@harness-lab/cli 0.2.2 → 0.2.3
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 +1 -0
- package/assets/workshop-bundle/SKILL.md +22 -11
- package/assets/workshop-bundle/bundle-manifest.json +82 -22
- package/assets/workshop-bundle/content/challenge-cards/deck.md +3 -3
- package/assets/workshop-bundle/content/challenge-cards/locales/en/deck.md +38 -0
- package/assets/workshop-bundle/content/czech-editorial-review-checklist.md +88 -0
- package/assets/workshop-bundle/content/facilitation/master-guide.md +9 -9
- package/assets/workshop-bundle/content/project-briefs/locales/en/code-review-helper.md +31 -0
- package/assets/workshop-bundle/content/project-briefs/locales/en/devtoolbox-cli.md +31 -0
- package/assets/workshop-bundle/content/project-briefs/locales/en/doc-generator.md +31 -0
- package/assets/workshop-bundle/content/project-briefs/locales/en/metrics-dashboard.md +31 -0
- package/assets/workshop-bundle/content/project-briefs/locales/en/standup-bot.md +31 -0
- package/assets/workshop-bundle/content/style-examples.md +1 -1
- package/assets/workshop-bundle/content/style-guide.md +4 -2
- package/assets/workshop-bundle/content/talks/codex-demo-script.md +3 -3
- package/assets/workshop-bundle/content/talks/context-is-king.md +8 -8
- package/assets/workshop-bundle/docs/harness-cli-foundation.md +32 -1
- package/assets/workshop-bundle/docs/learner-reference-gallery.md +6 -6
- package/assets/workshop-bundle/docs/learner-resource-kit.md +20 -20
- package/assets/workshop-bundle/docs/locales/en/learner-reference-gallery.md +82 -0
- package/assets/workshop-bundle/docs/locales/en/learner-resource-kit.md +126 -0
- package/assets/workshop-bundle/materials/locales/en/participant-resource-kit.md +72 -0
- package/assets/workshop-bundle/materials/participant-resource-kit.md +5 -5
- package/assets/workshop-bundle/workshop-blueprint/README.md +8 -1
- package/assets/workshop-bundle/workshop-skill/analyze-checklist.md +1 -1
- package/assets/workshop-bundle/workshop-skill/commands.md +7 -7
- package/assets/workshop-bundle/workshop-skill/facilitator.md +16 -6
- package/assets/workshop-bundle/workshop-skill/follow-up-package.md +1 -1
- package/assets/workshop-bundle/workshop-skill/install.md +6 -6
- package/assets/workshop-bundle/workshop-skill/locales/en/commands.md +44 -0
- package/assets/workshop-bundle/workshop-skill/locales/en/follow-up-package.md +35 -0
- package/assets/workshop-bundle/workshop-skill/locales/en/recap.md +22 -0
- package/assets/workshop-bundle/workshop-skill/locales/en/reference.md +80 -0
- package/assets/workshop-bundle/workshop-skill/locales/en/setup.md +84 -0
- package/assets/workshop-bundle/workshop-skill/reference.md +12 -12
- package/assets/workshop-bundle/workshop-skill/setup.md +5 -5
- package/assets/workshop-bundle/workshop-skill/template-agents.md +4 -4
- package/package.json +1 -1
- package/src/workshop-bundle.js +3 -0
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
# Doc Generator
|
|
2
|
+
|
|
3
|
+
## Problem
|
|
4
|
+
|
|
5
|
+
Documentation becomes stale almost immediately. Once maintenance is fully manual, the team starts postponing it, and after a few iterations nobody knows whether the docs still describe reality.
|
|
6
|
+
|
|
7
|
+
Your task is to design a tool that generates baseline technical documentation or a structured project overview from an existing codebase.
|
|
8
|
+
|
|
9
|
+
## User stories
|
|
10
|
+
|
|
11
|
+
- As a developer, I want basic technical documentation from a project without writing everything from scratch.
|
|
12
|
+
- As a reviewer, I want to understand the module structure and main entry points within minutes.
|
|
13
|
+
- As the team after rotation, I want to discover the architecture without long detective work.
|
|
14
|
+
|
|
15
|
+
## Architecture notes
|
|
16
|
+
|
|
17
|
+
- The input can be a local repo, seed directory, or simplified dataset.
|
|
18
|
+
- The output can be Markdown, HTML, or a simple text report.
|
|
19
|
+
- The important part is to explain what the tool can infer reliably and what remains heuristic.
|
|
20
|
+
- Design the structure from the start so another output type can be added later.
|
|
21
|
+
|
|
22
|
+
## Done when
|
|
23
|
+
|
|
24
|
+
- The tool produces at least one readable documentation page or report.
|
|
25
|
+
- It is clear how the tool runs locally and which input it expects.
|
|
26
|
+
- The output separates facts from estimates or heuristics.
|
|
27
|
+
- Another team can add a new output type without chaos in the repo.
|
|
28
|
+
|
|
29
|
+
## First step for the agent
|
|
30
|
+
|
|
31
|
+
First describe which signals you will read from the project, what you can infer from them, and where certainty stops. Only then propose the first implementation.
|
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
# Metrics Dashboard
|
|
2
|
+
|
|
3
|
+
## Problem
|
|
4
|
+
|
|
5
|
+
Teams often have the data, but not the screen that turns it into a readable shared view. Without that, decisions get harder, discussion gets noisier, and everyone walks away with a different interpretation of the numbers.
