ultracode-for-codex 0.3.1 → 0.3.2
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
CHANGED
|
@@ -17,7 +17,15 @@ npm install --save-dev ultracode-for-codex
|
|
|
17
17
|
npm exec -- ultracode-for-codex --llm-guide
|
|
18
18
|
```
|
|
19
19
|
|
|
20
|
-
|
|
20
|
+
Or install the CLI globally:
|
|
21
|
+
|
|
22
|
+
```bash
|
|
23
|
+
npm install -g ultracode-for-codex
|
|
24
|
+
ultracode-for-codex --version
|
|
25
|
+
ultracode-for-codex --llm-guide
|
|
26
|
+
```
|
|
27
|
+
|
|
28
|
+
Install the Codex skill commands from a project install:
|
|
21
29
|
|
|
22
30
|
```bash
|
|
23
31
|
mkdir -p "${CODEX_HOME:-$HOME/.codex}/skills"
|
|
@@ -27,6 +35,17 @@ cp -R ./node_modules/ultracode-for-codex/skills/ultracode-for-codex-cli \
|
|
|
27
35
|
"${CODEX_HOME:-$HOME/.codex}/skills/"
|
|
28
36
|
```
|
|
29
37
|
|
|
38
|
+
Or install the skill commands from a global npm install:
|
|
39
|
+
|
|
40
|
+
```bash
|
|
41
|
+
mkdir -p "${CODEX_HOME:-$HOME/.codex}/skills"
|
|
42
|
+
GLOBAL_NODE_MODULES="$(npm root -g)"
|
|
43
|
+
cp -R "$GLOBAL_NODE_MODULES/ultracode-for-codex/skills/ultracode-for-codex" \
|
|
44
|
+
"${CODEX_HOME:-$HOME/.codex}/skills/"
|
|
45
|
+
cp -R "$GLOBAL_NODE_MODULES/ultracode-for-codex/skills/ultracode-for-codex-cli" \
|
|
46
|
+
"${CODEX_HOME:-$HOME/.codex}/skills/"
|
|
47
|
+
```
|
|
48
|
+
|
|
30
49
|
Build and verify a local installable tarball from a source checkout:
|
|
31
50
|
|
|
32
51
|
```bash
|
package/package.json
CHANGED
|
@@ -51,7 +51,10 @@ Phase Inspect - 3 agents
|
|
|
51
51
|
file inspection, test execution, or integration planning.
|
|
52
52
|
7. As agents complete, report progress with a visual snapshot rather than a
|
|
53
53
|
dense sentence. Use the Default Live Snapshot golden shape from
|
|
54
|
-
`references/progress-visuals.md` by default.
|
|
54
|
+
`references/progress-visuals.md` by default. Select task-specific additions
|
|
55
|
+
from the Situation Choice Matrix in that reference. Within one user
|
|
56
|
+
request, keep a cumulative ledger: do not remove completed rows from later
|
|
57
|
+
snapshots; update their status and append newly discovered work below them.
|
|
55
58
|
|
|
56
59
|
```text
|
|
57
60
|
Phase E2E Validate
|
|
@@ -4,6 +4,67 @@ Use these golden examples for Codex-native Ultracode progress updates. The goal
|
|
|
4
4
|
is fast visual parsing in chat while staying portable across terminals and
|
|
5
5
|
renderers. Prefer ASCII symbols and short labels.
|
|
6
6
|
|
|
7
|
+
## Research Pattern Map
|
|
8
|
+
|
|
9
|
+
These patterns are adapted from established CLI/TUI conventions:
|
|
10
|
+
|
|
11
|
+
- multi-task progress: Rich-style multiple task rows with progress metadata;
|
|
12
|
+
- test progress: pytest-style progress/count/timing summaries;
|
|
13
|
+
- async build logs: Docker BuildKit-style numbered steps and plain progress;
|
|
14
|
+
- folded details: GitHub Actions-style grouped log sections;
|
|
15
|
+
- change plans: Terraform-style add/update/keep/destroy action symbols;
|
|
16
|
+
- rollout watches: Kubernetes rollout-style "N out of M" convergence messages;
|
|
17
|
+
- audit tables: npm audit-style severity/package/path/remediation columns.
