ai-employees 1.9.0 → 1.11.0
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 +37 -11
- package/docs/FAQ.md +1 -1
- package/docs/RELEASING.md +3 -3
- package/employees/ad-manager-employee/README.md +1 -1
- package/employees/chief-of-staff/CAPABILITIES.md +25 -2
- package/employees/chief-of-staff/CHANGELOG.md +9 -0
- package/employees/chief-of-staff/CONTRACT.md +14 -2
- package/employees/chief-of-staff/INSTALL-PROMPT.md +1 -1
- package/employees/chief-of-staff/README.md +5 -2
- package/employees/chief-of-staff/VERSION +1 -1
- package/employees/chief-of-staff/employee.json +1 -1
- package/employees/chief-of-staff/routines/cos-charter-and-fleet-audit/SKILL.md +29 -3
- package/employees/chief-of-staff/scripts/voice-check.mjs +458 -0
- package/employees/customer-satisfaction-employee/AGENTS.md +1 -0
- package/employees/customer-satisfaction-employee/CAPABILITIES.md +32 -4
- package/employees/customer-satisfaction-employee/CHANGELOG.md +13 -0
- package/employees/customer-satisfaction-employee/CONTRACT.md +1188 -1157
- package/employees/customer-satisfaction-employee/INSTALL-PROMPT.md +5 -5
- package/employees/customer-satisfaction-employee/README.md +15 -9
- package/employees/customer-satisfaction-employee/ROLE.md +5 -3
- package/employees/customer-satisfaction-employee/SCHEDULE.md +297 -294
- package/employees/customer-satisfaction-employee/VERSION +1 -1
- package/employees/customer-satisfaction-employee/employee.json +273 -249
- package/employees/customer-satisfaction-employee/routines/csat-deflection-desk/SKILL.md +31 -3
- package/employees/customer-satisfaction-employee/routines/csat-desk-intake/SKILL.md +60 -6
- package/employees/customer-satisfaction-employee/routines/csat-desk-standup/SKILL.md +2 -2
- package/employees/customer-satisfaction-employee/routines/csat-reply-desk/SKILL.md +39 -2
- package/employees/customer-satisfaction-employee/routines/csat-review-requests/SKILL.md +418 -0
- package/employees/customer-satisfaction-employee/run/csat-review-requests.cmd.example +11 -0
- package/employees/customer-satisfaction-employee/scripts/runlog.mjs +779 -778
- package/employees/customer-satisfaction-employee/scripts/voice-check.mjs +458 -0
- package/employees/customer-satisfaction-employee/work-profile.json +71 -64
- package/employees/gtm-engineer/CAPABILITIES.md +26 -3
- package/employees/gtm-engineer/CHANGELOG.md +10 -0
- package/employees/gtm-engineer/CONTRACT.md +22 -6
- package/employees/gtm-engineer/INSTALL-PROMPT.md +2 -1
- package/employees/gtm-engineer/README.md +9 -5
- package/employees/gtm-engineer/VERSION +1 -1
- package/employees/gtm-engineer/employee.json +1 -1
- package/employees/gtm-engineer/routines/gtm-intake-and-dashboard/SKILL.md +58 -5
- package/employees/gtm-engineer/routines/gtm-launch-step-runner/SKILL.md +19 -1
- package/employees/gtm-engineer/routines/gtm-outreach-queue/SKILL.md +40 -2
- package/employees/gtm-engineer/scripts/voice-check.mjs +458 -0
- package/employees/sales-employee/CAPABILITIES.md +27 -4
- package/employees/sales-employee/CHANGELOG.md +24 -0
- package/employees/sales-employee/CONTRACT.md +36 -7
- package/employees/sales-employee/INSTALL-PROMPT.md +2 -1
- package/employees/sales-employee/README.md +38 -10
- package/employees/sales-employee/ROLE.md +3 -1
- package/employees/sales-employee/SCHEDULE.md +2 -2
- package/employees/sales-employee/VERSION +1 -1
- package/employees/sales-employee/employee.json +2 -1
- package/employees/sales-employee/examples/README.md +3 -0
- package/employees/sales-employee/examples/crm/bookings.jsonl +2 -0
- package/employees/sales-employee/examples/queue/2026-03-05-booked-calls.md +103 -0
- package/employees/sales-employee/examples/strategy/booked-calls.md +28 -0
- package/employees/sales-employee/playbooks/booked-calls/PLAYBOOK.md +276 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/branches.md +17 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/call-brief.md +31 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/call-now.md +30 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/config.md +52 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/email.md +22 -0
- package/employees/sales-employee/playbooks/booked-calls/templates/texts.md +36 -0
- package/employees/sales-employee/routines/sales-desk-setup/SKILL.md +65 -5
- package/employees/sales-employee/routines/sales-desk-standup/SKILL.md +23 -2
- package/employees/sales-employee/routines/sales-first-touch-drafts/SKILL.md +66 -7
- package/employees/sales-employee/routines/sales-followup-sweep/SKILL.md +54 -5
- package/employees/sales-employee/scripts/booked-call-check.mjs +223 -0
- package/employees/sales-employee/scripts/voice-check.mjs +458 -0
- package/employees/seo-employee/CAPABILITIES.md +25 -2
- package/employees/seo-employee/CHANGELOG.md +10 -0
- package/employees/seo-employee/CONTRACT.md +21 -5
- package/employees/seo-employee/INSTALL-PROMPT.md +2 -2
- package/employees/seo-employee/README.md +7 -3
- package/employees/seo-employee/VERSION +1 -1
- package/employees/seo-employee/employee.json +1 -1
- package/employees/seo-employee/routines/seo-draft-run/SKILL.md +38 -3
- package/employees/seo-employee/routines/seo-intake-and-map/SKILL.md +53 -1
- package/employees/seo-employee/scripts/voice-check.mjs +458 -0
- package/employees/social-media-employee/CAPABILITIES.md +25 -2
- package/employees/social-media-employee/CHANGELOG.md +12 -0
- package/employees/social-media-employee/CONTRACT.md +21 -6
- package/employees/social-media-employee/INSTALL-PROMPT.md +2 -2
- package/employees/social-media-employee/README.md +7 -4
- package/employees/social-media-employee/VERSION +1 -1
- package/employees/social-media-employee/employee.json +1 -1
- package/employees/social-media-employee/routines/soc-draft-queue/SKILL.md +41 -3
- package/employees/social-media-employee/routines/soc-engagement-sweep/SKILL.md +8 -0
- package/employees/social-media-employee/routines/soc-intake-and-voice/SKILL.md +54 -3
- package/employees/social-media-employee/routines/soc-performance-review/SKILL.md +4 -0
- package/employees/social-media-employee/scripts/voice-check.mjs +458 -0
- package/employees/web-dev-employee/README.md +1 -1
- package/package.json +2 -2
package/README.md
CHANGED
|
@@ -29,7 +29,10 @@
|
|
|
29
29
|
</p>
|
|
30
30
|
|
|
31
31
|
<p align="center">
|
|
32
|
-
<a href="https://club.reinventing.ai/register?next=%2Fmembers%2Fhire&utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=
|
|
32
|
+
<a href="https://club.reinventing.ai/register?next=%2Fmembers%2Fhire&utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=hero-hire"><img src="assets/cta-hire.png" width="900" alt="Hire your first AI Employee in one click: pick a role, copy one prompt, your agent installs it. Free account, no card."></a>
|
|
33
|
+
</p>
|
|
34
|
+
|
|
35
|
+
<p align="center">
|
|
33
36
|
<a href="https://club.reinventing.ai/register?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=btn-join"><img src="assets/btn-join.png" width="199" height="60" alt="Join the Agent Ops Club free"></a>
|
|
34
37
|
<a href="https://club.reinventing.ai/events?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=btn-sessions"><img src="assets/btn-sessions.png" width="163" height="60" alt="Agent Ops Club live sessions"></a>
|
|
35
38
|
<a href="https://club.reinventing.ai/?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=btn-club"><img src="assets/btn-club.png" width="186" height="60" alt="Visit the Agent Ops Club"></a>
|
|
@@ -43,29 +46,51 @@
|
|
|
43
46
|
⭐ <em>Found something useful? Star the repo. It takes a second and helps the next person find it.</em>
