new-domain
Skill1 starsChanged 4 months agoMITView on GitHub
Add or retire a life domain by executing EVERY step of the docs/adding-domains.md checklist — skeleton, registry, router, external systems, goals, then a verified trigger test. Use when Alex says "/new-domain", "add a domain", "new life domain", "spin up a domain for X", "retire the X domain", "kill the X domain", or "archive a domain". NOT /domain-hygiene (that scans existing domains for staleness; this one creates or retires them).
2 other projects use a version of this.
Adopted by
Most recently taken up by
- andrefigueira/.contexttheir own version, first seen Oct 8, 2026
- cogotemartinez/ai-second-brain-templatetheir own version, first seen Feb 11, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- andrefigueira/.context★ 253same name
- cogotemartinez/ai-second-brain-template★ 9same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository22 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- commsSkill
- composeSkill
- daily-briefSkill
- domain-hygieneSkill
- grill-meSkill
- idea-deep-diveSkill
- laila-headless-conductSkill
- laila-judgmentSkill
- laila-ops-debuggingSkill
- level-upSkill
- os-auditSkill
- personaSkill
- quarterly-refreshSkill
- roastSkill
- running-briefSkill
- session-wrapSkill
- setupSkill
- specSkill
- validate-ideaSkill
- whats-nextSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 5 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 37
Step A — Skeleton (Tier 1): Create:
- Very specific: line 41
Step B — Registry (Tier 1): Add an entry to config/domain-triggers.json: patterns, context_file, state: "active", plus a meeting_routing …
- Distinctive: line 43
Step C — Router pointer (Tier 3 — propose the diff): If it's a headline domain, prepare a one-line addition to the root AGENTS.md Domains …
- Very specific: line 47
Step E — Goals: If the domain has quarterly goals, add them to the goal source of truth (the external system if one exists, else the …
- Distinctive: line 61
External system: if one exists, propose marking the domain's record retired so dashboards stop counting it as live.
- Very specific: line 63
Sweep for zombie references: search skills, scripts, and configs for the old domains/<name>/ path.
- Very specific: line 80
Never summarize a partially-executed run as complete.
- Distinctive: line 90
Before declaring this skill done (and again after edits):
- Very specific: line 91
Add round-trip: add a throwaway domain (e.g.
- Distinctive: line 97
If a step was skipped, mis-ordered, or a real add/retire later turned out partial (a zombie trigger, an empty goal tree), this skill is …
- Very specific: line 98
If docs/adding-domains.md gains or changes a step, update this skill in the same session — the doc and the skill must never disagree.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
New Domain Skill
Operationalizes docs/adding-domains.md. Domains are cheap to add and cheap to retire — the failure mode is doing either partially: a router pointing at files that don't exist, or an archived domain whose trigger still fires. This skill exists so no step gets skipped. Read docs/adding-domains.md first every run; it is the checklist of record and this skill executes it.
Setup note: the registry is
config/domain-triggers.jsonand the router is the rootAGENTS.md— both must exist. Step 4 (external CRM/goal-tree registration) only applies if you run one; the skill asks rather than assumes. No scheduled jobs are created here — if a new domain later needs a background loop, that job uses the$AGENT_RUNenv convention like every other loop.
Tier classification
- Tier 1 (auto-execute + notify): creating files under
domains/<name>/and adding the registry entry toconfig/domain-triggers.json. Deterministic, reversible, invisible to others. Log tostate/autonomy-audit.json, notify the command channel. - Tier 3 (propose-and-wait): any edit to the root
AGENTS.md— it is the router every session loads, so present the exact diff and wait for approval. External CRM/goal-tree writes are also proposed first. - Retiring is always Tier 3. Moving a domain to the archive is destructive-adjacent; confirm the kill explicitly before touching anything.
Mode: Add (default)
/new-domain or /new-domain <domain-name>
1. Interview (brief — five questions, then move)
- Name — lowercase slug for the directory (
domains/<name>/), display name for headers. - What is it — one or two sentences: scope, what "checking" it should surface.
- Trigger phrases — at least "Check
<name>"; ask for any natural variants Alex would actually say. - Meeting-notes auto-filing — should meeting notes route here? If yes, collect the signals: participant email domains, people, company names, title keywords.
- Headline domain? — does it get a mention in the root
AGENTS.mdDomains line, or is it registry-only?
Also ask whether the domain has goals to seed and whether an external CRM/goal-tree system is in play (skip step 5 cleanly if not).
2. Execute — every step, in order
Track each step in a live checklist (see Reporting). A step is pending -> done -> verified; nothing ships at done.
Step A — Skeleton (Tier 1). Create:
domains/<name>/AGENTS.md— overview, a "Check<Name>" workflow (Read -> Detect -> Act -> Report, with Act split into Tier 1 vs Tier 3 actions), domain rules, trigger phrases. Model it ondomains/ideas/AGENTS.md.domains/<name>/tracking/status.md— with a**Last updated:** YYYY-MM-DDheader (the domain-hygiene scanner keys off it), Current Focus, and an Active Items table. Model ondomains/ideas/tracking/status.md.
