recsys-pipeline-architect
A spec-and-scaffold skill for building composable recommendation, ranking, and feed pipelines. It encodes the six-stage pattern — Source → Hydrator → Filter → Scorer → Selector → SideEffect — popularized by xAI's open-sourced For You algorithm (Apache 2.0). This skill is an independent reimplementation of the pattern (MIT) — no code copied from the original.
Upstream: https://github.com/mturac/recsys-pipeline-architect
When to Use
- User wants to build any system that picks "the top K items for a user/context"
- User asks "how should I rank X" or describes a feed/personalization problem
- User has a scoring function and needs the pipeline plumbing around it
- User wants to migrate from a single relevance score to multi-action prediction with tunable weights
- User is wrapping an LLM/ML scorer and needs filters, hydrators, side-effects, and a runnable scaffold in their stack (TypeScript / Go / Python)
- Triggers: "recommendation system", "feed algorithm", "ranking pipeline", "for you feed", "candidate pipeline", "content recommender", "pipeline architecture for recsys", "RAG retrieval reranker"
When NOT to Use
- Model architecture work (transformer design, two-tower retrieval, embedding training) — this skill is plumbing around the model, not the model itself
- Pure ML training pipelines — the scoring function is the user's responsibility
- Operating a deployed pipeline (monitoring, autoscaling) — out of scope
The six-stage framework
| # |
Stage |
Job |
Parallel? |
| 1 |
Source |
Fetch candidates from one or more origins |
Yes — multiple sources run in parallel |
| 2 |
Hydrator |
Enrich each candidate with metadata needed for filtering and scoring |
Yes — independent hydrators run in parallel |
| 3 |
Filter |
Drop candidates that should never be shown (blocked, expired, duplicate, ineligible) |
Sequential — each filter sees fewer items |
| 4 |
Scorer |
Assign each surviving candidate one or more scores |
Sequential — later scorers see earlier scores |
| 5 |
Selector |
Sort by final score, return top K |
Single op |
| 6 |
SideEffect |
Cache served IDs, log impressions, emit events, update counters |
Async — must never block the response |
Why this exact order
- Sources before hydration: know what candidates exist before paying to enrich them
- Hydration before filtering: many filters need metadata the source did not provide
- Filtering before scoring: scoring is the expensive stage; drop the ineligible first
- Scorer chain (not single scorer): real systems compose ML scoring + diversity reranking + business rules
- Selector after scoring: keeps scoring deterministic and cacheable
- SideEffects last and async: side effects must never block the user response
Workflow when invoked
Walk the user through these eight steps:
- Clarify the use case (one round, three questions): items being ranked? input context? language/runtime?
- Identify the candidate sources: usually in-network (followed/owned/subscribed) + out-of-network (ML retrieval / trending / similar-to-liked)
- List required hydrations: for each filter and scorer, what data does it need that the source did not provide?
- List the filters: duplicate, self, age, block/mute, previously-served, eligibility. Order matters — cheap before expensive.
- Design the scorer chain: primary (ML) → combiner (multi-action with weights) → diversity → business rules
- Selector: sort descending by final score, take top K (or stratified mix for in-network/out-of-network)
- SideEffects: cache served IDs, emit impression events, update counters, log analytics — all fire-and-forget
- Generate the scaffold in the user's stack
Key trade-offs to surface (don't default silently)
1. Single score vs multi-action prediction
- Single score: train one model to predict relevance. To change behavior → retrain.
- Multi-action: predict
P(action) for many actions (read, like, share, skip, report), combine with weights at serving time. To change behavior → change weights. No retraining.
The X For You system uses multi-action with both positive and negative weights. Recommend multi-action when the user expects to tune frequently.
2. Candidate isolation in scoring
- Isolated: each candidate scored independently. Deterministic, cacheable.
- Joint: candidates attend to each other during scoring (e.g., transformer over batch). More expressive but non-deterministic across batches.
Default to isolation. Joint only when there's a specific reason (e.g., explicit batch-aware diversity).
3. Online vs offline
- Request-time (online): pipeline runs on each request. Latency budget: 100–300ms. Default.
- Pre-computed (offline batch): pipeline runs periodically, results cached. Lower latency, lower freshness.
- Hybrid: candidate retrieval offline, ranking online.
Hard rules
- Do not invent benchmark numbers. "How much faster?" → "depends on workload, run it yourself."
- Attribution discipline. When the pattern is referenced, attribute as "popularized by xAI's open-sourced For You algorithm" /
github.com/xai-org/x-algorithm (Apache 2.0).
- No trademark use. Do not name the user's artifact "X-like" or use "For You" branding. Pattern is free; brand is not. Suggested naming: "candidate pipeline", "feed pipeline", "ranking pipeline", "recsys pipeline".