|
|
44
47
|
</p>
|
|
45
48
|
|
|
49
|
+
<table>
|
|
50
|
+
<tr><td align="center" width="900">
|
|
51
|
+
|
|
52
|
+
<p><b>Free live session. Saturday, November 14, 10 AM Eastern, on Zoom.</b></p>
|
|
53
|
+
|
|
54
|
+
<p>Two New Premium AI Employees: Live Release. The replay goes to everyone who registers.</p>
|
|
55
|
+
|
|
56
|
+
<p><a href="https://club.reinventing.ai/events/final-countdown-lifetime-last-call?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=nov14-session"><b>Save a free seat</b></a></p>
|
|
57
|
+
|
|
58
|
+
<p><sub>The eight AI Employees stay free and MIT. The club has Free, Pro and Lifetime plans. <a href="https://club.reinventing.ai/pricing?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=nov15-pricing">Plans</a>.</sub></p>
|
|
59
|
+
|
|
60
|
+
</td></tr>
|
|
61
|
+
</table>
|
|
62
|
+
|
|
46
63
|
## AI Employees
|
|
47
64
|
|
|
48
65
|
An AI Employee is a folder of scheduled routines that covers one business role. It runs on your own machine, on the AI agent you already use, and briefs you every morning. They drive your browser and your PC the way you do, and every run makes the next one better. Eight roles, every routine and schedule in this repo, nothing held back.
|
|
49
66
|
|
|
50
67
|
Created by [Mark Fulton](https://www.reinventing.ai/?utm_source=github&utm_medium=readme&utm_campaign=ai-employees) of Reinventing.AI, founder of [Vibe Coding is Life](https://facebook.com/groups/vibecodinglife) (340,000+ members). Running my own business every weekday since August 27, 2026.
|
|
51
68
|
|
|
69
|
+
New to building with AI agents? Start with [What is vibe coding](https://insights.reinventing.ai/vibe-coding?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=vibe-coding-hub), the free reference that explains the practice, the tools and how to do it well.
|
|
70
|
+
|
|
71
|
+
<p align="center">
|
|
72
|
+
<img src="assets/employees/team.webp" width="960" alt="The eight AI Employees, each with its own identity and working tools">
|
|
73
|
+
</p>
|
|
74
|
+
|
|
52
75
|
## The eight AI Employees
|
|
53
76
|
|
|
54
77
|
| Employee | What it owns | Routines |
|
|
55
78
|
|---|---|---|
|
|
56
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/gtm-engineer.webp" width="72" height="72" alt=""><br>[**GTM Engineer**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-gtm-engineer#gtm-engineer) | Your launch: positioning, the launch board, outbound drafts, directory and press forms, the weekly scoreboard | 8 |
|
|
57
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/seo-employee.webp" width="72" height="72" alt=""><br>[**SEO/AEO Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-seo-employee#seo-employee) | Keyword research, one article a weekday, publishing, indexing, rank review and your visibility in AI answers | 8 |
|
|
58
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/web-dev-employee.webp" width="72" height="72" alt=""><br>[**Web Dev Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-web-dev-employee#web-dev-employee) | Site health, error triage, small changes on a branch, dependency review | 8 |
|
|
59
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/social-media-employee.webp" width="72" height="72" alt=""><br>[**Social Media Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-social-media-employee#social-media-employee) | Posts drafted in your voice for each platform, a veto window, replies drafted for you | 7 |
|
|
60
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/ad-manager-employee.webp" width="72" height="72" alt=""><br>[**Ad Manager Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-ad-manager-employee#ad-manager-employee) | Account reads, creative sets, build sheets and the weekly change list. Money moves only when you approve | 7 |
|
|
61
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/sales-employee.webp" width="72" height="72" alt=""><br>[**Sales Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-sales-employee#sales-employee) | Prospect sweeps, first touches into your own drafts, follow ups that never go quiet | 7 |
|
|
62
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/customer-satisfaction-employee.webp" width="72" height="72" alt=""><br>[**Customer Satisfaction Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-customer-satisfaction-employee#customer-satisfaction-employee) | Inbox sweep, replies drafted hardest first, churn flags with evidence |
|
|
63
|
-
| <img src="https://club.reinventing.ai/img/employees/thumbs/chief-of-staff.webp" width="72" height="72" alt=""><br>[**Chief of Staff**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-chief-of-staff#chief-of-staff) | Reads every other employee's run log, names what quietly stopped, brings you three moves | 7 |
|
|
79
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/gtm-engineer.webp" width="72" height="72" alt=""><br>[**GTM Engineer**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-gtm-engineer#gtm-engineer) | Your launch: positioning, the launch board, outbound drafts, directory and press forms, the weekly scoreboard | 8 |
|
|
80
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/seo-employee.webp" width="72" height="72" alt=""><br>[**SEO/AEO Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-seo-employee#seo-employee) | Keyword research, one article a weekday, publishing, indexing, rank review and your visibility in AI answers | 8 |
|
|
81
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/web-dev-employee.webp" width="72" height="72" alt=""><br>[**Web Dev Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-web-dev-employee#web-dev-employee) | Site health, error triage, small changes on a branch, dependency review | 8 |
|
|
82
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/social-media-employee.webp" width="72" height="72" alt=""><br>[**Social Media Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-social-media-employee#social-media-employee) | Posts drafted in your voice for each platform, a veto window, replies drafted for you | 7 |
|
|
83
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/ad-manager-employee.webp" width="72" height="72" alt=""><br>[**Ad Manager Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-ad-manager-employee#ad-manager-employee) | Account reads, creative sets, build sheets and the weekly change list. Money moves only when you approve | 7 |
|
|
84
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/sales-employee.webp" width="72" height="72" alt=""><br>[**Sales Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-sales-employee#sales-employee) | Prospect sweeps, first touches into your own drafts, follow ups that never go quiet | 7 |
|
|
85
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/customer-satisfaction-employee.webp" width="72" height="72" alt=""><br>[**Customer Satisfaction Employee**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-customer-satisfaction-employee#customer-satisfaction-employee) | Inbox sweep, replies drafted hardest first, churn flags with evidence | 9 |
|
|
86
|
+
| <img src="https://club.reinventing.ai/img/employees/v2/thumbs/chief-of-staff.webp" width="72" height="72" alt=""><br>[**Chief of Staff**](https://club.reinventing.ai/ai-employees?utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=table-chief-of-staff#chief-of-staff) | Reads every other employee's run log, names what quietly stopped, brings you three moves | 7 |
|
|
64
87
|
|
|
65
|
-
Sixty routines. Every kit, with its full schedule and a sample of its output, is in [employees/](employees).