|
|
6
|
+
|
|
7
|
+
Your task is to design a simple dashboard that turns a handful of metrics into a clear shared view.
|
|
8
|
+
|
|
9
|
+
## User stories
|
|
10
|
+
|
|
11
|
+
- As a team, I want several metrics on one screen so the current state is legible quickly.
|
|
12
|
+
- As a facilitator, I want to change seed data without touching UI logic.
|
|
13
|
+
- As the team after rotation, I want to understand the data structure, components, and screens within minutes.
|
|
14
|
+
|
|
15
|
+
## Architecture notes
|
|
16
|
+
|
|
17
|
+
- Separate seed data from UI from the first commit.
|
|
18
|
+
- Mobile-first is an advantage, but the dashboard must remain readable on a projected screen too.
|
|
19
|
+
- README and monitoring should explain what already works, what is still mock, and what is still missing.
|
|
20
|
+
- Design the structure so adding another metric does not require rewriting the whole screen.
|
|
21
|
+
|
|
22
|
+
## Done when
|
|
23
|
+
|
|
24
|
+
- The dashboard shows at least 3 metrics and one trend or comparison.
|
|
25
|
+
- The repo documents the data sources and a mock fallback.
|
|
26
|
+
- It is clear where a new metric is added and how layout changes are verified.
|
|
27
|
+
- A new team can extend the dashboard without breaking the structure.
|
|
28
|
+
|
|
29
|
+
## First step for the agent
|
|
30
|
+
|
|
31
|
+
Design a dashboard that survives handoff. First describe the data model, components, and `Done When` criteria, and only then start building the UI.
|
|
@@ -0,0 +1,31 @@
|
|
|
1
|
+
# Standup Bot
|
|
2
|
+
|
|
3
|
+
## Problem
|
|
4
|
+
|
|
5
|
+
Daily standups in chat often turn into a long thread with no structure. Blockers disappear in the noise, dependencies between people are hard to see, and after a few hours it is difficult to reconstruct what was actually agreed.
|
|
6
|
+
|
|
7
|
+
Your task is to design a tool that turns standup inputs into an overview people can continue working with.
|
|
8
|
+
|
|
9
|
+
## User stories
|
|
10
|
+
|
|
11
|
+
- As a team lead, I want standup responses collected into one readable summary.
|
|
12
|
+
- As a developer, I want to quickly see blockers, dependencies, and topics that need coordination.
|
|
13
|
+
- As the team after rotation, I want to understand the data flow and integration points without verbal handoff.
|
|
14
|
+
|
|
15
|
+
## Architecture notes
|
|
16
|
+
|
|
17
|
+
- Prefer a clear data model over complicated integration.
|
|
18
|
+
- Mock data is fine if the workflow feels realistic and is documented clearly.
|
|
19
|
+
- Separate ingest, processing, and presentation of the output.
|
|
20
|
+
- Prompts, runbooks, and decisions must live in the repo, not only in the heads of the original team.
|
|
21
|
+
|
|
22
|
+
## Done when
|
|
23
|
+
|
|
24
|
+
- The tool can ingest seed data and produce a readable summary.
|
|
25
|
+
- The output highlights blockers or items that need attention.
|
|
26
|
+
- The repo explains how to connect the solution to a real chat or another input channel.
|
|
27
|
+
- After rotation, another team can continue without further explanation.
|
|
28
|
+
|
|
29
|
+
## First step for the agent
|
|
30
|
+
|
|
31
|
+
Split the work into ingest, summarization, and context for the next team. First write the documentation the new team should open first, and only then propose implementation steps.
|
|
@@ -85,7 +85,7 @@ Dobře:
|
|
|
85
85
|
|
|
86
86
|
- Commands, file names, slash commands a literal terms piš v backticks.
|
|
87
87
|
- Nadpisy a běžný výklad piš normální češtinou.
|
|
88
|
-
- V
|
|
88
|
+
- V textech pro účastníky nezahlcuj čtenáře dlouhými bloky textu.
|
|
89
89
|
|
|
90
90
|
## Audience Calibration
|
|
91
91
|
|
|
@@ -97,10 +97,12 @@ Workshop je pro developery. Nepiš, jako by publikum bylo netechnické.
|
|
|
97
97
|
|
|
98
98
|
## Final Check Before Publishing
|
|
99
99
|
|
|
100
|
-
Před uložením
|
|
100
|
+
Před uložením textu pro účastníky si ověř:
|
|
101
101
|
|
|
102
102
|
1. Zní to jako přirozená čeština?
|
|
103
103
|
2. Pochopí to český developer napoprvé?
|
|
104
104
|
3. Nepřekládám něco, co má zůstat v angličtině?
|
|
105
105
|
4. Nezní to jako marketing nebo korporát?
|
|
106
106
|
5. Je z textu jasné, co má člověk udělat?
|
|
107
|
+
|
|
108
|
+
Pro finální workshop-ready revizi použijte i [`content/czech-editorial-review-checklist.md`](./czech-editorial-review-checklist.md).
|
|
@@ -2,7 +2,7 @@
|
|
|
2
2
|
|
|
3
3
|
## Cíl
|
|
4
4
|
|
|
5
|
-
Jedna příběhová ukázka, ne seznam
|
|
5
|
+
Jedna příběhová ukázka, ne seznam funkcí. Publikum má během 15 minut pochopit, jak vypadá dobrý workflow s agentem.