- Surface trade-offs. Multi-action vs single, isolation vs joint, online vs offline — never default silently.
- The generated scaffold must run. No pseudocode passing as code.
- Filter order matters. Cheap before expensive. Universal before user-specific.
- Side effects never block. Wrap in fire-and-forget patterns (goroutines / promises without await / asyncio tasks).
Anti-Patterns
- Scoring before filtering (wastes compute on candidates that will be dropped anyway)
- Synchronous side effects (cache writes / impression emits blocking the response)
- A single "relevance" score when the product needs to tune for multiple objectives (engagement vs safety vs diversity vs ads)
- Joint scoring as default (non-deterministic, harder to cache, doesn't compose with reranking stages)
- Generating pseudocode "for illustration" — the scaffold must actually run
Upstream contents
The upstream repository at https://github.com/mturac/recsys-pipeline-architect ships:
- Full
SKILL.md with the complete 8-step workflow
- 5 load-on-demand reference docs: interfaces in 4 languages (TS/Go/Python/Rust), multi-action scoring pattern, candidate isolation, filter cookbook (12 patterns), scorer cookbook (weighted sum, MMR, diversity penalty, position debiasing)
- 3 runnable example scaffolds, every one green on its test suite:
- Strapi v5 plugin (TypeScript / Jest — 3/3 pass)
- Zentra-compatible pipeline (Go with generics — 3/3 pass)
- PMAI task prioritizer (Python / FastAPI / pytest — 3/3 pass)
- v0.1.0 release tagged
- MIT license; pattern attributed to xAI X For You algorithm (Apache 2.0)
Install via skills.sh: npx skills add mturac/recsys-pipeline-architect
1---
2name: recsys-pipeline-architect
3description: Design composable recommendation, ranking, and feed pipelines using the six-stage Source→Hydrator→Filter→Scorer→Selector→SideEffect framework popularized by xAI's open-sourced For You algorithm. Use this skill whenever the user is building any system that picks "the top K items for a (user, context)" — social feeds, content CMSs, RAG rerankers, task prioritizers, notification triage, search reranking, ad ranking.
4origin: community
5metadata:
6 source: https://skillmd.com/skills/yanacuti1121/recsys-pipeline-architect
7---
8
9# recsys-pipeline-architect
10
11A spec-and-scaffold skill for building composable recommendation, ranking, and feed pipelines. It encodes the **six-stage pattern** — Source → Hydrator → Filter → Scorer → Selector → SideEffect — popularized by xAI's open-sourced [For You algorithm](https://github.com/xai-org/x-algorithm) (Apache 2.0). This skill is an independent reimplementation of the pattern (MIT) — no code copied from the original.
12
13Upstream: <https://github.com/mturac/recsys-pipeline-architect>
14
15## When to Use
16
17- User wants to build any system that picks "the top K items for a user/context"
18- User asks "how should I rank X" or describes a feed/personalization problem
19- User has a scoring function and needs the pipeline plumbing around it
20- User wants to migrate from a single relevance score to multi-action prediction with tunable weights
21- User is wrapping an LLM/ML scorer and needs filters, hydrators, side-effects, and a runnable scaffold in their stack (TypeScript / Go / Python)
22- Triggers: "recommendation system", "feed algorithm", "ranking pipeline", "for you feed", "candidate pipeline", "content recommender", "pipeline architecture for recsys", "RAG retrieval reranker"
23
24## When NOT to Use
25
26- Model architecture work (transformer design, two-tower retrieval, embedding training) — this skill is plumbing *around* the model, not the model itself
27- Pure ML training pipelines — the scoring function is the user's responsibility
28- Operating a deployed pipeline (monitoring, autoscaling) — out of scope
29
30## The six-stage framework
31
32| # | Stage | Job | Parallel? |
33|---|---|---|---|
34| 1 | **Source** | Fetch candidates from one or more origins | Yes — multiple sources run in parallel |
35| 2 | **Hydrator** | Enrich each candidate with metadata needed for filtering and scoring | Yes — independent hydrators run in parallel |
36| 3 | **Filter** | Drop candidates that should never be shown (blocked, expired, duplicate, ineligible) | Sequential — each filter sees fewer items |
37| 4 | **Scorer** | Assign each surviving candidate one or more scores | Sequential — later scorers see earlier scores |
38| 5 | **Selector** | Sort by final score, return top K | Single op |
39| 6 | **SideEffect** | Cache served IDs, log impressions, emit events, update counters | Async — must never block the response |
40
41### Why this exact order
42
43- Sources before hydration: know what candidates exist before paying to enrich them
44- Hydration before filtering: many filters need metadata the source did not provide
45- Filtering before scoring: scoring is the expensive stage; drop the ineligible first
46- Scorer chain (not single scorer): real systems compose ML scoring + diversity reranking + business rules
47- Selector after scoring: keeps scoring deterministic and cacheable
48- SideEffects last and async: side effects must never block the user response
49
50## Workflow when invoked
51
52Walk the user through these eight steps:
53
541. **Clarify the use case** (one round, three questions): items being ranked? input context? language/runtime?