|
|
88
|
+
Sixty one routines. Every kit, with its full schedule and a sample of its output, is in [employees/](employees).
|
|
66
89
|
|
|
67
90
|
## Install in two steps
|
|
68
91
|
|
|
92
|
+
**The fastest way:** [hire your first AI Employee in the Agent Ops Club](https://club.reinventing.ai/register?next=%2Fmembers%2Fhire&utm_source=github&utm_medium=readme&utm_campaign=ai-employees&utm_content=install-hire). Pick the role, and the club writes the one prompt your agent runs from start to finish: it downloads the kit, researches your business from your website, builds your dashboard and schedules its own routines. A free account, no card, and the Hire Your First AI Employee walkthrough is waiting on the other side. The two steps below are the same install done by hand.
|
|
93
|
+
|
|
69
94
|
**1. Download a kit and extract it to your PC.** Click **Code, Download ZIP** above, or run `npx ai-employees hire gtm-engineer --to <folder>`. Any folder not inside OneDrive, Dropbox, Google Drive or iCloud.
|
|
70
95
|
|
|
71
96
|
**2. Open your agent in that folder and say "Install the GTM Engineer from this folder."** It researches your business from your website, builds your dashboard, schedules its own routines, and stops once to show you its first drafts.
|
|
@@ -106,6 +131,7 @@ Then say "hire the GTM Engineer into D:\AgentOps\gtm-engineer", or run `/ai-empl
|
|
|
106
131
|
- **They get better at your business every run.** When a page changes, the routine fixes its own instructions and tells you in the next brief.
|
|
107
132
|
- **They drive your browser and your PC the way you do.** Signed in as you, on your own machine, from techniques learned on real runs.
|
|
108
133
|
- **You set how far they go.** Everything is drafted, filled and staged, and the last click is yours until you release a channel.
|
|
134
|
+
- **What they write sounds like you, not like AI.** The Anti-AI Voice Check gives every post, email, DM, reply and article a second pass for the shapes people spot as machine written, and rewrites them from the phrases you actually use in your own recordings. Each draft shows you the line it replaced.
|
|
109
135
|
- **One brief each morning.** One push to your phone, only when you are the blocker.
|
|
110
136
|
- **They run on the agent you already use.** Claude Code and twelve others, without a routine changing by one word.
|
|
111
137
|
- **You can direct any of them in chat.** Open a session in the employee's folder and tell it what to do.
|
|
@@ -167,7 +193,7 @@ On a Claude Pro or Max plan, nothing beyond the plan. They spend a share of your
|
|
|
167
193
|
|
|
168
194
|
## Go further
|
|
169
195
|
|
|
170
|
-
The eight are free for good. The Agent Ops Masterclass, the premium software library with a resale license, the live sessions and
|
|
196
|
+
The eight are free for good. The Agent Ops Masterclass, the premium software library with a resale license, the live sessions and premium AI Employees as they ship live in the [Agent Ops Club](https://club.reinventing.ai/?utm_source=github&utm_medium=readme&utm_campaign=ai-employees). A free club account gets you Masterclass Part 1 and its certificate, the session calendar and AI Employee updates.
|
|
171
197
|
|
|
172
198
|
## FAQ
|
|
173
199
|
|
package/docs/FAQ.md
CHANGED
|
@@ -38,6 +38,6 @@
|
|
|
38
38
|
|
|
39
39
|
**Can my fixes go back into the kits?** Yes, and the employee helps. When a routine repairs its own instructions it logs the change. Once a month the employee picks out the repairs that would be just as right on any business, takes your business out of them, and writes a draft to `improvements/contribution-draft-YYYY-MM.md`. The brief names it once. You read it, then paste it into an issue or delete it. No routine ever opens an issue or a pull request.
|
|
40
40
|
|
|
41
|
-
**What is the club, and what is not in this repo?** The eight employees are complete and free for good; nothing in them is held back. The Agent Ops Club
|
|
41
|
+
**What is the club, and what is not in this repo?** The eight employees are complete and free for good; nothing in them is held back. The Agent Ops Club holds the guided walkthrough, the Masterclass, the premium software library with a resale license, the live sessions and premium employees as they ship. A free club account gets you Masterclass Part 1 and its certificate, the session calendar and employee updates. The README's Go further section has the link.
|
|
42
42
|
|
|
43
43
|
**Where do I ask something not here?** Discussions. Issues are answered within a working day for the first month; after that, Discussions is where the community answers and I read every thread.
|
package/docs/RELEASING.md
CHANGED
|
@@ -1,10 +1,10 @@
|
|
|
1
1
|
# Publishing this update
|
|
2
2
|
|
|
3
|
-
Prepared package version: 1.
|
|
3
|
+
Prepared package version: 1.10.0. SEO/AEO kit: 1.11.0. Social Media, GTM Engineer, Sales, Customer Satisfaction and Chief of Staff: 1.10.0. Ad Manager and Web Dev: 1.9.0. Standard: 1.5.
|
|
4
4
|
|
|
5
5
|
Preparation checks completed on 2026-10-02: npm test, installer listing, whitespace check, package dry run, and strict validation of both the marketplace and plugin manifests passed. The package preview contains all eight work-cycle helpers and the GSC helper, with no operating state. npm reported 1.8.0 as the published version at that check. No publication or live scheduled model evaluation was performed.
|
|
6
6
|
|
|
7
|
-
GitHub and npm are separate distribution steps. Commit and push the reviewed source first, then publish from that same clean checkout. The
|
|
7
|
+
GitHub and npm are separate distribution steps. Commit and push the reviewed source first, then publish from that same clean checkout. The 61 routines are the roster. Do not run npm version again for this prepared release.
