loadout-ai 0.9.2 → 0.9.4

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.
Files changed (14) hide show
  1. package/CHANGELOG.md +25 -0
  2. package/README.md +160 -242
  3. package/catalog/discovered.json +28808 -27462
  4. package/dist/src/core/coordination/lock.js +25 -3
  5. package/dist/src/core/reporting/cli-guide.js +4 -2
  6. package/docs/DISCOVERED.md +248 -248
  7. package/docs/REFERENCE.md +6 -2
  8. package/docs/USER_TEST_GUIDE.md +4 -3
  9. package/docs/assets/loadout-social-preview.png +0 -0
  10. package/docs/assets/loadout-unified-workflow-v2.webp +0 -0
  11. package/docs/superpowers/plans/2026-09-07-public-surface-demo-release.md +181 -0
  12. package/docs/superpowers/plans/2026-09-07-readme-and-unified-hero.md +380 -0
  13. package/docs/superpowers/specs/2026-09-07-readme-and-unified-hero-design.md +135 -0
  14. package/package.json +1 -1
package/CHANGELOG.md CHANGED
@@ -2,6 +2,31 @@
2
2
 
3
3
  ## Unreleased
4
4
 
5
+ ## 0.9.4 - 2026-09-07
6
+
7
+ ### Changed
8
+
9
+ - Updated the Stable first-run flow to show detailed safety findings before
10
+ applying and require explicit acknowledgement of the currently reported
11
+ instruction-like files.
12
+
13
+ ### Fixed
14
+
15
+ - Serialized coordination lock recovery within a process so competing
16
+ recoverers cannot enter the same critical section simultaneously.
17
+
18
+ ## 0.9.3 - 2026-09-07
19
+
20
+ ### Changed
21
+
22
+ - Unified README around the full Loadout lifecycle: discover, curate, activate,
23
+ handoff, and coordinate — with a single hero image covering all five stages.
24
+ - Added prominent documentation for one-command bounded Claude Code ↔ Codex
25
+ design discussions (`loadout coord discuss start`).
26
+ - Aligned the GitHub social-preview asset with the unified workflow narrative.
27
+ - Moved npm release authentication fully to trusted publishing (OIDC), removing
28
+ the legacy publish-token fallback.
29
+
5
30
  ## 0.9.2 - 2026-09-06
6
31
 
7
32
  ### Changed
package/README.md CHANGED
@@ -1,16 +1,9 @@
1
- <p align="center">
2
- <img src="./docs/assets/loadout-discover-activate.webp" alt="Loadout discovers agent skills, tools, and MCP servers; screens and pins them; activates the right set for a repository; and lets users preview or roll back every change." width="960">
3
- </p>
4
-
5
1
  <h1 align="center">Loadout</h1>
6
2
 
7
- <p align="center"><strong>Agent extensions, under control.</strong></p>
3
+ <p align="center"><strong>Manage skills for 12 coding agents. Hand off and coordinate work between Claude Code and Codex.</strong></p>
8
4
 
9
5
  <p align="center">
10
- <strong>The package manager for your AI coding setup.</strong><br>
11
- Find skills worth having, activate the right ones per project,<br>
12
- install them across 12 agents, hand work from Claude Code to Codex,<br>
13
- and undo any of it.