|
|
18
|
+
|
|
19
|
+
Map those patterns to Ultracode situations instead of using one universal
|
|
20
|
+
status format.
|
|
21
|
+
|
|
22
|
+
## Situation Choice Matrix
|
|
23
|
+
|
|
24
|
+
Choose one row per task situation. Each row has at most three user-facing
|
|
25
|
+
shapes: primary, support, and finish. Do not present profile names to the user.
|
|
26
|
+
If a task mixes situations, choose the dominant row for live progress and borrow
|
|
27
|
+
at most one support shape from another row.
|
|
28
|
+
|
|
29
|
+
| Situation | Primary | Support | Finish |
|
|
30
|
+
| --- | --- | --- | --- |
|
|
31
|
+
| Ordinary or mixed work | Default Live Snapshot | Verification Gate Matrix | Plan-Style Result Summary |
|
|
32
|
+
| Design or planning | Decision Tournament | Context Coverage Matrix | User Decision Gate |
|
|
33
|
+
| Implementation | Implementation Work Ledger | Verification Gate Matrix | Completion Impact Summary |
|
|
34
|
+
| Review or audit | Agent Lens Matrix | Context Coverage Matrix | Evidence To Finding Trace |
|
|
35
|
+
| Release or install | Artifact Inventory | Rollout Or Convergence Watch | Risk Or Audit Table |
|
|
36
|
+
| Retry, cancellation, or long-running work | Recovery Ledger | Resource Budget Snapshot | Rollout Or Convergence Watch |
|
|
37
|
+
|
|
38
|
+
Phase Plan Preview is baseline behavior before a phase starts, not an extra
|
|
39
|
+
choice. Long Async Timeline is only a temporary diagnostic appendix when the
|
|
40
|
+
user explicitly asks to debug detailed event order.
|
|
41
|
+
|
|
42
|
+
## Cumulative Ledger Rule
|
|
43
|
+
|
|
44
|
+
Within one user request, progress snapshots are cumulative. Do not let completed
|
|
45
|
+
work scroll out of the next snapshot just because new work starts. Keep completed
|
|
46
|
+
rows, update their status, and append newly discovered work below them. This
|
|
47
|
+
makes the current answer self-contained even if the user only sees the latest
|
|
48
|
+
snapshot.
|
|
49
|
+
|
|
50
|
+
```text
|
|
51
|
+
Phase Commit Prep
|
|
52
|
+
|
|
53
|
+
+ README install flow done local + global skill install
|
|
54
|
+
+ Registry/install check done <version> published and installed
|
|
55
|
+
+ Verification done npm run test:all
|
|
56
|
+
|
|
57
|
+
> Commit running staging release changes
|
|
58
|
+
- Push queued origin/main
|
|
59
|
+
|
|
60
|
+
Checks 3 passed | 0 failed | 2 running/queued
|
|
61
|
+
Next: commit release changes
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
Use `+` for completed work, `>` for running work, `-` for queued work, and `!`
|
|
65
|
+
for blocked or failed work. Keep earlier completed rows visible in every later
|
|
66
|
+
snapshot for the same request.
|
|
67
|
+
|
|
7
68
|
## Default Live Snapshot
|
|
8
69
|
|
|
9
70
|
Use this for ordinary phase progress. It is inspired by test-runner summaries:
|
|
@@ -24,8 +85,7 @@ Checks 5 passed | 0 failed | 2 running
|
|
|
24
85
|
Elapsed 1m 12s
|
|
25
86
|
```
|
|
26
87
|
|
|
27
|
-
|
|
28
|
-
items, and `-` for queued items. Keep each row to one line when possible.