|
|
8
8
|
|
|
9
9
|
From the repository root in PowerShell:
|
|
10
10
|
|
|
@@ -18,7 +18,7 @@ if ($LASTEXITCODE -ne 0) { throw "Package inspection failed" }
|
|
|
18
18
|
npm view ai-employees version
|
|
19
19
|
```
|
|
20
20
|
|
|
21
|
-
Confirm that 1.
|
|
21
|
+
Confirm that 1.10.0 has not already been published and that the packed files contain `scripts/voice-check.mjs` in the six copy kits, with no member state. Validate the plugin with `claude plugin validate . --strict` on a machine with the Claude CLI. JSON/version checks in npm test are useful but do not replace the vendor validator.
|
|
22
22
|
|
|
23
23
|
For the maintainer to publish after GitHub is current:
|
|
24
24
|
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
# Ad Manager
|
|
2
2
|
|
|
3
|
-
<img src="https://club.reinventing.ai/img/employees/ad-manager-employee.webp" alt="Ad Manager Employee" width="100%">
|
|
3
|
+
<img src="https://club.reinventing.ai/img/employees/v2/ad-manager-employee.webp" alt="Ad Manager Employee" width="100%">
|
|
4
4
|
|
|
5
5
|
|
|
6
6
|
**Role:** paid advertising, end to end, up to the spend button.
|
|
@@ -32,7 +32,7 @@ Ask it to read one file inside another Employee's folder, by absolute path, and
|
|
|
32
32
|
Ask it for the current local time and the timezone id, then check both against your own clock. Every routine's first act after the pause check is a window check, and a routine that cannot read a clock records `failed` and stops. It will never assume a timezone and it will never trust one remembered from a previous run, because you might have moved.
|
|
33
33
|
|
|
34
34
|
**4. Can it run a local command?**
|
|
35
|
-
Ask it to run `node --version`. You need Node 18 or newer for the
|
|
35
|
+
Ask it to run `node --version`. You need Node 18 or newer for the scripts inside the kit, `scripts/runlog.mjs`, `scripts/copy-check.mjs` and `scripts/voice-check.mjs`. All are dependency free. There is no install step and no package file.
|
|
36
36
|
|
|
37
37
|
**And a fifth, which decides one routine rather than the kit.** Can it drive a browser that carries your own logged in sessions? The kit never logs in. It never creates an account, never types a password, never completes a captcha. It inherits a browser you are already signed in to. So a harness that launches a clean automated browser has handed it a browser with no session, and every read of a surface behind a sign in lands on a wall. The routine will do the correct thing, which is to record `blocked-login`, change nothing, and tell you. It will do that every week.
|
|
38
38
|
|
|
@@ -215,7 +215,7 @@ Run a local command and read its output.
|
|
|
215
215
|
| Codex | Its shell, inside the sandbox | `expected` |
|
|
216
216
|
| Antigravity | Its shell | `expected` |
|
|
217
217
|
|
|
218
|
-
**Absent:** `runlog.append` and `
|
|
218
|
+
**Absent:** `runlog.append`, `copy.check` and `voice.check` take their in agent routes instead, described in section 6. All three still happen. None is skipped. The run record says which route it took.
|
|
219
219
|
|
|
220
220
|
---
|
|
221
221
|
|
|
@@ -595,6 +595,29 @@ node "«COS_ROOT»/scripts/copy-check.mjs" --file <path> --dest <destination> [-
|
|
|
595
595
|
|
|
596
596
|
**The banned word, opener, and closer lists ship inside the script.** This kit has no voice file, because nothing it writes is outbound copy: every page it produces is read by one person. If a word keeps coming through that you do not want, write one line into the `## Corrections` section of the routine that keeps producing it, and that routine amends its own instructions on its next run.
|
|
597
597
|
|
|
598
|
+
### `voice.check`
|
|
599
|
+
The Anti-AI Voice Check: the second pass on the one file this kit writes for somebody other than you to read, the monthly contribution draft you may paste into a public issue, after `copy.check` passes.
|
|
600
|
+
|
|
601
|
+
| Harness | Route | Confidence |
|
|
602
|
+
|---|---|---|
|
|
603
|
+
| Claude Code | `shell.run` on `scripts/voice-check.mjs`, then the same rules applied in agent | `confirmed` |
|
|
604
|
+
| OpenClaw | `shell.run` on `scripts/voice-check.mjs`, then in agent | `expected` |
|
|
605
|
+
| Hermes | `shell.run` if present, then in agent | `unknown` |
|
|
606
|
+
| OpenCode | `shell.run` on `scripts/voice-check.mjs`, then in agent | `expected` |
|
|
607
|
+
| Grok Bot | `shell.run` on its cloud computer, then in agent | `expected` |
|
|
608
|
+
| Codex | `shell.run` on `scripts/voice-check.mjs`, then in agent | `expected` |
|
|
609
|
+
| Antigravity | `shell.run` on `scripts/voice-check.mjs`, then in agent | `expected` |
|
|
610
|
+
|
|
611
|
+
**Absent the script:** the in agent route reads the rule list in `CONTRACT.md` under `voice.check`, applies it to the lines line by line, and the run record says `voice-check: in-agent`. **It is never skipped.**
|
|
612
|
+
|
|
613
|
+
One interface, used verbatim at every call site:
|
|
614
|
+
|
|
615
|
+
```
|
|
616
|
+
node "«COS_ROOT»/scripts/voice-check.mjs" --file <path> --kind <kind> [--json]
|
|
617
|
+
```
|
|
618
|
+
|
|
619
|
+
`--kind` is one of `post`, `dm`, `email`, `reply`, `script`, `article`, `brief`, and this kit calls only `brief`. `--selftest` confirms the script runs. `--profile` is there because one script serves every AI Employee on the machine, and this kit never calls it. **This kit has no voice file**, so the script runs its neutral rules alone and says so in its own output, which is correct here.
|
|
620
|
+
|
|
598
621
|
### `schedule.register`
|
|
599
622
|
Register, inspect, or change a recurring job named after a routine id **in this kit**.
|
|
600
623
|
|
|
@@ -1,5 +1,14 @@
|
|
|
1
1
|
# Chief of Staff: release history
|
|
2
2
|
|
|
3
|
+
## 1.10.0, 2026-11-19
|
|
4
|
+
|
|
5
|
+
The Anti-AI Voice Check: a second pass on everything this Employee writes for somebody else to read, rewritten in your own voice.
|
|
6
|
+
|
|
7
|
+
- **`cos-charter-and-fleet-audit` B4a.3** runs the check (`--kind brief`) on the lines it writes into the monthly contribution draft, the one file you may paste somewhere public. The quoted kit text is never rewritten, and an item that still carries a tell is taken out of the draft. The Chief of Staff keeps no voice file, so the check runs its neutral rules.