6
+ <img src="./docs/assets/loadout-unified-workflow-v2.webp" alt="Loadout discovers, curates, and activates skills, tools, and MCP servers for coding agents, then helps Claude Code and Codex handoff and coordinate work through durable tasks, bundled context, ownership, contracts, and decisions." width="960">
14
7
  </p>
15
8
 
16
9
  <p align="center">
@@ -23,186 +16,160 @@
23
16
  </p>
24
17
 
25
18
  <p align="center">
26
- <a href="#what-it-looks-like">See it</a> ·
27
- <a href="#install">Install</a> ·
28
- <a href="#use-it-from-inside-your-agent">Use it in your agent</a> ·
29
- <a href="#passing-work-between-two-agents">Two agents</a> ·
30
- <a href="#why-loadout">Why</a> ·
19
+ <a href="#try-it-in-30-seconds">Quick start</a> ·
20
+ <a href="#use-claude-code-and-codex-together">Two agents</a> ·
21
+ <a href="#try-these-prompts">Agent skills</a> ·
22
+ <a href="#safety-and-trust">Safety</a> ·
31
23
  <a href="#command-reference">Commands</a>
32
24
  </p>
33
25
 
34
- ## What it looks like
35
-
36
- ```console
37
- $ loadout doctor
38
- loadout doctor — HEALTHY
39
-
40
- Platform: darwin
41
- State: ~/.loadout ✓ writable
42
- Agents: 3 detected, 9 available
43
-
44
- DETECTED AGENTS
45
- ✓ Claude Code
46
- ~/.claude/skills
47
- 43 skills | supports: skill, command, agent, mcp, plugin, root
48
- ✓ Codex
49
- ~/.agents/skills
50
- 30 skills | supports: skill, command, agent, mcp, plugin, root
51
-
52
- $ loadout recommend --project .
53
- Project: checkout-service
54
- Detected: TypeScript, Zod, Vitest, next.js, react
55
-
56
- Rule-based project suggestions:
57
- superpowers [high, skill library] — Useful engineering planning, testing, and review workflows.
58
- context7 [high, skill library] — Current library documentation helps agents avoid outdated APIs.
59
- ui-ux-pro-max [high, skill library] — Frontend framework detected: next.js, react.
60
- playwright-mcp [medium, MCP/runtime setup] — Browser verification may help test the detected frontend.
26
+ ## Try it in 30 seconds
27
+
28
+ You need Node.js 20 or newer and Git.
29
+
30
+ ```bash
31
+ npm install --global loadout-ai
32
+ loadout setup --mode stable --details
33
+ ```
34
+
35
+ The second command detects your coding agents and previews the 30-skill Stable
36
+ loadout. It does not change agent files. Review the plan, then apply it:
37
+
38
+ ```bash
39
+ loadout setup --mode stable --yes --approve-risk
40
+ loadout status
61
41
  ```
62
42
 
63
- It reads your repository rather than a config file, knows which agents you
64
- actually have, and previews and snapshots every write before it touches them.
43
+ Loadout saves a rollback snapshot before applying changes. Run
44
+ `loadout rollback` to restore the previous managed state.
65
45
 
66
- ### Demo
46
+ `--approve-risk` acknowledges the instruction-like files reported by the current
47
+ Stable preview. Use it only after reading those findings; it does not replace the
48
+ preview or the pinned-source checks.
67
49
 
68
- <p align="center">
69
- <a href="https://www.youtube.com/watch?v=opNqJKX7xMw">
70
- <img src="https://img.youtube.com/vi/opNqJKX7xMw/maxresdefault.jpg" alt="Watch the 72-second Loadout demo" width="880">
71
- </a>
72
- </p>
50
+ A later `--yes` invocation recomputes the plan from pinned sources and current agent and filesystem state; it does not persist or prove identity with the earlier preview.
73
51
 
74
- **[Watch the 72-second Loadout demo on YouTube](https://www.youtube.com/watch?v=opNqJKX7xMw).**
52
+ ## What Loadout does
75
53
 
76
- ## Install
54
+ | Capability | What you get |
55
+ | -------------- | --------------------------------------------------------------------- |
56
+ | **Discover** | Find skills, tools, and MCP servers worth reviewing |
57
+ | **Curate** | Inspect, screen, and pin sources before trusting them |
58
+ | **Activate** | Install a focused set for this repository across your agents |
59
+ | **Handoff** | Pass durable tasks with bundled context between Claude Code and Codex |
60
+ | **Coordinate** | Share file ownership, contracts, decisions, and acknowledgements |
77
61
 
78
- You need Node.js 20 or newer and Git.