|
|
88
|
+
Keep each row to one line when possible.
|
|
29
89
|
|
|
30
90
|
## Dense Meter Snapshot
|
|
31
91
|
|
|
@@ -123,3 +183,236 @@ Details
|
|
|
123
183
|
```
|
|
124
184
|
|
|
125
185
|
This is best for final handoff or after multiple agents have returned.
|
|
186
|
+
|
|
187
|
+
## Building Block Examples
|
|
188
|
+
|
|
189
|
+
Use these shapes as building blocks selected by the Situation Choice Matrix
|
|
190
|
+
above. Do not present this full list as user-facing choices.
|
|
191
|
+
|
|
192
|
+
### Phase Plan Preview
|
|
193
|
+
|
|
194
|
+
Use this immediately before spawning agents for a phase.
|
|
195
|
+
|
|
196
|
+
```text
|
|
197
|
+
Phase Inspect - planned fan-out
|
|
198
|
+
|
|
199
|
+
- Runtime contracts check execution authority and failure semantics
|
|
200
|
+
- UX/progress inspect visible status and completion summaries
|
|
201
|
+
- Tests/package verify E2E, tarball, and installed skill contents
|
|
202
|
+
|
|
203
|
+
Parallelism: 3 agents
|
|
204
|
+
Synthesis: merge material findings and decide next phase
|
|
205
|
+
```
|
|
206
|
+
|
|
207
|
+
### Agent Lens Matrix
|
|
208
|
+
|
|
209
|
+
Use this when several agents are reviewing the same artifact from different
|
|
210
|
+
angles.
|
|
211
|
+
|
|
212
|
+
```text
|
|
213
|
+
Review Lenses
|
|
214
|
+
|
|
215
|
+
+ Runtime correctness no material issue workflow-runtime.ts
|
|
216
|
+
+ Capability boundary no material issue subagent-backend.ts
|
|
217
|
+
! User visibility needs fix progress visuals
|
|
218
|
+
> Package contract running tarball contents
|
|
219
|
+
|
|
220
|
+
Findings 1 open | 2 clear | 1 running
|
|
221
|
+
Next: fix user visibility contract
|
|
222
|
+
```
|
|
223
|
+
|
|
224
|
+
### Implementation Work Ledger
|
|
225
|
+
|
|
226
|
+
Use this when work is split into edit scopes.
|
|
227
|
+
|
|
228
|
+
```text
|
|
229
|
+
Implementation Ledger
|
|
230
|
+
|
|
231
|
+
+ Skill routing docs updated skills/ultracode-for-codex
|
|
232
|
+
+ CLI skill command added skills/ultracode-for-codex-cli
|
|
233
|
+
> E2E package checks editing scripts/e2e-installed-...
|
|
234
|
+
- README install flow queued local/global install examples
|
|
235
|
+
|
|
236
|
+
Files touched 6 | Tests queued 2 | Risk low
|
|
237
|
+
```
|
|
238
|
+
|
|
239
|
+
### Verification Gate Matrix
|
|
240
|
+
|
|
241
|
+
Use this after tests, package checks, reviews, or publish dry-runs.
|
|
242
|
+
|
|
243
|
+
```text
|
|
244
|
+
Verification Gates
|
|
245
|
+
|
|
246
|
+
+ unit/integration passed 35 tests
|
|
247
|
+
+ installed E2E passed fake Codex boundary
|
|
248
|
+
+ package validation passed 35 files in tarball
|
|
249
|
+
+ publish dry-run passed would publish <version>
|
|
250
|
+
|
|
251
|
+
Gates 4 passed | 0 failed | 0 skipped
|
|
252
|
+
Residual: live Codex smoke remains opt-in
|
|
253
|
+
```
|
|
254
|
+
|
|
255
|
+
### Decision Tournament
|
|
256
|
+
|
|
257
|
+
Use this when comparing approaches before implementation.
