feat: bilingual role SOP packs — 66 new skills, knowledge library, 4 role loops - #21
Merged
Merged
Conversation
…rary, 4 role loops Adds four role-based standard-operating-procedure packs in both editions, taking the library from 72 to 138 skills. VN cluster (v3.5.0) — skills/vi/35-67: - Content system (35-40): brand voice, content brief, organic caption, community seeding, content audit, next content plan - Design production (41-50): campaign asset list, image/carousel/video/design-tool briefs, brand guideline, design review, quick visual brief, HTML email, asset resize - Performance ops (51-57): audience research, account structure, tracking setup, media plan, scaling, retargeting, next ads plan - Leader ops (58-67): positioning, GTM, launch playbook, budget planning, marketing review, retrospective, team brief, performance review, crisis, vendor brief Global cluster (v3.6.0) — skills/en/35-67 *-global: English adaptations, not translations. Frameworks and process logic carry over unchanged; currency rebased to USD against references/benchmarks-global.md, channels swapped to the global stack (Klaviyo/Shopify/Meta US), compliance moved to FTC, GDPR, CCPA and CAN-SPAM. Community seeding was rewritten rather than localized. Three tactics in the VN version do not survive US/EU regulation and were removed: persona accounts (astroturfing), brand-written customer reviews (FTC 16 CFR Part 465 carries civil penalties), and staged threads answered by your own accounts. They are replaced with a named-participant roster, permission-backed customer stories, and disclosed answers in existing threads, at Reddit's 9:1 value-to-promotion norm. Also added: - knowledge/ — 24 foundation and implementation files for role-based AI project context - agents/design-producer.md — sixth agent covering the visual production pipeline - 4 VN workflows: content-engine, performance-loop, design-pipeline, leader-cadence - references/marketing-templates-library.md — structures of 28 operational templates mapped to the skill that regenerates each - references/ai-tool-orchestration.md — multi-tool AI workflow for marketers Changed: - 17 existing skills enriched with SOP-level detail without structural breaks - benchmarks-vietnam.md extended: frequency, TikTok VTR, organic group/IG rates, warn-vs-kill thresholds, target-relative decision rules, CPO/ROAS by industry, B2B - progressive disclosure applied to the 9 skills over the 500-line spec guideline; detail moved into per-skill references/, no content removed - marketplace.json 72 -> 138 skills, version 3.6.0 Fixed: - stale reference paths skills/references/ -> skills/vi/references/ (moved in v3.0.0) - .gitignore now covers output/ (was outputs/, so client deliverables were untracked-visible) Validator: 89 passed, 0 issues. All 66 new skills PASS with no warnings.
minhnv0807
marked this pull request as ready for review
August 7, 2026 04:47
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Four commits. The bulk is the latest one: four role-based standard-operating-procedure packs in both editions, taking the library from 72 to 138 skills, plus a foundation knowledge library and the workflows that tie them together. The earlier three commits add the growth/AI-Marketing-OS skills and README diagram work.
What this adds
33 Vietnamese skills (
skills/vi/35-67)Each skill is a full operating procedure — steps, required inputs, decision rules, quality gates, timelines — not a single prompt. Example:
54-media-plancarries the reverse-math from revenue to max CPL to budget with the 30/50/15/5 split;55-scaling-adshas a 7-condition gate before scaling and a learning-phase reset risk table.33 Global skills (
skills/en/35-67 *-global)Adaptations, not translations. Frameworks and process logic carry over unchanged. What changes: currency rebased to USD against
references/benchmarks-global.md, channels swapped to the global stack (Klaviyo/Mailchimp instead of Zalo OA, Shopify/Amazon instead of Shopee/Lazada, Meta US/EU instead of Facebook VN), and compliance moved to FTC, GDPR, CCPA and CAN-SPAM.Community seeding was rewritten, not localized. Three tactics in the Vietnamese version do not survive US/EU regulation and were removed rather than translated:
Replaced with a named-participant roster (real team members, affiliation visible), permission-backed customer story amplification, and disclosed answers in existing threads, at Reddit's 9:1 value-to-promotion norm.
No invented numbers. Where a Vietnamese threshold had no global equivalent in the benchmark file, it became a relative rule with the gap stated in-file — minimum ad set budget is now a derived formula
(50 conversions × target CPA) / 7rather than a currency conversion, and CTR cells for platforms without a repo median read "use account history" instead of a guess.Also added
knowledge/— 24 foundation and implementation files, sized for upload as context into role-based AI projectsagents/design-producer.md— sixth agent, owns the visual production pipelinecontent-engine,performance-loop,design-pipeline,leader-cadencereferences/marketing-templates-library.md— structures of 28 operational templates, each mapped to the skill that regenerates itreferences/ai-tool-orchestration.md— multi-tool AI workflow for marketersChanged
benchmarks-vietnam.mdextended — frequency thresholds, TikTok VTR, Facebook Group and Instagram organic rates, warn-vs-kill thresholds, target-relative scale/pause/test rules, CPO and ROAS by industry, B2B benchmarksreferences/, nothing removedmarketplace.json72 → 138 skills, version 3.6.0Fixed
skills/references/→skills/vi/references/in 12 places (folder moved in v3.0.0, links were never updated).gitignorecoveredoutputs/but the real folder isoutput/, so client deliverables were sitting untracked-but-visible and onegit add -Aaway from a public repoTest plan
validate-skills.ps1→ 89 passed, 0 issues; all 66 new skills PASS with no warningsrelated:entry in the 66 new skills resolves to an existing skill foldermarketplace.jsonpaths exist and containSKILL.mdinstall.shandinstall.ps1syntax valid; both now copyknowledge/references/benchmarks-global.mdMerge state
GitHub reports the branch as mergeable with no conflicts against
master, including the two commits master has that this branch does not (63cf44d,c30a009).