62
+ The package manager works across supported agents. Handoff and coordination are
63
+ currently designed for Claude Code and Codex working in the same repository.
64
+
65
+ ## Use Claude Code and Codex together
66
+
67
+ ### Handoff: pass a task that survives sessions
79
68
 
80
69
  ```bash
81
- npm install --global loadout-ai
82
- loadout setup --mode stable
70
+ loadout handoff codex "write unit tests for auth" --bundle src/auth.ts src/types.ts --verify "tests pass" --verify-command npm --verify-args '["test"]'
71
+ ```
72
+
73
+ Codex sees the task when it checks its inbox, works from the attached bounded
74
+ context, and marks it done. Claude Code can then read the result. Handoffs live
75
+ in an append-only project log, so restarting either agent does not erase them.
76
+ Bundles contain secret-redacted text and are capped at 50 KiB total; still
77
+ review them before committing and never attach credential files.
78
+
79
+ ### Coordinate: work at the same time without stepping on files
80
+
81
+ ```bash
82
+ loadout coord start --agents claude-code,codex # preview ownership
83
+ loadout coord start --agents claude-code,codex --yes # apply ownership
84
+ loadout coord snapshot codex # inspect shared state
85
+ loadout coord discuss start "REST or GraphQL?" --agents claude-code,codex --rounds 2 --max-turns 5
83
86
  ```
84
87
 
85
- The second command detects your agents and previews the 30-skill Stable loadout.
86
- Nothing changes until you approve it. If anything goes wrong, start with the
87
- [user test guide](./docs/USER_TEST_GUIDE.md).
88
+ Coordination is structured shared project state, not shared memory or a merged
89
+ context window. Events reach an agent at safe turn boundaries or when it checks
90
+ its snapshot; Loadout does not interrupt a turn in progress.
88
91
 
89
- For a reproducible install, pin the release: `npm install --global loadout-ai@0.9.2`.
92
+ ### Let Claude and Codex debate one decision
90
93
 
91
- ## Use it from inside your agent
94
+ You prompt once and Loadout calls both providers in bounded turns—propose,
95
+ critique, synthesize—then records the decision. Neither agent can inject into
96
+ the other's in-progress turn; events arrive at safe turn boundaries. Each
97
+ round consumes paid provider turns from your configured Claude and Codex quota.
98
+ The bounded design discussion reports its turn budget before it starts.
92
99
 
93
- A CLI you have to leave your agent to run is a context switch. Loadout ships two
94
- skills so you do not have to: one for agent handoffs and one for choosing a
95
- focused skill set for the current repository.
100
+ See the [live coordination guide](./docs/LIVE_COLLABORATION.md) for ownership,
101
+ contracts, acknowledgements, discussions, the local dashboard, and limitations.
102
+
103
+ ## Try these prompts
104
+
105
+ Install Loadout's two first-party skills once so your agent knows the workflow:
96
106
 
97
107
  ```bash
98
108
  loadout skills install loadout-handoff --yes
99
109
  loadout skills install loadout-curator --yes
100
110
  ```
101
111
 
102
- Start a new agent session and just ask, in the conversation you are already in:
112
+ Then ask naturally:
103
113
 
104
114
  > _"Hand the test writing to Codex."_
115
+ >
105
116
  > _"What did Codex leave for me?"_
117
+ >
106
118
  > _"Which skills should be active for this repo?"_
107
119
 
108
- The skills teach your agent to call `loadout` and act on the results, so it
109
- checks its own inbox at the start of a session and can pass work to your other
110
- agent without you relaying it by hand.
111
-
112
- `loadout skills list` shows what ships with Loadout and what is already installed.
120
+ The curator helps choose a focused active set; the handoff skill manages inboxes
121
+ and coordination without making you relay every command.