|
|
6
6
|
|
|
7
7
|
## Příběh
|
|
8
8
|
|
|
@@ -35,9 +35,9 @@ Jedna příběhová ukázka, ne seznam feature. Publikum má během 15 minut poc
|
|
|
35
35
|
## Co explicitně neukazovat
|
|
36
36
|
|
|
37
37
|
- pět různých režimů práce
|
|
38
|
-
- složitou
|
|
38
|
+
- složitou přehlídku funkcí
|
|
39
39
|
- dlouhé čekání na generování
|
|
40
40
|
|
|
41
41
|
## Pointa pro místnost
|
|
42
42
|
|
|
43
|
-
Nejde o to ukázat „kouzelný
|
|
43
|
+
Nejde o to ukázat „kouzelný výsledek“. Jde o to ukázat, jak rychle roste kvalita, když přidáme kontext, plán a review.
|
|
@@ -6,36 +6,36 @@ Tento workshop skill i dashboard vznikly stejným způsobem, jakým dnes budeme
|
|
|
6
6
|
|
|
7
7
|
## Klíčová linka
|
|
8
8
|
|
|
9
|
-
Harness engineering je práce s instrukcemi, kontextem a workflow tak, aby agent dělal správné věci opakovaně a předvídatelně. Team lead přece neříká každých třicet sekund jednomu vývojáři, co má dělat.
|
|
9
|
+
Harness engineering je práce s instrukcemi, kontextem a workflow tak, aby agent dělal správné věci opakovaně a předvídatelně. Team lead přece neříká každých třicet sekund jednomu vývojáři, co má dělat. Staví systém, ve kterém tým funguje. A přesně tohle dnes budeme dělat pro agenty.
|
|
10
10
|
|
|
11
|
-
Moje hlavní
|
|
11
|
+
Moje hlavní věta pro dnešek:
|
|
12
12
|
|
|
13
13
|
> Neučíme se "lépe promptovat". Učíme se postavit repo a workflow, ve kterém agent i cizí tým dokážou bezpečně pokračovat.
|
|
14
14
|
|
|
15
15
|
## Mikro-cvičení
|
|
16
16
|
|
|
17
|
-
Všichni dostanou stejný malý úkol. Jedna varianta bude prompt blob. Druhá varianta bude krátké zadání se 4 prvky a s odkazem na
|
|
17
|
+
Všichni dostanou stejný malý úkol. Jedna varianta bude prompt blob. Druhá varianta bude krátké zadání se 4 prvky a s odkazem na kontext zapsaný v repu. Pak porovnáme výsledky. Nehledáme „nejhezčí prompt“. Hledáme způsob práce, který přenese záměr, omezení a kritéria hotovo i do dalšího kroku.
|
|
18
18
|
|
|
19
19
|
## Hlavní teze
|
|
20
20
|
|
|
21
21
|
- Kontext je páka, ne kosmetika.
|
|
22
22
|
- `AGENTS.md`, skills a runbooky jsou týmová infrastruktura.
|
|
23
23
|
- `AGENTS.md` nemá být encyklopedie. Má to být mapa, která ukáže, kam sáhnout dál.
|
|
24
|
-
- Co není v repu, neexistuje. Slack, ústní dovysvětlení a "to si pamatujeme" se při
|
|
24
|
+
- Co není v repu, neexistuje. Slack, ústní dovysvětlení a "to si pamatujeme" se při návaznosti rozpadají.
|
|
25
25
|
- Testy jsou hranice důvěry. Když agent pracuje samostatněji, musíte mnohem líp ověřovat, že udělal právě to, co jste chtěli.
|
|
26
26
|
- Jednoduché mantinely zrychlují práci. Agentovi pomáhá jasný build/test flow, viditelné hranice a předvídatelná struktura.
|
|
27
27
|
- U UI práce je výchozí pattern: agent exploration, potom repeatable browser test, potom lidské review.
|
|
28
|
-
- „Nech model jezdit v mém běžném přihlášeném browseru“ není
|
|
28
|
+
- „Nech model jezdit v mém běžném přihlášeném browseru“ není výchozí doporučení. Bezpečnější je izolované lokální prostředí a jasné mantinely.
|
|
29
29
|
- RED → GREEN není s agenty pomalejší. Naopak často zlevňuje korekce, protože agent dostane přesnější mantinely.
|
|
30
30
|
- Úklid není bonus po workshopu. Když narazíte na opakující se chaos, je čas ho proměnit v lepší template, check nebo runbook.
|
|
31
|
-
- Odpolední
|
|
31
|
+
- Odpolední návaznost prověří, jestli váš kontext funguje i bez vás.
|
|
32
32
|
|
|
33
33
|
## Co chci, aby si adoptovali
|
|
34
34
|
|
|
35
|
-
- Než začnu generovat
|
|
35
|
+
- Než začnu generovat další funkci, udělám z repa místo, kde se dá orientovat.
|
|
36
36
|
- Když řekneme nějaké pravidlo dvakrát nahlas, patří do repa.
|
|
37
37
|
- Když agent dělá víc, já musím lépe ověřovat.
|
|
38
|
-
- Handoff není závěr dne. Je to
|
|
38
|
+
- Handoff není závěr dne. Je to průběžná podmínka celé práce.