|
|
258
|
+
|
|
259
|
+
```text
|
|
260
|
+
Decision Tournament
|
|
261
|
+
|
|
262
|
+
A CLI orchestration rejected low visibility
|
|
263
|
+
B Native main orchestration selected highest chat visibility
|
|
264
|
+
C Hybrid auto-router deferred more moving parts
|
|
265
|
+
|
|
266
|
+
Winner: B
|
|
267
|
+
Reason: best fit for Codex-native progress and direct subagent control
|
|
268
|
+
```
|
|
269
|
+
|
|
270
|
+
### Blocked Or Risk Snapshot
|
|
271
|
+
|
|
272
|
+
Use this when a dependency, permission, failing gate, or missing capability
|
|
273
|
+
blocks progress.
|
|
274
|
+
|
|
275
|
+
```text
|
|
276
|
+
Blocked Snapshot
|
|
277
|
+
|
|
278
|
+
+ Package contents verified
|
|
279
|
+
! Native subagent surface blocked multi-agent tools unavailable
|
|
280
|
+
- Parallel review paused needs subagent surface
|
|
281
|
+
|
|
282
|
+
Blocker: native parallel orchestration unavailable in this session
|
|
283
|
+
Fallback: continue single-context review and record residual risk
|
|
284
|
+
```
|
|
285
|
+
|
|
286
|
+
### Retry Or Recovery Ledger
|
|
287
|
+
|
|
288
|
+
Use this for transient failures, retry loops, cancellation, or recovery.
|
|
289
|
+
|
|
290
|
+
```text
|
|
291
|
+
Recovery Ledger
|
|
292
|
+
|
|
293
|
+
+ attempt 1 failed workflow_agent_stalled
|
|
294
|
+
+ retry policy applied retry 1 / 2
|
|
295
|
+
> attempt 2 running narrowed review prompt
|
|
296
|
+
- synthesis queued after terminal result
|
|
297
|
+
|
|
298
|
+
Retries 1 used | 1 remaining
|
|
299
|
+
Next: wait for attempt 2 terminal state
|
|
300
|
+
```
|
|
301
|
+
|
|
302
|
+
### Artifact Inventory
|
|
303
|
+
|
|
304
|
+
Use this when the output is files, package artifacts, generated docs, or local
|
|
305
|
+
state.
|
|
306
|
+
|
|
307
|
+
```text
|
|
308
|
+
Artifact Inventory
|
|
309
|
+
|
|
310
|
+
+ npm tarball artifacts/ultracode-for-codex-<version>.tgz
|
|
311
|
+
+ native skill skills/ultracode-for-codex/SKILL.md
|
|
312
|
+
+ CLI skill skills/ultracode-for-codex-cli/SKILL.md
|
|
313
|
+
+ progress examples skills/ultracode-for-codex/references/progress-visuals.md
|
|
314
|
+
|
|
315
|
+
Artifacts 4 ready | Sensitive local state not included
|
|
316
|
+
```
|
|
317
|
+
|
|
318
|
+
### Rollout Or Convergence Watch
|
|
319
|
+
|
|
320
|
+
Use this when waiting for a target state: publish propagation, installed package
|
|
321
|
+
availability, test shards, deployment checks, or background jobs.
|
|
322
|
+
|
|
323
|
+
```text
|
|
324
|
+
Convergence Watch
|
|
325
|
+
|
|
326
|
+
+ npm registry visible <version> latest
|
|
327
|
+
+ global CLI updated /opt/homebrew/bin/ultracode-for-codex
|
|
328
|
+
> Codex skill reload pending next session boundary
|
|
329
|
+
- downstream smoke queued user project install
|
|
330
|
+
|
|
331
|
+
Converged 2 / 4
|
|
332
|
+
Next: verify downstream smoke after reload
|
|
333
|
+
```
|
|
334
|
+
|
|
335
|
+
### Risk Or Audit Table
|
|
336
|
+
|
|
337
|
+
Use this for security, capability boundaries, provenance, dependency audit,
|
|
338
|
+
license review, or data exposure checks.