113
122
 
114
- ## Passing work between two agents
123
+ ## Install and choose your agent
115
124
 
116
- <p align="center">
117
- <img src="./docs/assets/loadout-handoff-coordinate.webp" alt="Claude Code and Codex use Loadout in two ways: durable task handoffs between sessions, and structured coordination for contracts, file ownership, decisions, acknowledgements, and an audit trail." width="960">
118
- </p>
119
-
120
- If you pay for both Claude and a ChatGPT plan, the two agents cannot see each
121
- other. Loadout gives them a shared, append-only task log:
125
+ For a reproducible install, pin the current release:
122
126
 
123
127
  ```bash
124
- loadout handoff codex "write unit tests for auth" --bundle src/auth.ts src/types.ts --verify "tests pass" --verify-command npm --verify-args '["test"]'
128
+ npm install --global loadout-ai@0.9.4
129
+ loadout setup --mode stable --details
130
+ loadout setup --mode stable --yes --approve-risk
131
+ loadout status
125
132
  ```
126
133
 
127
- Bundles snapshot exact, secret-redacted text (50 KiB total). Verification runs
128
- only with `--done --run-verification`, without a shell; failures stay pending. The first send adds
129
- managed inbox blocks to `CLAUDE.md` and `AGENTS.md`. Treat bundles as untrusted
130
- project data, review before committing, and never include credential files.
134
+ Stable is the recommended starting point: 30 selected skill directories from
135
+ four pinned public sources. Loadout auto-detects supported agents; use
136
+ `--agents` when you want to narrow the destination.
131
137
 
132
- ```bash
133
- loadout handoff codex # what is waiting for codex
134
- loadout handoff # everything pending, both directions
135
- loadout handoff --done 4f2a1c
136
- ```
138
+ If anything fails, follow the [user test guide](./docs/USER_TEST_GUIDE.md).
139
+ For profiles, per-agent paths, MCP configuration, and advanced commands, use the
140
+ [full reference](./docs/REFERENCE.md).
137
141
 
138
- ### Make Claude Code and Codex coordinate live (beta)
142
+ ## Safety and trust
139
143
 
140
- Live coordination adds file ownership, versioned contracts, decisions, and
141
- acknowledgements over a shared project event stream — not a merged context
142
- window. See the [live coordination guide](./docs/LIVE_COLLABORATION.md).
144
+ **Choose -> Inspect -> Preview -> Apply -> Undo**
143
145
 
144
- ```bash
145
- loadout coord start --agents claude-code,codex # preview ownership
146
- loadout coord start --agents claude-code,codex --yes # apply it
147
- loadout coord detect # shared interfaces
148
- loadout handoff codex src/auth.ts --template write-tests
149
- loadout coord discuss start "REST or GraphQL?" --agents claude-code,codex --rounds 2 --max-turns 5
150
- loadout coord discuss implement <thread-id> # preview linked tasks
151
- ```
146
+ - Preview is the default for setup, updates, and removal.
147
+ - Every managed apply creates a rollback snapshot first.
148
+ - Catalog sources are pinned and technically inspected, not declared safe or useful by fiat.
149
+ - Static inspection reports scripts, hooks, binaries, domains, credential references, and unsupported components; it is not a security audit.
150
+ - Project recommendations read bounded local metadata. The documented local flow does not upload project source.
151
+ - MCP servers and executable tools stay behind separate preview, permission, and setup steps.
152
+ - Shared manifests store environment-variable or OS-keychain references, not secret values.
152
153
 
153
- Contract publication and bounded design discussion implementation require a second `--yes`
154
- approval. Exact supported declarations are marked current or stale; ambiguous
155
- ones require manual review. The optional MCP server and provider bridge deliver
156
- events at safe turn boundaries, never during an active turn. The
157
- [coordination guide](./docs/LIVE_COLLABORATION.md) covers templates, Git-author
158
- mappings, the local dashboard, limitations, and the complete workflow.