|
|
39
39
|
|
|
40
40
|
## Závěr
|
|
41
41
|
|
|
@@ -63,7 +63,7 @@ Preferred interactive path:
|
|
|
63
63
|
1. facilitator starts `harness auth login`
|
|
64
64
|
2. CLI opens browser or emits a device-code style flow
|
|
65
65
|
3. dashboard auth/backend completes facilitator identity verification
|
|
66
|
-
4. CLI stores the resulting session material in local session storage
|
|
66
|
+
4. CLI stores the resulting platform session material in local session storage
|
|
67
67
|
|
|
68
68
|
The facilitator skill should call the CLI for privileged operations instead of storing raw auth/session material itself.
|
|
69
69
|
|
|
@@ -73,6 +73,37 @@ Control model:
|
|
|
73
73
|
- CLI invokes shared protected dashboard APIs
|
|
74
74
|
- APIs enforce authz, validation, idempotency, and audit logging
|
|
75
75
|
|
|
76
|
+
Current auth layering:
|
|
77
|
+
|
|
78
|
+
- `harness auth login` authenticates the facilitator to the platform
|
|
79
|
+
- workshop commands then target a concrete workshop instance explicitly or through a default-selection rule
|
|
80
|
+
- instance grants are checked when a workshop is targeted, not during login bootstrap
|
|
81
|
+
|
|
82
|
+
Current target-selection order:
|
|
83
|
+
|
|
84
|
+
1. an explicit command or route instance id
|
|
85
|
+
2. the selected control-room instance in the dashboard
|
|
86
|
+
3. the deployment default `HARNESS_WORKSHOP_INSTANCE_ID`
|
|
87
|
+
4. a repository fallback only on surfaces that intentionally present a workspace-level default
|
|
88
|
+
|
|
89
|
+
Current command posture:
|
|
90
|
+
|
|
91
|
+
- explicit target required:
|
|
92
|
+
- `harness workshop update-instance <instance-id>`
|
|
93
|
+
- `harness workshop prepare <instance-id>`
|
|
94
|
+
- `harness workshop remove-instance <instance-id>`
|
|
95
|
+
- explicit target is the created resource:
|
|
96
|
+
- `harness workshop create-instance [<instance-id>]`
|
|
97
|
+
- deployment-default target:
|
|
98
|
+
- `harness workshop status`
|
|
99
|
+
- `harness workshop archive`
|
|
100
|
+
- `harness workshop phase set <phase-id>`
|
|
101
|
+
|
|
102
|
+
Current discoverability path:
|
|
103
|
+
|
|
104
|
+
- facilitators can inspect available workshop instances through the dashboard workspace view
|
|
105
|
+
- the same platform-scoped list lives behind the authenticated instances API surface at `/api/workshop/instances`
|
|
106
|
+
|
|
76
107
|
For the current facilitator lifecycle slice:
|
|
77
108
|
|
|
78
109
|
- `workshop status`, `create-instance`, `update-instance`, and `prepare` are brokered-session commands
|
|
@@ -12,19 +12,19 @@ Jakmile tato stránka začne připomínat „awesome list“, je moc dlouhá.
|
|
|
12
12
|
## Oficiální dokumentace
|
|
13
13
|
|
|
14
14
|
- [OpenAI Codex documentation](https://developers.openai.com/codex)
|
|
15
|
-
Použijte jako hlavní zdroj pro aktuální Codex
|
|
15
|
+
Použijte jako hlavní zdroj pro aktuální Codex workflow, skills, `AGENTS.md`, subagenty a bezpečnostní doporučení.
|
|
16
16
|
|
|
17
17
|
- [OpenAI Codex best practices](https://developers.openai.com/codex/learn/best-practices)
|
|
18
18
|
Nejlepší rychlý vstup pro to, jak Codex používat jako dlouhodobého spolupracovníka: kontext, `AGENTS.md`, review, verification, MCP a automatizace.
|
|
19
19
|
|
|
20
20
|
- [OpenAI Codex skills documentation](https://developers.openai.com/codex/skills)
|
|
21
|
-
Hodí se ve chvíli, kdy chcete z opakovaných promptů udělat znovupoužitelné
|
|
21
|
+
Hodí se ve chvíli, kdy chcete z opakovaných promptů udělat znovupoužitelné skills zapsané přímo v repu.
|
|
22
22
|
|
|
23
23
|
- [OpenAI Codex plugins](https://developers.openai.com/codex/plugins)
|
|
24
|
-
Dobré pro pochopení, kdy v Codexu použít pluginy a marketplace místo samotných
|
|
24
|
+
Dobré pro pochopení, kdy v Codexu použít pluginy a marketplace místo samotných skills zapsaných v repu. Pro workshop je berte jako volitelný Codex akcelerátor, ne jako základní bootstrap.
|
|
25
25
|
|
|
26
26
|
- [OpenAI Codex build plugins](https://developers.openai.com/codex/plugins/build)
|
|
27
|
-
Hodí se maintainerům, když chtějí pochopit marketplace model, repo-local marketplace nebo
|
|
27
|
+
Hodí se maintainerům, když chtějí pochopit marketplace model, repo-local marketplace nebo balení skills, app integrací a MCP serverů do jednoho Codex balíčku.
|
|
28
28
|
|
|
29
29
|
- [OpenAI Codex workflows](https://developers.openai.com/codex/workflows)
|
|
30
30
|
Dobré pro převedení workshopových návyků do reálných projektových workflow.
|
|
@@ -66,12 +66,12 @@ Jakmile tato stránka začne připomínat „awesome list“, je moc dlouhá.