|
|
339
|
+
|
|
340
|
+
```text
|
|
341
|
+
Risk Audit
|
|
342
|
+
|
|
343
|
+
severity area status evidence
|
|
344
|
+
high provider credentials clear env stripping test
|
|
345
|
+
medium local state paths clear .ultracode-for-codex ignored
|
|
346
|
+
low docs ambiguity open install wording
|
|
347
|
+
|
|
348
|
+
Open risk 1 low | Material risk 0
|
|
349
|
+
Next: clarify install wording
|
|
350
|
+
```
|
|
351
|
+
|
|
352
|
+
### Context Coverage Matrix
|
|
353
|
+
|
|
354
|
+
Use this when the quality of a result depends on which evidence was actually
|
|
355
|
+
read, searched, tested, or left unverified.
|
|
356
|
+
|
|
357
|
+
```text
|
|
358
|
+
Context Coverage
|
|
359
|
+
|
|
360
|
+
+ runtime source read src/runtime/workflow-runtime.ts
|
|
361
|
+
+ package scripts read scripts/package-...
|
|
362
|
+
+ installed E2E executed npm run test:e2e:...
|
|
363
|
+
! live Codex smoke skipped opt-in local environment
|
|
364
|
+
|
|
365
|
+
Coverage 3 verified | 1 residual
|
|
366
|
+
Residual: live smoke remains user-triggered
|
|
367
|
+
```
|
|
368
|
+
|
|
369
|
+
### User Decision Gate
|
|
370
|
+
|
|
371
|
+
Use this when the next step needs a product or risk choice rather than more
|
|
372
|
+
execution.
|
|
373
|
+
|
|
374
|
+
```text
|
|
375
|
+
Decision Gate
|
|
376
|
+
|
|
377
|
+
A Publish now ready all gates green
|
|
378
|
+
B Add live smoke first safer needs local Codex run
|
|
379
|
+
C Defer release safest no user impact yet
|
|
380
|
+
|
|
381
|
+
Recommended: A
|
|
382
|
+
Why: package and dry-run gates are green; live smoke is optional
|
|
383
|
+
```
|
|
384
|
+
|
|
385
|
+
### Resource Budget Snapshot
|
|
386
|
+
|
|
387
|
+
Use this for long work where elapsed time, agent count, retry budget, or token
|
|
388
|
+
budget matters.
|
|
389
|
+
|
|
390
|
+
```text
|
|
391
|
+
Resource Budget
|
|
392
|
+
|
|
393
|
+
agents active 3 / 6
|
|
394
|
+
retries used 1 / 2
|
|
395
|
+
elapsed 7m 20s
|
|
396
|
+
timeout none
|
|
397
|
+
token budget not capped
|
|
398
|
+
|
|
399
|
+
Pressure: low
|
|
400
|
+
Next: wait for active agents before synthesis
|
|
401
|
+
```
|
|
402
|
+
|
|
403
|
+
### Evidence To Finding Trace
|
|
404
|
+
|
|
405
|
+
Use this when translating many observations into a smaller set of findings or
|
|
406
|
+
fixes.
|
|
407
|
+
|
|
408
|
+
```text
|
|
409
|
+
Evidence Trace
|
|
410
|
+
|
|
411
|
+
evidence finding action
|
|
412
|
+
package files include skill refs packaging contract ok keep
|
|
413
|
+
npm exec run path untested E2E gap add test
|
|
414
|
+
progress rows disappear visibility gap add ledger rule
|
|
415
|
+
|
|
416
|
+
Findings 2 actionable | 1 keep
|
|
417
|
+
Next: implement E2E gap and ledger rule
|
|
418
|
+
```
|