154
+ <!-- loadout:current-limits:start -->
159
155
 
160
- ## How it works
156
+ - All catalog records have identified SPDX licenses. See the [recorded license decisions](./docs/UPSTREAM_LICENSE_DECISIONS.md) for the source-by-source record.
161
157
 
162
- **Choose -> Inspect -> Preview -> Apply -> Undo**
158
+ <!-- loadout:current-limits:end -->
163
159
 
164
- 1. **Choose** Stable, Power, Maximum, or your own package list.
165
- 2. **Inspect** where each extension comes from and what it can do.
166
- 3. **Preview** every planned change without changing agent files.
167
- 4. **Apply** with `--yes`; Loadout saves a rollback snapshot first.
168
- 5. **Undo** with `loadout rollback` if you change your mind.
169
-
170
- ### Abridged terminal transcript
171
-
172
- This is an explicitly abridged transcript from a disposable Stable run. A literal `…` marks omitted fetch output; `<snapshot-id>` is a variable placeholder because snapshot IDs vary. Loadout auto-detects installed agents; `--agents` narrows the selection when needed.
173
-
174
- ```console
175
- $ loadout setup --mode stable
176
- …
177
- Loadout: Stable
178
- Detected agents: Claude Code, Cursor, Codex
179
- Catalog selection: 4 repositories
180
- Ready to install: 4 skill repositories (30 agent skill directories)
181
- Preview complete; nothing was changed. Re-run with --yes to install this exact screened plan.
182
-
183
- $ loadout setup --mode stable --yes
184
- …
185
- Loadout installed 4 repositories for 3 agent(s). Snapshot: <snapshot-id>
186
-
187
- $ loadout rollback
188
- Restored snapshot <snapshot-id>
189
-
190
- $ loadout handoff codex "write tests for the auth module" --context "zod schemas exist"
191
- created .handoff/
192
- told codex to check its inbox
193
- Sent to codex: write tests for the auth module
194
-
195
- $ loadout doctor
196
- loadout doctor — HEALTHY
197
- Platform: darwin
198
- State: ~/.loadout ✓ writable
199
- Agents: 3 detected, 9 available
200
- ```
160
+ Read the [security policy](./SECURITY.md), [catalog policy](./docs/CATALOG_POLICY.md),
161
+ and [credential and update policy](./docs/CREDENTIAL_AND_UPDATE_POLICY.md) before
162
+ trusting third-party content.
163
+
164
+ ## Demo
201
165
 
202
- The final preview sentence above is captured CLI wording. A later `--yes` invocation recomputes the plan from pinned sources and current agent and filesystem state; it does not persist or prove identity with the earlier preview.
166
+ <p align="center">
167
+ <a href="https://www.youtube.com/watch?v=opNqJKX7xMw">
168
+ <img src="https://img.youtube.com/vi/opNqJKX7xMw/maxresdefault.jpg" alt="Watch the 72-second Loadout demo" width="880">
169
+ </a>
170
+ </p>
203
171
 
204
- Preview may fill Loadout's private download cache, but it does not change your agent
205
- files. Review the summary and warnings before approving an apply command.
172
+ **[Watch the 72-second Loadout demo on YouTube](https://www.youtube.com/watch?v=opNqJKX7xMw).**
206
173
 
207
174
  ## Why Loadout
208
175
 
@@ -225,7 +192,7 @@ structured project facts when Claude Code and Codex work together.
225
192
 
226
193
  Everything on this page is enforced. `docs/evidence/readme-claims.json` records
227
194
  each material claim with the code or command that proves it, and CI fails the
228
- build when the README and the implementation disagree — including the pinned
195
+ build when the README and the implementation disagree—including the pinned
229
196
  version in the install line above.
230
197
 
231
198
  Loadout watches a much wider catalog than it activates. You can keep thousands of
@@ -243,53 +210,30 @@ everything into every prompt.