|
|
8
|
+
- **`scripts/voice-check.mjs`, new, the same file in every kit that writes copy.** No dependencies, with its own self test. It flags the shapes people spot as machine written, which a banned word list cannot see: three sentences in a row that start with "your", a standalone "Not someday." fragment, "It's not X, it's Y", a prophecy about who will win, an opener that orders the reader about, stock phrases and the lines a general chatbot writes for any business ("Here's your sign."), engagement bait such as "Tag someone", three clipped fragments on one line, and copy with nobody in it. `--kind` (`post`, `dm`, `email`, `reply`, `script`, `article`, `brief`) decides which shapes apply, so a support reply may say "your order" five times and an article may have no "I". `--profile` counts how you actually talk from your own recordings and posts.
|
|
9
|
+
- **It reads you, never anybody else.** An optional `## How I talk` section in your voice file holds your own phrases, how you open and close, the stock lines that are not you, and any shape you really do use, each quoted with its source and a count. Your phrases are never flagged, your "not me" list always is, and every rewrite takes its words from that section and your own samples.
|
|
10
|
+
- `CONTRACT.md` and `CAPABILITIES.md` add the `voice.check` capability, its one interface and its in agent route. The install runs its self test.
|
|
11
|
+
|
|
3
12
|
## 1.9.0, 2026-10-02
|
|
4
13
|
|
|
5
14
|
- Shared work cycle: evidence of useful delivery, scoped blockers, bounded experiments, configured handoffs and compact progress in existing briefs.
|
|
@@ -152,8 +152,9 @@ These ship with the kit. No routine rewrites them. Each ends with a `## Correcti
|
|
|
152
152
|
|---|---|---|
|
|
153
153
|
| `scripts/runlog.mjs` | ships with the kit | the `runlog.append` capability |
|
|
154
154
|
| `scripts/copy-check.mjs` | ships with the kit | the `copy.check` capability |
|
|
155
|
+
| `scripts/voice-check.mjs` | ships with the kit, identical in every kit that writes copy | the `voice.check` capability |
|
|
155
156
|
|
|
156
|
-
|
|
157
|
+
All three are dependency free and take one interface each, defined in section 3.4. None is optional and none may be described in the present tense by any file until it exists on disk.
|
|
157
158
|
|
|
158
159
|
### 2.3 Charter
|
|
159
160
|
|
|
@@ -490,7 +491,7 @@ Each capability below carries a route preference order. **A route is tried in or
|
|
|
490
491
|
| `file.read` | Read a file as text | harness file read, then shell | None. The kit does not run without it |
|
|
491
492
|
| `file.write` | Write a file, temp path plus rename for anything a crash could truncate | harness file write, then shell | None |
|
|
492
493
|
| `file.list` | List paths under a folder | harness glob, then shell | Enumerate from the known paths in section 2 and note the degradation. **The bounded discovery search in `cos-charter-and-fleet-audit` needs this**, so without it the fleet map holds this Employee's own root alone and says so |
|
|
493
|
-
| `shell.run` | Run a local command and read its output | harness shell | `runlog.append` and `
|
|
494
|
+
| `shell.run` | Run a local command and read its output | harness shell | `runlog.append`, `copy.check` and `voice.check` fall back to their in-agent routes below |
|
|
494
495
|
|
|
495
496
|
### 3.2 Browser
|
|
496
497
|
|
|
@@ -533,6 +534,7 @@ Every capability in this table degrades the same way when the harness has no bro
|
|
|
533
534
|
|---|---|---|---|
|
|
534
535
|
| `runlog.append` | Append exactly one validated run record. Validates the shape, validates `status` against the closed list of eight, refuses secret-shaped substrings, writes UTF-8 with no byte order mark, and repairs a stray mark at the head of the file | `shell.run` on `scripts/runlog.mjs`, then a direct append performing the same validation in the agent | If neither is possible, write the record as the last line of `brief-latest.md` under a heading `UNRECORDED RUN` and stop. A run with no record is a run that will be repeated |
|
|
535
536
|
| `copy.check` | The scripted judge for any text about to be written into a brief, a dossier, a weekly page, a charter file, or a dashboard partial | `shell.run` on `scripts/copy-check.mjs`, then the same rule set applied in the agent, marked in the run record as `copy-check: in-agent` | Never skip it. The in-agent route is a degradation, not an exemption |
|
|
537
|
+
| `voice.check` | The Anti-AI Voice Check: the second pass on any copy a person other than the member will read. In this kit that is one file, the monthly contribution draft the member may paste into a public issue. Finds the shapes readers recognise as machine written and names each one as a tell or a warn | `shell.run` on `scripts/voice-check.mjs`, then the same rules applied in the agent, marked in the run record as `voice-check: in-agent` | Never skip it where it applies. It runs after `copy.check` passes, every tell is rewritten in plain words from the source line, and both run again on the rewrite |
|
|
536
538
|
| `schedule.register` | Register, inspect, or change a recurring job named after a routine id **in this kit** | harness scheduler, then the OS scheduler through `shell.run`, then write the exact commands to `«COS_ROOT»/schedule-commands.txt` and name that file in the report and the brief | The kit still runs when launched by hand. Nothing about a routine's behaviour depends on which of the three registered it |
|
|
537
539
|
|
|
538
540
|
**A registered job's only content is the invocation that runs one routine unattended in `«COS_ROOT»`.** What that invocation looks like is a property of the harness, so it lives in `CAPABILITIES.md` section 9.2a as one row per harness and nowhere else. Three rules sit above every route: one job per routine, never a chained job; one routine proved by hand before seven are registered; and **never a job belonging to any routine outside this kit**, whatever it looks like from the outside.
|
|
@@ -559,6 +561,16 @@ node "«COS_ROOT»/scripts/copy-check.mjs" --file <path> --dest <destination> [-
|
|
|
559
561
|
|
|
560
562
|
Do not eyeball any of these. **The script is the judge.** A stated preference has never been enough.
|
|
561
563
|
|
|
564
|
+
**`voice.check` has exactly one interface and every call site uses it verbatim:**
|
|
565
|
+
|
|
566
|
+
```
|
|
567
|
+
node "«COS_ROOT»/scripts/voice-check.mjs" --file <path> --kind <kind> [--json]
|
|
568
|
+
```
|
|
569
|
+
|
|
570
|
+
`--kind` is one of `post`, `dm`, `email`, `reply`, `script`, `article`, `brief`, and decides which shape rules apply. **This kit calls one of the seven, `brief`**, at its one call site, `cos-charter-and-fleet-audit` Step B4a.3, on the lines it wrote into `improvements/contribution-draft-YYYY-MM.md`. That is the only file this kit writes for somebody other than the member to read: the brief, the dossiers, the weekly pages, the charter and the dashboard are each read by one person, and the paste ready correction lines in a dossier are read by an agent. `--selftest` confirms the script runs. `--profile <files>` exists because one script serves every AI Employee on the machine, and this kit never calls it. Exit `0` is clean or warns only, `1` is one or more tells, `2` is bad usage.