|
|
|
66
66
|
Když se stejný task vrací napříč více sessions nebo repy, je to kandidát na skill.
|
|
67
67
|
|
|
68
68
|
- Pluginy a marketplace berte jako Codex-specific distribuční vrstvu, ne jako definici workshop metody.
|
|
69
|
-
Když něco učíme jako Harness
|
|
69
|
+
Když něco učíme jako výchozí přístup v Harness Labu, mělo by to dávat smysl i mimo Codex. Plugin má smysl tam, kde skutečně přidává Codex integrace nebo pohodlnější distribuci.
|
|
70
70
|
|
|
71
71
|
- Testy, tracer bullets a checklisty berte jako hranici důvěry.
|
|
72
72
|
Čím víc autonomie agent dostane, tím méně stačí „rychle jsem projel diff“.
|
|
73
73
|
|
|
74
|
-
-
|
|
74
|
+
- Příklady pro účastníky držte menší než backstage systémy.
|
|
75
75
|
Dobrý learner artefakt je kopírovatelný a čitelný, ne vyčerpávající.
|
|
76
76
|
|
|
77
77
|
## Pravidlo čerstvosti
|
|
@@ -1,39 +1,39 @@
|
|
|
1
1
|
# Learner Resource Kit
|
|
2
2
|
|
|
3
|
-
Tato stránka definuje
|
|
3
|
+
Tato stránka definuje účastnický kit zdrojů pro Harness Lab.
|
|
4
4
|
|
|
5
5
|
Je to malá, kopírovatelná část repa, kterou si mají účastníci odnést, projít a znovu použít ve vlastních projektech.
|
|
6
6
|
|
|
7
7
|
## Základní pravidlo
|
|
8
8
|
|
|
9
|
-
Learner kit není celý backstage
|
|
9
|
+
Learner kit není celý backstage workshopu.
|
|
10
10
|
|
|
11
11
|
Má odpovídat na otázky:
|
|
12
12
|
- co si mám z workshopu zkopírovat do vlastního repa?
|
|
13
13
|
- co má agent udělat jako první?
|
|
14
14
|
- jak zařídit, aby práce přežila handoff?
|
|
15
|
-
- co musím ověřit, než
|
|
15
|
+
- co musím ověřit, než výsledku uvěřím?
|
|
16
16
|
|
|
17
|
-
Výchozí
|
|
17
|
+
Výchozí pracovní rytmus, který má learner kit posilovat, je:
|
|
18
18
|
- `workshop` pro orientaci
|
|
19
19
|
- `brainstorm` nebo `plan` před větším řezem
|
|
20
20
|
- `work` proti jednomu ověřitelnému cíli
|
|
21
21
|
- `review` před důvěrou
|
|
22
|
-
- `compound` a
|
|
22
|
+
- `compound` a průběžný úklid pro věci, které mají přežít session
|
|
23
23
|
|
|
24
24
|
## Jádro learner kitu
|
|
25
25
|
|
|
26
26
|
### 1. Workshop skill
|
|
27
27
|
|
|
28
|
-
Hlavní
|
|
28
|
+
Hlavní rozhraní pro účastníky:
|
|
29
29
|
- [`SKILL.md`](../SKILL.md)
|
|
30
30
|
- [`workshop-skill/setup.md`](../workshop-skill/setup.md)
|
|
31
31
|
- [`workshop-skill/reference.md`](../workshop-skill/reference.md)
|
|
32
32
|
- [`workshop-skill/recap.md`](../workshop-skill/recap.md)
|
|
33
33
|
|
|
34
34
|
Proč sem patří:
|
|
35
|
-
- ukazuje, jak může
|
|
36
|
-
- je to reálné repo
|
|
35
|
+
- ukazuje, jak může skill pro účastníky vést setup, workflow i ověřování v češtině
|
|
36
|
+
- je to reálné rozhraní opřené o repo, ne workshopový slide artefakt
|
|
37
37
|
|
|
38
38
|
### 2. `AGENTS.md` příklad
|
|
39
39
|
|
|
@@ -44,7 +44,7 @@ Proč sem patří:
|
|
|
44
44
|
- je to nejmenší znovupoužitelný příklad trvalého kontextu v repu
|
|
45
45
|
- účastníci ho mohou upravit přímo pro vlastní projekt
|
|
46
46
|
|
|
47
|
-
### 3. Příklad pro
|
|
47
|
+
### 3. Příklad pro ověření a review
|
|
48
48
|
|
|
49
49
|
Výchozí checklist:
|
|
50
50
|
- [`workshop-skill/analyze-checklist.md`](../workshop-skill/analyze-checklist.md)
|
|
@@ -80,17 +80,17 @@ Použijte:
|
|
|
80
80
|
- [`workshop-skill/setup.md`](../workshop-skill/setup.md)
|
|
81
81
|
- [`workshop-skill/reference.md`](../workshop-skill/reference.md)
|
|
82
82
|
|
|
83
|
-
### Během
|
|
83
|
+
### Během build fáze 1
|
|
84
84
|
|
|
85
85
|
Použijte:
|
|
86
86
|
- [`workshop-skill/template-agents.md`](../workshop-skill/template-agents.md)
|
|
87
|
-
-
|
|
87
|
+
- karty `Context Engineering` v [`content/challenge-cards/deck.md`](../content/challenge-cards/deck.md)
|
|
88
88
|
|
|
89
|
-
### Během
|
|
89
|
+
### Během odpoledního převzetí
|
|
90
90
|
|
|
91
91
|
Použijte:
|
|
92
92
|
- [`workshop-skill/analyze-checklist.md`](../workshop-skill/analyze-checklist.md)
|
|
93
|
-
-
|
|
93
|
+
- karty `Workflow` v [`content/challenge-cards/deck.md`](../content/challenge-cards/deck.md)
|
|
94
94
|
|
|
95
95
|
### Po workshopu
|
|
96
96
|
|
|
@@ -109,18 +109,18 @@ Použijte:
|
|
|
109
109
|
- context
|
|
110
110
|
- constraints
|
|
111
111
|
- done when
|
|
112
|
-
2. Do `context` napište, co má agent číst jako první a které docs jsou
|
|
113
|
-
3. Do `done when` napište konkrétní ověření a další
|
|
112
|
+
2. Do `context` napište, co má agent číst jako první a které docs jsou zdroj pravdy.
|
|
113
|
+
3. Do `done when` napište konkrétní ověření a další bezpečný krok, když práce zůstane rozdělaná.
|
|
114
114
|
4. Přidejte build/test příkazy, které zvládne spustit další tým nebo agent bez ústního dovysvětlení.
|
|
115
115
|
5. Přidejte jeden review nebo handoff checklist.
|
|
116
116
|
6. Přesuňte jedno trvalé pravidlo z chatu do repa.
|
|
117
117
|
7. Před větší implementací použijte plan a po větší změně review nebo check.
|
|
118
|
-
8. Když už používáte externí workflow skill pack, napojte ho až nad tento základ místo toho, abyste jím nahrazovali
|
|
118
|
+
8. Když už používáte externí workflow skill pack, napojte ho až nad tento základ místo toho, abyste jím nahrazovali kontext zapsaný v repu a ověřování.
|
|
119
119
|
|
|
120
120
|
## Co sem nepatří
|
|
121
121
|
|
|
122
|
-
Learner kit nemá obsahovat plný backstage
|
|
123
|
-
- workshop
|
|
122
|
+
Learner kit nemá obsahovat plný backstage provozní detail, například:
|
|
123
|
+
- runbooky pro konkrétní workshop instance
|
|
124
124
|
- privátní runtime architekturu a operace
|
|
125
|
-
-
|
|
126
|
-
- maintainer-level deployment a security postupy, pokud se zrovna neučí explicitně jako
|
|
125
|
+
- monitoring nebo řídicí postupy jen pro facilitátory
|
|
126
|
+
- maintainer-level deployment a security postupy, pokud se zrovna neučí explicitně jako skill pro účastníky
|
|
@@ -0,0 +1,82 @@
|
|
|
1
|
+
# Learner Reference Gallery
|
|
2
|
+
|
|
3
|
+
This is a short list of resources for participants after the workshop.
|
|
4
|
+
|
|
5
|
+
The rule is simple:
|
|
6
|
+
- official documentation first
|
|
7
|
+
- then a few strong public repos
|
|
8
|
+
- and finally a small number of patterns directly relevant to Harness Lab
|
|
9
|
+
|
|
10
|
+
Once this page starts to look like an “awesome list,” it has become too long.
|
|
11
|
+
|
|
12
|
+
## Official documentation
|
|
13
|
+
|
|
14
|
+
- [OpenAI Codex documentation](https://developers.openai.com/codex)
|
|
15
|
+
Use this as the main source for current Codex workflows, skills, `AGENTS.md`, subagents, and security guidance.
|
|
16
|
+
|
|
17
|
+
- [OpenAI Codex best practices](https://developers.openai.com/codex/learn/best-practices)
|
|
18
|
+
The best fast entry point for using Codex as a long-term collaborator: context, `AGENTS.md`, review, verification, MCP, and automation.
|
|
19
|
+
|
|
20
|
+
- [OpenAI Codex skills documentation](https://developers.openai.com/codex/skills)
|
|
21
|
+
Useful once you want to turn repeated prompts into reusable repo-native skills.
|
|
22
|
+
|
|
23
|
+
- [OpenAI Codex plugins](https://developers.openai.com/codex/plugins)
|
|
24
|
+
Good for understanding when to use plugins and marketplace distribution in Codex instead of repo-native skills alone. In this workshop, treat plugins as an optional Codex accelerator, not as the core bootstrap.
|
|
25
|
+
|
|
26
|
+
- [OpenAI Codex build plugins](https://developers.openai.com/codex/plugins/build)
|
|
27
|
+
Useful for maintainers who want to understand the marketplace model, repo-local marketplace patterns, or bundling skills, app integrations, and MCP servers into one Codex package.
|
|
28
|
+
|
|
29
|
+
- [OpenAI Codex workflows](https://developers.openai.com/codex/workflows)
|
|
30
|
+
Good for turning workshop habits into real project workflows.
|
|
31
|
+
|
|
32
|
+
- [OpenAI: Harness engineering: leveraging Codex in an agent-first world](https://openai.com/index/harness-engineering/)
|
|
33
|
+
Useful for understanding why repo knowledge should be the system of record and why plans, review, and garbage collection belong to engineering discipline rather than bonus process.
|
|
34
|
+
|
|
35
|
+
- [Next.js AI Coding Agents](https://nextjs.org/docs/app/guides/ai-agents)
|
|
36
|
+
Important mainly for Next.js projects: it shows why agents should read version-matched framework docs instead of relying on stale model memory.