Step B — Registry (Tier 1). Add an entry to config/domain-triggers.json: patterns, context_file, state: "active", plus a meeting_routing block (email_domains, people, company_names, title_keywords, filing.correspondence_dir) if the interview said notes should auto-file. Validate the JSON parses after the edit.
Step C — Router pointer (Tier 3 — propose the diff). If it's a headline domain, prepare a one-line addition to the root AGENTS.md Domains section (the active-domains list). Present the diff, wait for approval, then apply. The router routes — one line, no content.
Step D — External CRM/goal tree (Tier 3, only if Alex runs one). Propose the record creation AND the select-option additions on every object type that can be tagged with a domain (goals, tasks, KPIs). Surface the doc's warning verbatim in the proposal: the select-option step is the one that gets skipped — the domain appears in the system but nothing can be tagged to it, so its goal tree renders empty. When an API replaces an options array wholesale: read the full existing array, append the new option, write it ALL back — omitting an existing option deletes it.
Step E — Goals. If the domain has quarterly goals, add them to the goal source of truth (the external system if one exists, else the domain's status.md) and regenerate any exported views (state/goals.md).
Step F — Verify end-to-end. Have Alex (or a fresh session) say the trigger phrase and confirm: the registry resolves it, the domain's AGENTS.md loads, and the Check workflow runs against the new status.md. Only after this does the checklist read fully verified.
3. Close out
Log the Tier 1 actions to state/autonomy-audit.json, notify the command channel ("Domain <name> added — 6/6 steps verified"), and note the decision in knowledge/decisions/YYYY-MM.md.
Mode: Retire
/new-domain retire <domain-name> — all Tier 3. Confirm first: "Retiring <name>: archive the directory, kill the trigger, remove the router line. Confirm?"
- Archive the directory:
git mv domains/<name>/ domains/_archive/<name>-killed-YYYY-MM-DD/(plainmvif not tracked). WriteARCHIVED.mdinside it: kill reason (Alex's words), last known state (pull from itsstatus.md), and revival cost — what it would take to bring it back. - Flip the registry: in
config/domain-triggers.json, remove the trigger entry and append a row to thearchivedarray (domain,state: "archived", kill date, archive path). Validate the JSON. - External system: if one exists, propose marking the domain's record retired so dashboards stop counting it as live. Do NOT remove its select-option — historical records still reference it.
- Router: propose the root
AGENTS.mddiff removing the domain from the active list (and mentioningdomains/_archive/if it isn't already). - Sweep for zombie references: search skills, scripts, and configs for the old
domains/<name>/path; fix or flag each hit. - Verify: say the old trigger phrase in a fresh session — it must surface "that domain was killed
<date>, archived at<path>" instead of running a Check workflow. A retirement that skips steps 2-4 leaves a zombie trigger that errors weeks later.
Reporting
Maintain and print the checklist at every pause point:
# Add domain: <name>
A. Skeleton (AGENTS.md + status.md) [verified]
B. Registry entry (domain-triggers.json) [verified]
C. Router pointer (root AGENTS.md) [awaiting approval — diff above]
D. External CRM/goal tree [n/a — none configured]
E. Goals seeded + exports regenerated [pending]
F. Fresh-session trigger test [pending]Never summarize a partially-executed run as complete. If the session must end mid-checklist, record the remaining steps in state/active-tasks.json (owner: laila) so the heartbeat nags until F is verified.
When to invoke proactively
- Alex describes a new sustained commitment (a client engagement, a health program, a build) and starts asking Laila to track it — that's a domain forming; offer this skill instead of letting the tracking sprawl into daily notes.
- A domain's status has read "nothing active" for a quarter and its Check runs come back empty — propose retirement rather than letting it rot (killed is a verdict, not a deletion; the archive keeps the post-mortem).
- An engagement domain (the Acme pattern — created for one client commitment) hits the end of its contract: retirement is part of the close-out, not an afterthought.
Verification (MANDATORY)
Before declaring this skill done (and again after edits):
- Add round-trip: add a throwaway domain (e.g.
testbed), confirm all six steps reachverifiedincluding the fresh-session trigger test, then retire it via this same skill and confirm the trigger surfaces the archived state. - JSON integrity:
config/domain-triggers.jsonparses after both the add and the retire edits. - Tier discipline: confirm the run never edited the root
AGENTS.mdwithout showing a diff and getting approval first.
Self-improvement (MANDATORY)
- If a step was skipped, mis-ordered, or a real add/retire later turned out partial (a zombie trigger, an empty goal tree), this skill is wrong, not Alex. Propose the checklist edit immediately, or log
{"skill": "new-domain", "correction": "...", "date": "YYYY-MM-DD"}tostate/skill-feedback.jsonl. - If
docs/adding-domains.mdgains or changes a step, update this skill in the same session — the doc and the skill must never disagree.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