552. **Identify the candidate sources**: usually in-network (followed/owned/subscribed) + out-of-network (ML retrieval / trending / similar-to-liked)
563. **List required hydrations**: for each filter and scorer, what data does it need that the source did not provide?
574. **List the filters**: duplicate, self, age, block/mute, previously-served, eligibility. Order matters — cheap before expensive.
585. **Design the scorer chain**: primary (ML) → combiner (multi-action with weights) → diversity → business rules
596. **Selector**: sort descending by final score, take top K (or stratified mix for in-network/out-of-network)
607. **SideEffects**: cache served IDs, emit impression events, update counters, log analytics — all fire-and-forget
618. **Generate the scaffold** in the user's stack
62
63## Key trade-offs to surface (don't default silently)
64
65### 1. Single score vs multi-action prediction
66
67- **Single score**: train one model to predict relevance. To change behavior → retrain.
68- **Multi-action**: predict `P(action)` for many actions (read, like, share, skip, report), combine with weights at serving time. To change behavior → change weights. No retraining.
69
70The X For You system uses multi-action with both positive and negative weights. Recommend multi-action when the user expects to tune frequently.
71
72### 2. Candidate isolation in scoring
73
74- **Isolated**: each candidate scored independently. Deterministic, cacheable.
75- **Joint**: candidates attend to each other during scoring (e.g., transformer over batch). More expressive but non-deterministic across batches.
76
77Default to isolation. Joint only when there's a specific reason (e.g., explicit batch-aware diversity).
78
79### 3. Online vs offline
80
81- **Request-time (online)**: pipeline runs on each request. Latency budget: 100–300ms. Default.
82- **Pre-computed (offline batch)**: pipeline runs periodically, results cached. Lower latency, lower freshness.
83- **Hybrid**: candidate retrieval offline, ranking online.
84
85## Hard rules
86
871. **Do not invent benchmark numbers.** "How much faster?" → "depends on workload, run it yourself."
882. **Attribution discipline.** When the pattern is referenced, attribute as "popularized by xAI's open-sourced For You algorithm" / `github.com/xai-org/x-algorithm` (Apache 2.0).
893. **No trademark use.** Do not name the user's artifact "X-like" or use "For You" branding. Pattern is free; brand is not. Suggested naming: "candidate pipeline", "feed pipeline", "ranking pipeline", "recsys pipeline".
904. **Surface trade-offs.** Multi-action vs single, isolation vs joint, online vs offline — never default silently.
915. **The generated scaffold must run.** No pseudocode passing as code.
926. **Filter order matters.** Cheap before expensive. Universal before user-specific.
937. **Side effects never block.** Wrap in fire-and-forget patterns (goroutines / promises without await / asyncio tasks).
94
95## Anti-Patterns
96
97- Scoring before filtering (wastes compute on candidates that will be dropped anyway)
98- Synchronous side effects (cache writes / impression emits blocking the response)
99- A single "relevance" score when the product needs to tune for multiple objectives (engagement vs safety vs diversity vs ads)
100- Joint scoring as default (non-deterministic, harder to cache, doesn't compose with reranking stages)
101- Generating pseudocode "for illustration" — the scaffold must actually run
102
103## Upstream contents
104
105The upstream repository at <https://github.com/mturac/recsys-pipeline-architect> ships:
106
107- Full `SKILL.md` with the complete 8-step workflow
108- 5 load-on-demand reference docs: interfaces in 4 languages (TS/Go/Python/Rust), multi-action scoring pattern, candidate isolation, filter cookbook (12 patterns), scorer cookbook (weighted sum, MMR, diversity penalty, position debiasing)
109- 3 runnable example scaffolds, every one green on its test suite:
110 - Strapi v5 plugin (TypeScript / Jest — 3/3 pass)
111 - Zentra-compatible pipeline (Go with generics — 3/3 pass)
112 - PMAI task prioritizer (Python / FastAPI / pytest — 3/3 pass)
113- v0.1.0 release tagged
114- MIT license; pattern attributed to xAI X For You algorithm (Apache 2.0)
115
116Install via skills.sh: `npx skills add mturac/recsys-pipeline-architect`