243
210
  | Manually remember what was changed | Scan, reconcile, remove, roll back, or completely uninstall |
244
211
 
245
212
  Loadout is local, open source, and preview-first. It does not need an LLM API key
246
- to manage skills. MCP servers and executable tools stay behind their own explicit
247
- setup and permission steps.
248
-
249
- ## Stable workflow
250
-
251
- ### Stable: install the essentials and start building
252
-
253
- Stable is the recommended daily driver: **30 selected skill directories from four
254
- pinned public sources**, installed into each agent you choose.
255
-
256
- ```bash
257
- loadout setup --mode stable
258
- loadout setup --mode stable --yes
259
- loadout status
260
- loadout scan
261
- loadout rollback
262
- ```
263
-
264
- Stable is Loadout's strongest general starting point, not a claim that one setup is
265
- best for every person or project.
213
+ to manage skills.
266
214
 
267
215
  ## Profiles
268
216
 
269
- | Mode | Sources | Skills | Active by default |
270
- | --------- | --------------------- | ------ | -------------------------------- |
271
- | `stable` | 4 | 30 | yes — recommended starting point |
272
- | `power` | 8 | 56 | yes |
273
- | `maximum` | all reviewed | all | **no — downloaded but disabled** |
274
- | `custom` | your `--package` list | varies | yes |
275
-
276
- **Maximum** downloads the entire reviewed library and leaves every skill _disabled_.
277
- Nothing reaches an agent prompt until a project activates what it needs.
217
+ | Mode | Sources | Skills | Active by default |
218
+ | --------- | --------------------- | ------ | ------------------------------ |
219
+ | `stable` | 4 | 30 | yes—recommended starting point |
220
+ | `power` | 8 | 56 | yes |
221
+ | `maximum` | all reviewed | all | **no—downloaded but disabled** |
222
+ | `custom` | your `--package` list | varies | yes |
278
223
 
279
- For detailed profile tables, source lists, and custom configuration, see the
280
- **[full reference](./docs/REFERENCE.md)**.
224
+ Maximum downloads the reviewed library and leaves every skill disabled. Nothing
225
+ reaches an agent prompt until a project activates what it needs. See the
226
+ [full reference](./docs/REFERENCE.md) for source lists and custom configuration.
281
227
 
282
228
  ## Catalog and discovery
283
229
 
284
- The catalog is not a frozen list. Loadout separates **discovery** from
285
- **installation** so a viral repo can be noticed quickly without being trusted
286
- blindly.
230
+ Discovery is separate from trust and installation. A popular new repository can
231
+ enter the review queue without being installed or promoted automatically.
287
232
 
288
233
  ```bash
289
234
  loadout discover --source all --queue
290
235
  loadout review-queue
291
236
  loadout candidate inspect owner/repository
292
- loadout update
293
237
  ```
294
238
 
295
239
  <!-- loadout:catalog-coverage:start -->
@@ -306,26 +250,9 @@ Catalog maturity: **53 sourced**, **53 technically inspected**, and **4 selected
306
250
 
307
251
  <!-- loadout:daily-discovery:start -->
308
252
 
309
- **Discovery snapshot (generated 2026-09-06):** [237 repositories observed](./docs/DISCOVERED.md), including 221 uncataloged review candidates and 16 repositories already in the inspected catalog.
253
+ **Discovery snapshot (generated 2026-09-07):** [237 repositories observed](./docs/DISCOVERED.md), including 221 uncataloged review candidates and 16 repositories already in the inspected catalog.
310
254
  <!-- loadout:daily-discovery:end -->
311
255
 
312
- ## Trust and limits
313
-
314
- - A pinned commit identifies source bytes; it does not prove safety, correct licensing, usefulness, or future compatibility.
315
- - Static inspection reports scripts, hooks, binaries, domains, credential references, and unsupported components. It is not a security audit.