|
|
37
|
+
|
|
38
|
+
## Public repositories
|
|
39
|
+
|
|
40
|
+
- [openai/codex](https://github.com/openai/codex)
|
|
41
|
+
The official CLI repo and the best anchor for how the tool itself evolves.
|
|
42
|
+
|
|
43
|
+
- [openai/skills](https://github.com/openai/skills)
|
|
44
|
+
The official skills catalog and the best reference for what a Codex-native skill looks like.
|
|
45
|
+
|
|
46
|
+
- [openai/codex-action](https://github.com/openai/codex-action)
|
|
47
|
+
A strong example of narrow, safe automation around Codex in CI.
|
|
48
|
+
|
|
49
|
+
- [vercel-labs/skills](https://github.com/vercel-labs/skills)
|
|
50
|
+
Useful when you want skill packaging that stays portable across more than one coding agent.
|
|
51
|
+
|
|
52
|
+
- [vercel-labs/agent-skills](https://github.com/vercel-labs/agent-skills)
|
|
53
|
+
Good examples of practical, high-quality skills, especially for frontend and React work.
|
|
54
|
+
|
|
55
|
+
## Optional workflow packs
|
|
56
|
+
|
|
57
|
+
- [EveryInc/compound-engineering-plugin](https://github.com/EveryInc/compound-engineering-plugin)
|
|
58
|
+
A strong optional choice for people who want an explicit `brainstorm -> plan -> work -> review -> compound` loop. This repo supports installation for both Codex and pi, but treat it as an accelerator on top of the workshop default, not as required setup.
|
|
59
|
+
|
|
60
|
+
## Practical patterns
|
|
61
|
+
|
|
62
|
+
- Start with repo context before you start repeating the prompt.
|
|
63
|
+
In practice: first add `AGENTS.md`, build/test commands, and a concrete definition of done.
|
|
64
|
+
|
|
65
|
+
- Use skills for repeated workflows, not as one-off chat macros.
|
|
66
|
+
If the same task returns across multiple sessions or repos, it is a good skill candidate.
|
|
67
|
+
|
|
68
|
+
- Treat plugins and marketplace distribution as a Codex-specific delivery layer, not as the definition of the workshop method.
|
|
69
|
+
If something is taught as a Harness Lab default, it should still make sense outside Codex. A plugin makes sense where it genuinely adds Codex integrations or easier distribution.
|
|
70
|
+
|
|
71
|
+
- Treat tests, tracer bullets, and checklists as the trust boundary.
|
|
72
|
+
The more autonomy the agent gets, the less “I skimmed the diff quickly” is enough.
|
|
73
|
+
|
|
74
|
+
- Keep participant-facing examples smaller than backstage systems.
|
|
75
|
+
A good learner artifact is copyable and readable, not exhaustive.
|
|
76
|
+
|
|
77
|
+
## Freshness rule
|
|
78
|
+
|
|
79
|
+
Review this list:
|
|
80
|
+
- before every workshop run
|
|
81
|
+
- after a major change in Codex capabilities
|
|
82
|
+
- when a recommended repo becomes stale, noisy, or stops being the best example
|
|
@@ -0,0 +1,126 @@
|
|
|
1
|
+
# Learner Resource Kit
|
|
2
|
+
|
|
3
|
+
This page defines the participant-facing resource kit for Harness Lab.
|
|
4
|
+
|
|
5
|
+
It is the small, copyable slice of the repo that participants should take away, revisit, and reuse in their own projects.
|
|
6
|
+
|
|
7
|
+
## Core rule
|
|
8
|
+
|
|
9
|
+
The learner kit is not the full backstage harness.
|
|
10
|
+
|
|
11
|
+
It should answer:
|
|
12
|
+
- what should I copy from the workshop into my own repo?
|
|
13
|
+
- what should the agent do first?
|
|
14
|
+
- how do I make the work survive handoff?
|
|
15
|
+
- what must I verify before I trust the output?
|
|
16
|
+
|
|
17
|
+
The default participant pattern this kit should reinforce is:
|
|
18
|
+
- `workshop` for orientation
|
|
19
|
+
- `brainstorm` or `plan` before a larger cut
|
|
20
|
+
- `work` against one verifiable goal
|
|
21
|
+
- `review` before trust
|
|
22
|
+
- `compound` and cleanup for things that should survive the session
|
|
23
|
+
|
|
24
|
+
## Core learner kit
|
|
25
|
+
|
|
26
|
+
### 1. Workshop skill
|
|
27
|
+
|
|
28
|
+
Primary participant interface:
|
|
29
|
+
- [`SKILL.md`](../../../SKILL.md)
|
|
30
|
+
- [`workshop-skill/locales/en/setup.md`](../../../workshop-skill/locales/en/setup.md)
|
|
31
|
+
- [`workshop-skill/locales/en/reference.md`](../../../workshop-skill/locales/en/reference.md)
|
|
32
|
+
- [`workshop-skill/locales/en/recap.md`](../../../workshop-skill/locales/en/recap.md)
|
|
33
|
+
|
|
34
|
+
Why it belongs here:
|
|
35
|
+
- it shows how a participant-facing skill can guide setup, workflow, and verification in reviewed workshop language
|
|
36
|
+
- it is a real repo-native interface, not a workshop slide artifact
|
|
37
|
+
|
|
38
|
+
### 2. `AGENTS.md` example
|
|
39
|
+
|
|
40
|
+
Default starter:
|
|
41
|
+
- [`workshop-skill/template-agents.md`](../../../workshop-skill/template-agents.md)
|
|
42
|
+
|
|
43
|
+
Why it belongs here:
|
|
44
|
+
- it is the smallest reusable example of durable context in the repo
|
|
45
|
+
- participants can adapt it directly for their own project
|
|
46
|
+
|
|
47
|
+
### 3. Verification / review example
|
|
48
|
+
|
|
49
|
+
Default checklist:
|
|
50
|
+
- [`workshop-skill/analyze-checklist.md`](../../../workshop-skill/analyze-checklist.md)
|
|
51
|
+
|
|
52
|
+
Why it belongs here:
|
|
53
|
+
- it provides a concrete standard for “can another team continue from here?”
