patterns

Name: patterns
Author: tkellogg

83stars19forks

How to install patterns?

tkellogg/open-strix/patterns

$npx -y skills add tkellogg/open-strix --skill patterns

Installs into the current project.

Summary

Index of recurring problem shapes — environmental awareness (world-scanning, OS event hooks, browser automation), communication (messaging, multi-agent handoff, async waits), coordination between parallel things (avoiding collisions, oscillation, double-work), and self-correction

Files · 1

View on GitHub

SKILL.md

1	Having problems that look like this?
2
3	* Out-of-date info — world-scanning.md
4	* Awareness of surroundings — world-scanning.md
5	* Getting ahead of problems — world-scanning.md
6	* Sending information or messages between computers — messaging.md
7	* Notifying another agent that something happened — messaging.md
8	* Reacting when the user opens their laptop, locks the screen, plugs in a USB, joins Wi-Fi, etc. — os-events-macos.md / os-events-windows.md / os-events-linux.md
9	* Hooking into anything the operating system already knows about — os-events-{macos,windows,linux}.md
10	* Scraping a site that has no API, especially behind a login wall — browser-automation.md
11	* Driving a SaaS dashboard to click a button or download a file — browser-automation.md
12	* Personal-data dashboards from services that don't expose one — browser-automation.md
13	* Need to wait for a human action (click, login, file drop) without burning tokens — async-tasks.md
14	* "Wake me up when X happens" where X is a one-shot event — async-tasks.md
15	* Pausing this conversation across an indeterminate wait without losing context — async-tasks.md
16	* Handing work to another agent (sub-agent, peer, fresh self) — multi-agent-handoff.md
17	* What survives across a fresh-context turn (async wake-up, schedule fire, compaction) — context-boundaries.md
18	* How to write journal entries that future-you / introspection-you will actually find useful — journal-as-breadcrumbs.md
19	* Picking the right temporal primitive (poller / schedule / loop / async-block / OS cron) — scheduling.md
20	* Building redundancy into channels, scrapers, sources without overengineering — fallback-chains.md
21	* Recognizing your own loops / runaway behavior and stopping yourself — circuit-breaker.md
22	* Stuck on something — what to do instead of grit (edit blocks, edit checkpoint.md, find conflicts) — try-harder.md
23	* The instinct to "try harder" / "do better" / "be more careful" — try-harder.md (almost always the wrong move)
24	* Two parallel things stepping on each other (schedules colliding, pollers in storm, double-messaging, oscillating state files) — coordination.md
25	* "It worked fine for weeks then suddenly two of X happened" — coordination.md (S2 collision)
26	* No durable place to track what you've noticed, committed to, or finished — interest-backlog.md
27	* "I should remember to look into X" / "that was weird but I don't have time now" — interest-backlog.md (log it, drain it later)
28
29	## Two rules worth pulling forward
30
31	These show up repeatedly in the sub-pages but are easy to miss if you only skim the index.
32
33	* The artifact of trying harder is a diff someone else can verify, not a resolution someone else has to trust. If your "fix" is a commitment to do better, it dies the moment your turn ends. See `try-harder.md`.
34	* Journal entries should have handles, not prose. `j_abc123` and `/tmp/rl.md` survive a context boundary; "I spawned a job and will check on it" doesn't. See `journal-as-breadcrumbs.md`.
35
36	It's worth running a five-why's on your problem to see if some of this info applies.

People who liked this also viewed

6 picks

Type

patterns

83stars19forks

View on GitHub

How to install patterns?

tkellogg/open-strix/patterns

$npx -y skills add tkellogg/open-strix --skill patterns

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	Having problems that look like this?
2
3	* Out-of-date info — world-scanning.md
4	* Awareness of surroundings — world-scanning.md
5	* Getting ahead of problems — world-scanning.md
6	* Sending information or messages between computers — messaging.md
7	* Notifying another agent that something happened — messaging.md
8	* Reacting when the user opens their laptop, locks the screen, plugs in a USB, joins Wi-Fi, etc. — os-events-macos.md / os-events-windows.md / os-events-linux.md
9	* Hooking into anything the operating system already knows about — os-events-{macos,windows,linux}.md
10	* Scraping a site that has no API, especially behind a login wall — browser-automation.md
11	* Driving a SaaS dashboard to click a button or download a file — browser-automation.md
12	* Personal-data dashboards from services that don't expose one — browser-automation.md
13	* Need to wait for a human action (click, login, file drop) without burning tokens — async-tasks.md
14	* "Wake me up when X happens" where X is a one-shot event — async-tasks.md
15	* Pausing this conversation across an indeterminate wait without losing context — async-tasks.md
16	* Handing work to another agent (sub-agent, peer, fresh self) — multi-agent-handoff.md
17	* What survives across a fresh-context turn (async wake-up, schedule fire, compaction) — context-boundaries.md
18	* How to write journal entries that future-you / introspection-you will actually find useful — journal-as-breadcrumbs.md
19	* Picking the right temporal primitive (poller / schedule / loop / async-block / OS cron) — scheduling.md
20	* Building redundancy into channels, scrapers, sources without overengineering — fallback-chains.md
21	* Recognizing your own loops / runaway behavior and stopping yourself — circuit-breaker.md
22	* Stuck on something — what to do instead of grit (edit blocks, edit checkpoint.md, find conflicts) — try-harder.md
23	* The instinct to "try harder" / "do better" / "be more careful" — try-harder.md (almost always the wrong move)
24	* Two parallel things stepping on each other (schedules colliding, pollers in storm, double-messaging, oscillating state files) — coordination.md
25	* "It worked fine for weeks then suddenly two of X happened" — coordination.md (S2 collision)
26	* No durable place to track what you've noticed, committed to, or finished — interest-backlog.md
27	* "I should remember to look into X" / "that was weird but I don't have time now" — interest-backlog.md (log it, drain it later)
28
29	## Two rules worth pulling forward
30
31	These show up repeatedly in the sub-pages but are easy to miss if you only skim the index.
32
33	* The artifact of trying harder is a diff someone else can verify, not a resolution someone else has to trust. If your "fix" is a commitment to do better, it dies the moment your turn ends. See `try-harder.md`.
34	* Journal entries should have handles, not prose. `j_abc123` and `/tmp/rl.md` survive a context boundary; "I spawned a job and will check on it" doesn't. See `journal-as-breadcrumbs.md`.
35
36	It's worth running a five-why's on your problem to see if some of this info applies.

People who liked this also viewed

6 picks

Type