|
|
571
|
+
|
|
572
|
+
**What `voice.check` flags as a tell**: an em or en dash; a stock phrase every generated draft reaches for; a label that announces the next line instead of saying it ("Here's the honest answer."); a line a general chatbot writes for any business ("Here's your sign.", "that one friend who"); engagement bait ("Tag someone") outside a support reply, an article or a brief; a phrase listed under `### Tells that are not me`; coach-speak; consultant-memo vocabulary; a prophecy about who will win; a standalone "Not X." fragment; the "It's not X, it's Y" frame; three or more sentences in a row starting with the same drum word; three clipped fragments on one line; three or more "X, Y and Z" lists in one piece; an opener that orders the reader about or asks a rhetorical question; an abstract noun as the hook with nobody in the copy; no I, my, we or our anywhere; five or more "your" outnumbering the writer two to one; and a "Thoughts?" close. Which of these apply depends on `--kind`, and warns are read with judgment. **The script reads `## How I talk` in a kit's voice file, and this kit has none**, for the reason in rule 4 above, so it runs the neutral rules alone and says so in its own output. That is correct and is never fixed by creating a voice file: the contribution draft speaks for the kit's own repair, its rewrite takes its words from the changelog line it reports, and it never borrows the member's voice or anybody else's.
|
|
573
|
+
|
|
562
574
|
**Two failures are the ones this kit actually causes, and each has one mechanical answer.** A dash inside a blocker string or a quote written by somebody else becomes a comma, with the untouched original named one file away. A count that reads as a claim gains the file path it was read from, or is rewritten as a date. Neither answer is ever to add a line to `evidence/sourced.md` so a sentence passes: five of the seven routines are not appenders to that file at all, and the two that are may only append a figure they read out of a file inside this folder this run.
|
|
563
575
|
|
|
564
576
|
---
|
|
@@ -60,7 +60,7 @@ These apply from now until I remove them. They are not negotiable inside this se
|
|
|
60
60
|
1. Read these in full, in this order, including the `## Corrections` section at the bottom of each: `«COS_ROOT»/CONTRACT.md`, `ROLE.md`, `CAPABILITIES.md`, `SCHEDULE.md`. Where anything in this prompt and `CONTRACT.md` disagree, the contract wins. Where the contract and my own workspace rule file disagree, mine wins.
|
|
61
61
|
2. Confirm `«COS_ROOT»` exists and is writable, and that it is **not** inside OneDrive, Dropbox, Google Drive, or iCloud. If it is inside a synced folder, copy the whole kit to a local path that is not synced, continue there, leave the original in place with a short pointer file beside it naming the new root, and put the new path as the first line of your day one report. Do not ask me. These routines write state mid run and a sync client corrupts exactly the file that tells tomorrow's run what already happened. On a harness whose computer is not a machine I use (Grok Bot runs this kit on its own cloud computer), the sync rule does not apply: the kit lives on that computer, you install it there, and the brief reaches me through `brief.deliver` in `CONTRACT.md` section 3.2a, because I never open that computer's files.
|
|
62
62
|
3. Read the local timezone id and the local wall clock time using rule 16. Record both. Every time you write from here on is in that zone.
|
|
63
|
-
4. Check `node --version`. It must be 18 or newer. Then run the
|
|
63
|
+
4. Check `node --version`. It must be 18 or newer. Then run the four self tests once: `node "«COS_ROOT»/scripts/copy-check.mjs" --selftest`, `node "«COS_ROOT»/scripts/voice-check.mjs" --selftest`, `node "«COS_ROOT»/scripts/runlog.mjs" --selftest` and `node "«COS_ROOT»/scripts/guard.mjs" --selftest`. If Node is missing or a self test fails, record the blocker, take the in agent routes for `copy.check`, `voice.check` and `runlog.append` described in `CAPABILITIES.md` section 6, and carry on. None of them is ever skipped. **Then check the login, and stop if it is missing:** run `claude auth status` through `shell.run`, or your harness's own equivalent. If it reports `loggedIn: false`, or no signed in account, stop here and tell me in one plain sentence that nothing after this phase can run until I open a terminal, run `claude`, and complete `/login` myself. A scheduled run that is not logged in exits in under a second with `Not logged in` and writes nothing, so there is nothing to gain by continuing. Never try to log in for me, and never enter or write a key.
|
|
64
64
|
5. Detect your own capabilities live, using the framing in `CAPABILITIES.md` section 1.2. Answer it for yourself rather than asking me. **Settle two things specifically and write down both answers.** First: can you read a file by absolute path in a folder outside this one, because that is this Employee's whole job and a harness that sandboxes reads to one directory will only ever watch itself. Second: does your browser control attach to a browser I am already signed in to, or does it start a clean one. Cache nothing. Detection happens at the top of every run.
|
|
65
65
|
- For each row of `CAPABILITIES.md` section 4b, say whether that connection is present on this harness, under what name, and whether it is read only. Write the answers in your working notes. A row that is absent costs nothing today: the browser lane in section 4 is the route, and section 7 says what that read produces without one.
|
|
66
66
|
6. Confirm seven folders exist under `«COS_ROOT»/routines/` and that each holds a `SKILL.md` whose YAML `name` equals its folder name exactly. The seven are `cos-fleet-reconcile`, `cos-fault-dossier`, `cos-market-sweep`, `cos-metrics-review`, `cos-decision-brief`, `cos-charter-and-fleet-audit`, `cos-decision-review`. If one is missing or its name key differs, record it as a blocker and carry on with the ones that are correct.