|
|
54
|
+
- it turns repo quality into a checklist people can actually run
|
|
55
|
+
|
|
56
|
+
### 4. Challenge cards
|
|
57
|
+
|
|
58
|
+
Recommended subset:
|
|
59
|
+
- [`content/challenge-cards/locales/en/deck.md`](../../../content/challenge-cards/locales/en/deck.md)
|
|
60
|
+
|
|
61
|
+
Use:
|
|
62
|
+
- as small interventions during the workshop
|
|
63
|
+
- as prompts for what to improve later in a real repo
|
|
64
|
+
|
|
65
|
+
### 5. Follow-up package
|
|
66
|
+
|
|
67
|
+
Post-workshop reinforcement:
|
|
68
|
+
- [`workshop-skill/locales/en/follow-up-package.md`](../../../workshop-skill/locales/en/follow-up-package.md)
|
|
69
|
+
- [`materials/locales/en/participant-resource-kit.md`](../../../materials/locales/en/participant-resource-kit.md)
|
|
70
|
+
|
|
71
|
+
Why it belongs here:
|
|
72
|
+
- it turns the workshop from a one-day event into a repeatable prompt for behavior change
|
|
73
|
+
- it provides a literal handout you can send or print without further explanation
|
|
74
|
+
|
|
75
|
+
## When to use which artifact
|
|
76
|
+
|
|
77
|
+
### During setup
|
|
78
|
+
|
|
79
|
+
Use:
|
|
80
|
+
- [`workshop-skill/locales/en/setup.md`](../../../workshop-skill/locales/en/setup.md)
|
|
81
|
+
- [`workshop-skill/locales/en/reference.md`](../../../workshop-skill/locales/en/reference.md)
|
|
82
|
+
|
|
83
|
+
### During Build Phase 1
|
|
84
|
+
|
|
85
|
+
Use:
|
|
86
|
+
- [`workshop-skill/template-agents.md`](../../../workshop-skill/template-agents.md)
|
|
87
|
+
- the `Context Engineering` cards in [`content/challenge-cards/locales/en/deck.md`](../../../content/challenge-cards/locales/en/deck.md)
|
|
88
|
+
|
|
89
|
+
### During the continuation shift
|
|
90
|
+
|
|
91
|
+
Use:
|
|
92
|
+
- [`workshop-skill/analyze-checklist.md`](../../../workshop-skill/analyze-checklist.md)
|
|
93
|
+
- the `Workflow` cards in [`content/challenge-cards/locales/en/deck.md`](../../../content/challenge-cards/locales/en/deck.md)
|
|
94
|
+
|
|
95
|
+
### After the workshop
|
|
96
|
+
|
|
97
|
+
Use:
|
|
98
|
+
- [`workshop-skill/locales/en/recap.md`](../../../workshop-skill/locales/en/recap.md)
|
|
99
|
+
- [`workshop-skill/locales/en/follow-up-package.md`](../../../workshop-skill/locales/en/follow-up-package.md)
|
|
100
|
+
- [`materials/locales/en/participant-resource-kit.md`](../../../materials/locales/en/participant-resource-kit.md)
|
|
101
|
+
- [`learner-reference-gallery.md`](learner-reference-gallery.md)
|
|
102
|
+
|
|
103
|
+
## What to carry into a real project
|
|
104
|
+
|
|
105
|
+
Participants should leave with these concrete moves:
|
|
106
|
+
|
|
107
|
+
1. Add `AGENTS.md` with:
|
|
108
|
+
- goal
|
|
109
|
+
- context
|
|
110
|
+
- constraints
|
|
111
|
+
- done when
|
|
112
|
+
2. In `context`, write what the agent should read first and which docs are the source of truth.
|
|
113
|
+
3. In `done when`, write explicit verification and the next safe move if work remains unfinished.
|
|
114
|
+
4. Add build/test commands that another team or agent can run without verbal backfilling.
|
|
115
|
+
5. Add one review or handoff checklist.
|
|
116
|
+
6. Move one durable rule from chat into the repo.
|
|
117
|
+
7. Use plan before a larger implementation and review or another check after a larger change.
|
|
118
|
+
8. If you already use an external workflow skill pack, layer it on top of this foundation instead of replacing repo-native context and verification.
|
|
119
|
+
|
|
120
|
+
## What does not belong here
|
|
121
|
+
|
|
122
|
+
The learner kit should not include the full backstage operating detail, for example:
|
|
123
|
+
- workshop-instance runbooks
|
|
124
|
+
- private runtime architecture and operations
|
|
125
|
+
- facilitator-only monitoring or control procedures
|
|
126
|
+
- maintainer-level deployment and security procedures unless they are being taught explicitly as participant skill content
|