316
- - No bundled source is called proven until human review and recorded local outcomes support it.
317
- - Project recommendations read bounded local metadata. The documented local flow does not upload project source.
318
- - MCP servers and executable tools have separate preview and approval paths.
319
- - Shared manifests hold environment-variable or OS-keychain references, not secret values.
320
-
321
- <!-- loadout:current-limits:start -->
322
-
323
- - All catalog records have identified SPDX licenses. See the [recorded license decisions](./docs/UPSTREAM_LICENSE_DECISIONS.md) for the source-by-source record.
324
-
325
- <!-- loadout:current-limits:end -->
326
-
327
- Read the [security policy](./SECURITY.md), [catalog policy](./docs/CATALOG_POLICY.md), and [credential and update policy](./docs/CREDENTIAL_AND_UPDATE_POLICY.md) before trusting third-party content.
328
-
329
256
  ## Agent support
330
257
 
331
258
  <!-- loadout:support-summary:start -->
@@ -344,42 +271,27 @@ Configured CI platforms describe a manually triggered workflow, not evidence tha
344
271
 
345
272
  ## Command reference
346
273
 
347
- | What it does | Command |
348
- | --------------------------------------- | ---------------------------------------------------------- |
349
- | Beginner-friendly guided path | `loadout guide` |
350
- | Preview the 30-skill Stable setup | `loadout setup --mode stable` |
351
- | Apply after reviewing the preview | `loadout setup --mode stable --yes` |
352
- | Show managed packages and active skills | `loadout status` · `loadout library` |
353
- | What fits this repository | `loadout recommend --project .` |
354
- | Project-specific active set | `loadout optimize --project . --limit 30` |
355
- | Scan existing skills across agents | `loadout scan` |
356
- | Check for source updates | `loadout update` |
357
- | Find newly launched candidates | `loadout discover --source all --queue` |
358
- | Install Loadout's own skill | `loadout skills install loadout-handoff --yes` |
359
- | Send a task to another agent | `loadout handoff codex "write tests"` |
360
- | Use a reusable handoff template | `loadout handoff codex src/auth.ts --template write-tests` |
361
- | Preview two-agent ownership setup | `loadout coord start --agents claude-code,codex` |
362
- | Detect shared contract candidates | `loadout coord detect` |
363
- | Inspect shared agent state | `loadout coord snapshot codex` |
364
- | Detect live provider runtimes | `loadout coord agents detect` |
365
- | Debate one design with both providers | `loadout coord discuss start "<topic>" ...` |
366
- | Turn a decision into linked tasks | `loadout coord discuss implement <thread-id>` |
367
- | Start the coordination MCP server | `loadout serve` |
368
- | Agent health check | `loadout doctor` |
369
- | Rollback the latest managed change | `loadout rollback` |
370
- | Preview complete removal | `loadout uninstall` |
371
- | Full CLI reference | `loadout --help` · `loadout advanced` |
372
-
373
- Most mutating commands are dry runs first. Add `--yes` to apply.
374
-
375
- ## Built with Claude and Codex
376
-
377
- Loadout was designed and built by [Viraj Mishra](https://github.com/VirajMishra1) with Claude Code and Codex.
378
-
379
- Loadout's core skill management does **not** call an LLM API or require an LLM API
380
- key. The opt-in provider bridge and design room do invoke your configured
381
- Claude/Codex sessions and spend their quota; neither is a hidden requirement for
382
- discovering, installing, or rolling back extensions.
