` | CodebaseQuery | 2-3 (find fails, grep fallback) | ~2s | Now works (was a retrieval bug, fixed) | Resolved | -- |
86+
| 28 | Read file | `"read file src/main.cpp"` | CodebaseQuery | 1 (`read src/main.cpp`) | ~100ms | OK | OK | -- |
8787

8888
---
8989

9090
## 5. CI / GITHUB INVESTIGATION
9191

9292
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
9393
|---|---|---|---|---|---|---|---|---|
94-
| 29 | Why did CI fail | `"why did CI fail"` | CICheck | 1-3 (gh + optional grep/read) | ~3-10s | `Investigating... → ✓ complete → answer` | OK | — |
95-
| 30 | Check workflow | `"check my CI workflow"` | CICheck | 1 (gh list) | ~3s | Same | OK | — |
96-
| 31 | Investigate run URL | paste of `github.com/.../actions/runs/12345` | GitHubInvestigation | 2 (gh run view + logs) | ~5-15s | Same | OK | — |
94+
| 29 | Why did CI fail | `"why did CI fail"` | CICheck | 1-3 (gh + optional grep/read) | ~3-10s | `Investigating... → ✓ complete → answer` | OK | -- |
95+
| 30 | Check workflow | `"check my CI workflow"` | CICheck | 1 (gh list) | ~3s | Same | OK | -- |
96+
| 31 | Investigate run URL | paste of `github.com/.../actions/runs/12345` | GitHubInvestigation | 2 (gh run view + logs) | ~5-15s | Same | OK | -- |
9797
| 32 | **Check this log** | `"can you check this log https://..."` | **GeneralChat** | **0** | **0ms** | AI answers without running `gh` | **Wrong. Should be GitHubInvestigation. The URL is present but "check this log" is a GeneralChat pattern that matches before the URL check.** | **Critical** |
9898

9999
**Pattern:** Line 32 is a priority-ordering bug. The GeneralChat exclusion check (Level 8) fires before the GitHubInvestigation URL check (Level 6) if the query starts with general-language patterns. The order matters: URL-containing queries should be checked early regardless of surrounding language.
@@ -104,9 +104,9 @@ The most fragmented intent in the system.
104104

105105
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
106106
|---|---|---|---|---|---|---|---|---|
107-
| 33 | What model am I on | `"what model am i on"` | SessionState | 0 | ~0ms | `cursor\n\n[state answer]` | OK | — |
108-
| 34 | Am I online | `"am i online"` | SessionState | 0 | ~0ms | OK | OK | — |
109-
| 35 | What provider | `"what provider am i using"` | SessionState | 0 | ~0ms | OK | OK | — |
107+
| 33 | What model am I on | `"what model am i on"` | SessionState | 0 | ~0ms | `cursor\n\n[state answer]` | OK | -- |
108+
| 34 | Am I online | `"am i online"` | SessionState | 0 | ~0ms | OK | OK | -- |
109+
| 35 | What provider | `"what provider am i using"` | SessionState | 0 | ~0ms | OK | OK | -- |
110110

111111
---
112112

@@ -127,11 +127,11 @@ The most fragmented intent in the system.
127127

128128
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
129129
|---|---|---|---|---|---|---|---|---|
130-
| 40 | How are you | `"how are you"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | — |
131-
| 41 | What can you do | `"what can you do"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | — |
132-
| 42 | How do I install python | `"how do I install python"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | — |
133-
| 43 | **What is the difference** | `"what is the difference between X and Y"` | GeneralChat | 0 | ~0ms | AI chat answer | OK (correct) | — |
134-
| 44 | Hello | `"hello"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | — |
130+
| 40 | How are you | `"how are you"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | -- |
131+
| 41 | What can you do | `"what can you do"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | -- |
132+
| 42 | How do I install python | `"how do I install python"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | -- |
133+
| 43 | **What is the difference** | `"what is the difference between X and Y"` | GeneralChat | 0 | ~0ms | AI chat answer | OK (correct) | -- |
134+
| 44 | Hello | `"hello"` | GeneralChat | 0 | ~0ms | AI chat answer | OK | -- |
135135

136136
---
137137

@@ -140,11 +140,11 @@ The most fragmented intent in the system.
140140
| # | Task | Input | GoalType | Tools | Time | What user sees | Friction | Sev |
141141
|---|---|---|---|---|---|---|---|---|
142142
| 45 | Explicit git prefix | `"git:status"` | (direct command) | 1 | ~100ms | Direct output | **Only if user knows `git:` prefix exists. Not discoverable.** | Medium |
143-
| 46 | Typo: comit | `"last comit"` | CommitHistory | 2 | ~200ms | OK (typo in keyword list) | OK | — |
144-
| 47 | Typo: codbease | `"codbease overview"` | CodebaseOverview | 2 | ~3s | Normalized by command_router before classify_goal() | OK (if routed through command_router) | — |
143+
| 46 | Typo: comit | `"last comit"` | CommitHistory | 2 | ~200ms | OK (typo in keyword list) | OK | -- |
144+
| 47 | Typo: codbease | `"codbease overview"` | CodebaseOverview | 2 | ~3s | Normalized by command_router before classify_goal() | OK (if routed through command_router) | -- |
145145
| 48 | **Ambiguous: plan** | `"plan the implementation"` | **CodeChange** | 5 | ~30s+ | Full investigation | **"Plan" is not in any keyword list. Might miss CodeChange path.** | Medium |
146-
| 49 | Long query | 200-character multi-sentence question | CodebaseQuery | 2-3 | ~2s | Works (keyword matching doesn't penalize length) | OK | — |
147-
| 50 | Multi-step: find + status | `"show me the last commit and the current branch"` | CommitHistory | 2 (log + status) | ~200ms | Both shown | OK (both are CommitHistory) | — |
146+
| 49 | Long query | 200-character multi-sentence question | CodebaseQuery | 2-3 | ~2s | Works (keyword matching doesn't penalize length) | OK | -- |
147+
| 50 | Multi-step: find + status | `"show me the last commit and the current branch"` | CommitHistory | 2 (log + status) | ~200ms | Both shown | OK (both are CommitHistory) | -- |
148148

149149
---
150150

@@ -223,6 +223,6 @@ Every UX change must be verified against these journeys before closing:
223223
| 13 | Linux terminal (gnome, xterm) | All features work on common Linux terminals |
224224
| 14 | macOS terminal (Terminal.app, iTerm2) | All features work on common macOS terminals |
225225
| 15 | First launch | Startup hint shown, disappears after first prompt |
226-
| 16 | Recovery — low confidence | Planner attempts recovery, user sees progress, no tool internals |
226+
| 16 | Recovery -- low confidence | Planner attempts recovery, user sees progress, no tool internals |
227227

228228
Each journey must pass before the change is considered complete. If a journey produces unexpected output or terminal corruption, the change is not ready.

0 commit comments

Comments
 (0)