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
# Chief of Staff
|
|
2
2
|
|
|
3
|
-
<img src="https://club.reinventing.ai/img/employees/chief-of-staff.webp" alt="Chief of Staff" width="100%">
|
|
3
|
+
<img src="https://club.reinventing.ai/img/employees/v2/chief-of-staff.webp" alt="Chief of Staff" width="100%">
|
|
4
4
|
|
|
5
5
|
|
|
6
6
|
**Role:** the Employee that watches the other Employees.
|
|
@@ -190,7 +190,7 @@ Ten things. The install checks the ones it can, and it stops in plain words on t
|
|
|
190
190
|
1. **An agent harness that can read and write files, run a command, read the clock, and drive the browser you are signed in to.** Any of the eleven this kit is built for: Claude Code, OpenClaw, Hermes, OpenCode, Grok Bot, Codex, Antigravity, Pi, Cline, Qwen Code or DeepSeek. On Claude Code that means a plan that includes it, Pro, Max, Team or Enterprise; the free plan does not. An API key works for the file routines and turns the browser lane off. The other ten run on whatever account they already run on.
|
|
191
191
|
2. **The harness installed, in a shape you can schedule.** On Claude Code that is either the Claude Desktop app, which includes it and has a local scheduler of its own, or the CLI, which you pair with the operating system's scheduler. `CAPABILITIES.md` section 2 gives the same answer for each of the other ten.
|
|
192
192
|
3. **Logged in, by a human, once.** On Claude Code, run `claude`, then `/login`, and confirm with `claude auth status`; on another harness, its own login check. A scheduled run that is not logged in exits in under a second and writes nothing, so the install checks this before it does anything else and the launcher writes a `failed` run record when it happens later.
|
|
193
|
-
4. **Node 18 or newer.** The
|
|
193
|
+
4. **Node 18 or newer.** The scripts in `scripts/` are dependency free and need nothing installed.
|
|
194
194
|
5. **On Windows with Claude Code, Git for Windows.** Claude Code needs it for its shell and the Desktop app needs it for local sessions. Not WSL if you want the browser lane, because the browser bridge does not run inside WSL. Another harness has its own Windows requirements, listed on its install page.
|
|
195
195
|
6. **The browser lane, if you want it:** a harness that drives a browser carrying your own logins rather than a clean automated profile. On Claude Code that is Google Chrome or Microsoft Edge, the Claude in Chrome extension, a login based session rather than an API key, and the site permissions granted in the extension before the first scheduled run; `CAPABILITIES.md` section 2 says what it is on each of the others. The routines share your browser's own login state and never sign in to anything.
|
|
196
196
|
7. **A scheduler.** Your harness's own where it has one: the Claude Desktop app's local scheduled tasks, the built in cron in OpenClaw, Hermes, Cline and Qwen Code, Codex scheduled runs, the Antigravity job runner, the DeepSeek plugin. Otherwise Windows Task Scheduler with the launchers in `run/`, launchd on macOS, or cron on Linux. `CAPABILITIES.md` section 9 has the commands, and the repo's `docs/INSTALL.md` walks each one.
|
|
@@ -270,6 +270,8 @@ This kit is **v1.2.0**. The version is in `VERSION` at the root, and `CHANGELOG.
|
|
|
270
270
|
|
|
271
271
|
Employees update one at a time. They share nothing but a scheduler, so running one at v1.0.0 and another at a later version is fine and needs no coordination.
|
|
272
272
|
|
|
273
|
+
Once a month it may also leave you a short draft of the repairs it made to its own instructions that look useful to everybody running this kit, ready for you to paste into the project's issue tracker if you choose. That draft is the one file it writes for other people to read, so it goes through the Anti-AI Voice Check first: any line that reads as machine written is rewritten in plain words, and the monthly report shows you what changed.
|
|
274
|
+
|
|
273
275
|
When an update lands, `CHANGELOG.md` says exactly which files changed and which are yours. The short version: everything under `strategy/`, `state/`, your ledgers, your learned `recipes/*.json`, and the `## Corrections` at the foot of every file are **yours and are never overwritten**. Everything else is ours and safe to replace. The one file holding both is a routine's `SKILL.md`, so copy your `## Corrections` block out before replacing it and paste it back after.
|
|
274
276
|
|
|
275
277
|
## When something goes wrong
|
|
@@ -304,6 +306,7 @@ No routine ever writes a secret, a credential, a quote, a person's name, or anot
|
|
|
304
306
|
scripts/runlog.mjs the only sanctioned way to append a run record
|
|
305
307
|
scripts/guard.mjs the pause, window and period guards, run before any document is read
|
|
306
308
|
scripts/copy-check.mjs the scripted judge for anything about to be written
|
|
309
|
+
scripts/voice-check.mjs the Anti-AI Voice Check, on the one file others read
|
|
307
310
|
charter/ business, constraints, metric-map, fleet-map,
|
|
308
311
|
priorities, CHANGELOG
|
|
309
312
|
fleet/ fleet.json, observations.jsonl, inbox.jsonl
|
|
@@ -1 +1 @@
|
|
|
1
|
-
1.
|
|
1
|
+
1.10.0
|
|
@@ -97,7 +97,7 @@ You open more folders than any other routine in this kit. You write into none of
|
|
|
97
97
|
| `schedule-commands.txt` | Only where `schedule.register` has no other route, written expanded and named first in the report |
|
|
98
98
|
| `recipes/<flow>.json` | Only flows whose `owner` reads `cos-charter-and-fleet-audit` |
|
|
99
99
|
| `state/cos-charter-and-fleet-audit.json` | Your own state, temp path plus rename |
|
|
100
|
-
| `state/kit-update.json`, and `improvements/contribution-draft-YYYY-MM.md` in a month that has one | Whole files, one writer, this routine, on the monthly pass. Step B4a. Your own state file, `state/kit-update.json`, and nothing else under `state
|
|
100
|
+
| `state/kit-update.json`, and `improvements/contribution-draft-YYYY-MM.md` in a month that has one | Whole files, one writer, this routine, on the monthly pass. Step B4a. Your own state file, `state/kit-update.json`, and nothing else under `state/`, except `state/contribution-lines.tmp.md`, deleted in Step B4a.3 |
|
|
101
101
|
| `improvements/CHANGELOG.md` | Appended, only when you amended this file |
|
|
102
102
|
| `archive/**` | Files older than ninety days, moved with their paths preserved. Nothing is ever deleted |
|
|
103
103
|
| `runlog.jsonl` | Exactly one record, through `runlog.append` |
|
|
@@ -261,13 +261,14 @@ Then check it. If any path segment matches, case insensitively, `OneDrive`, `Dro
|
|
|
261
261
|
|
|
262
262
|
**Do not stop to ask for a different folder.** Choose one: the nearest local path outside every synced tree, under the member's own home directory, named after the kit. Create the tree there. Copy everything already present in the launch folder across. **Leave the original in place**, because nothing in this kit is deleted, and write one short pointer file beside it naming the new root. Record the move in `assumptions[]`, write one line into `charter/CHANGELOG.md`, and name the new path in the first line of the report.
|
|
263
263
|
|
|
264
|
-
**A1.3 Confirm the
|
|
264
|
+
**A1.3 Confirm the three scripts.** `scripts/runlog.mjs`, `scripts/copy-check.mjs` and `scripts/voice-check.mjs` ship with the kit. Run the self tests:
|
|
265
265
|
|
|
266
266
|
```
|
|
267
267
|
node "«COS_ROOT»/scripts/copy-check.mjs" --selftest
|
|
268
|
+
node "«COS_ROOT»/scripts/voice-check.mjs" --selftest
|
|
268
269
|
```
|
|
269
270
|
|
|
270
|
-
If `shell.run` is unavailable, or the runtime is missing, or
|
|
271
|
+
If `shell.run` is unavailable, or the runtime is missing, or any script is absent, every one of these capabilities has a second route: `runlog.append` performs the same validation inside the agent, `copy.check` applies the same rule set inside the agent and marks the run record `copy-check: in-agent`, and `voice.check` applies the rules in `CAPABILITIES.md` section 6 inside the agent and marks the run record `voice-check: in-agent`. Take the second route and carry on. **The in agent route is a degradation, never an exemption, and you never skip the check.**
|
|
271
272
|
|
|
272
273
|
Put one line in the report naming what the member would gain by installing the runtime named in `CAPABILITIES.md`. One line, once, not a warning repeated every month.