274
+ | Goal | Command |
275
+ | ---------------------------------- | ---------------------------------------------------------------------------------- |
276
+ | Guided first run | `loadout guide` |
277
+ | Preview or apply Stable | `loadout setup --mode stable --details` · then `--yes --approve-risk` after review |
278
+ | Inspect managed skills | `loadout status` · `loadout library` |
279
+ | Recommend for this repository | `loadout recommend --project .` |
280
+ | Activate a project-specific set | `loadout optimize --project . --limit 30` |
281
+ | Scan or update | `loadout scan` · `loadout update` |
282
+ | Send a task | `loadout handoff codex "write tests"` |
283
+ | Use a handoff template | `loadout handoff codex src/auth.ts --template write-tests` |
284
+ | Preview two-agent ownership | `loadout coord start --agents claude-code,codex` |
285
+ | Inspect coordination state | `loadout coord snapshot codex` |
286
+ | Detect shared contract candidates | `loadout coord detect` |
287
+ | Turn a decision into tasks | `loadout coord discuss implement <thread-id>` |
288
+ | Start the coordination MCP server | `loadout serve` |
289
+ | Health check | `loadout doctor` |
290
+ | Restore the previous managed state | `loadout rollback` |
291
+ | Preview complete removal | `loadout uninstall` |
292
+ | Full CLI reference | `loadout --help` · `loadout advanced` |
293
+
294
+ Most mutating commands are previews first. Add `--yes` only after reviewing the plan.
383
295
 
384
296
  ## Development
385
297
 
@@ -395,25 +307,31 @@ npm run verify:full
395
307
 
396
308
  <!-- loadout:verification-summary:end -->
397
309
 
398
- The [testing guide](./docs/TESTING.md) documents the exact checks and their boundaries.
310
+ The [testing guide](./docs/TESTING.md) documents each check and its boundary.
399
311
 
400
312
  ## Documentation
401
313
 
402
- - [Full reference (profiles, MCP, discovery, tools)](./docs/REFERENCE.md)
314
+ - [Full CLI and profile reference](./docs/REFERENCE.md)
315
+ - [User test guide](./docs/USER_TEST_GUIDE.md)
316
+ - [Live Claude Code ↔ Codex coordination](./docs/LIVE_COLLABORATION.md)
403
317
  - [Catalog and upstream credits](./docs/CATALOG.md)
404
318
  - [Catalog evidence policy](./docs/CATALOG_POLICY.md)
405
319
  - [Feature and evidence matrix](./docs/FEATURE_TEST_MATRIX.md)
406
320
  - [Testing contract](./docs/TESTING.md)
407
- - [User test guide](./docs/USER_TEST_GUIDE.md)
408
- - [Live Codex ↔ Claude collaboration design](./docs/LIVE_COLLABORATION.md)
409
321
  - [Changelog](./CHANGELOG.md)
410
322
 
411
323
  ## Contributing, security, and attribution
412
324
 
413
- See [CONTRIBUTING.md](./CONTRIBUTING.md) · [Code of Conduct](./CODE_OF_CONDUCT.md) · Report vulnerabilities through [SECURITY.md](./SECURITY.md), without credentials, private source, or unredacted state. General bugs and proposals belong in the [issue tracker](https://github.com/VirajMishra1/loadout/issues).
325
+ See [CONTRIBUTING.md](./CONTRIBUTING.md) · [Code of Conduct](./CODE_OF_CONDUCT.md) ·
326
+ Report vulnerabilities through [SECURITY.md](./SECURITY.md), without credentials,
327
+ private source, or unredacted state. General bugs and proposals belong in the
328
+ [issue tracker](https://github.com/VirajMishra1/loadout/issues).
414
329
 
415
- The catalog contains 53 credited public repositories. Inclusion records discovery and attribution; it does not transfer ownership, imply endorsement, or relicense upstream work.
330
+ The catalog contains 53 credited public repositories. Inclusion records discovery
331
+ and attribution; it does not transfer ownership, imply endorsement, or relicense
332
+ upstream work.
416
333
 
417
334
  ## License
418
335
 
419
- Loadout is licensed under the [MIT License](./LICENSE). Catalog entries retain their upstream licenses and terms.
336
+ Loadout is licensed under the [MIT License](./LICENSE). Catalog entries retain
337
+ their upstream licenses and terms.