|
|
273
274
|
|
|
@@ -876,6 +877,29 @@ What changed: «one sentence, from the changelog line»
|
|
|
876
877
|
|
|
877
878
|
**You never send it.** Not an issue, not a pull request, not a `git` command, not a form. Opening an issue publishes under the member's name, which is guardrail 1, and nothing in `RELEASES.md` releases it, because the project's issue tracker is not one of the member's channels. You read no other routine's `SKILL.md` to write the draft. The changelog line is the whole of your evidence.
|
|
878
879
|
|
|
880
|
+
### B4a.3 The two judges, on the lines you wrote, before the draft is final
|
|
881
|
+
|
|
882
|
+
**The contribution draft is the one file this kit writes for somebody other than the member to read.** Every other page it produces is read by one person. This one is pasted, under the member's name, into a public issue that the kit's maintainers and anybody else following the project read. So the lines you wrote in it get both judges, and the Anti-AI Voice Check is the second, because an issue that reads as machine written is an issue a maintainer skims past, and it goes out with the member's name on it.
|
|
883
|
+
|
|
884
|
+
**Only your own lines are checked**: `What happened` and `What changed` for every item. The heading block is fixed text from this file and `What the kit said` is the kit's own instruction, quoted whole, which is never edited to please a checker. Write those lines alone, one item after another, to `state/contribution-lines.tmp.md`, and run:
|
|
885
|
+
|
|
886
|
+
```
|
|
887
|
+
node "«COS_ROOT»/scripts/copy-check.mjs" --file "«COS_ROOT»/state/contribution-lines.tmp.md" --dest plain --json
|
|
888
|
+
node "«COS_ROOT»/scripts/voice-check.mjs" --file "«COS_ROOT»/state/contribution-lines.tmp.md" --kind brief --json
|
|
889
|
+
```
|
|
890
|
+
|
|
891
|
+
That is the interface, verbatim. `--kind brief` is a plain report of what happened, which is allowed to have no "I" in it and no rhetorical opener. **This kit has no voice file**, so the script runs its neutral rules and says so in its own output; that is correct here, and you never create a voice file to make it quieter. Exit `0` is clean or warns only. Exit `1` means one or more tells. Where `shell.run` is unavailable, apply the same rules in the agent from `CAPABILITIES.md` section 6 and put `voice-check: in-agent` in `notes`. **The in agent route is a degradation, not an exemption.**
|
|
892
|
+
|
|
893
|
+
**Every `tell` is rewritten. A `warn` is read with judgment.** The rewrite is one pass, and only on the line a tell names:
|
|
894
|
+
|
|
895
|
+
**Then one read by judgment, because the script is a list of patterns and a draft can dodge every one of them.** Read the draft as the member saying it out loud to someone they know, and ask three questions of every line: would it fit another business's copy with only the name swapped; is it a clever line rather than something that happened; is the member, or the person they are writing to, actually in it. A line that fails any of the three is rewritten by the same rules as a tell and counted under the rule id `judgment`. Most drafts written from real material pass this read untouched, and those stay exactly as they are.
|
|
896
|
+
|
|
897
|
+
1. **Plain words, from the changelog line itself.** The source of every rewritten sentence is the trigger and the change recorded in `improvements/CHANGELOG.md`, said once, in the order it happened. A stock phrase becomes the fact it was decorating. A "Not X." fragment is cut. "It's not X, it's Y" becomes the one true statement about Y. A prediction about who the fix will help comes out.
|
|
898
|
+
2. **Never import a phrase, a story, a number or a habit that is not in that changelog line**, and never anybody's voice, the member's included, because this file speaks for the kit's own repair and not for them. The redaction from B4a.2 step 5 holds: a rewrite never puts back a name, a domain, a figure or a path that was redacted.
|
|
899
|
+
3. **Then run both judges again on the rewrite**, `copy.check` first, then `voice.check`.
|
|
900
|
+
|
|
901
|
+
**An item whose lines still carry a tell, or now fail `copy.check`, comes out of the draft**, the same way an item that cannot be told without the member's detail comes out in B4a.2. Name it in the run record with `voice-check` as its first failure reason. Where every item comes out, write no draft, advance the cursor, and say nothing about it in the report. Lower `contribution_items` to the count that stayed. In your monthly report, beside the draft's path, write one line `- voice check: N tells found, N fixed (rule ids)`, `0 tells found` included, and under it one `- was:` line per rewritten line, at most three, so the member can read the before beside the after before they paste it. The totals go in `notes`. Delete `state/contribution-lines.tmp.md` in this step and on every exit path.
|
|
902
|
+
|
|
879
903
|
## Step B5. Close the monthly pass
|
|
880
904
|
|
|
881
905
|
Update `registered_times{}` for anything you re-registered. Set `complete: true`. Write the report. Append exactly one run record.
|
|
@@ -972,6 +996,8 @@ If any of the five does not hold, the run is a failure regardless of what else i
|
|
|
972
996
|
| An Employee's schedule file will not parse | Keep the previous rows, mark the block, name the file. Never guess a cadence | continues |
|
|
973
997
|
| A charter file fails the copy check twice | Take the failing line out, replace it with a statement of what is missing, name it | continues, named in `notes` |
|
|
974
998
|
| A dashboard partial fails the copy check twice | Keep it out of the build, put a one line placeholder in its place, name it | continues, named in `notes` |
|
|
999
|
+
| `voice.check` has no shell route | The in agent route from `CAPABILITIES.md` section 6 on the contribution draft lines, one line in `notes`. Never skipped | continues, `voice-check: in-agent` |
|
|
1000
|
+
| A contribution draft item still carries a tell after one rewrite | Take the item out of the draft, `voice-check` as its first failure reason. Never write it in | continues, named in `notes` |
|
|
975
1001
|
| `schedule.register` has no route | Write `schedule-commands.txt`, expanded, and name it first in the report | continues |
|
|
976
1002
|
| Browser control rewrites the local file address | Verify off disk per A8.6 tier one and say it was not viewed. **Never stand up a local server to work around it** | continues |
|
|
977
1003
|
| A flow file this routine owns does not exist yet | `learn-a-recipe`, write only what you verified, one line in the record | continues |
|