research-ingestion

Name: research-ingestion
Author: minihellboy

95stars28forks

How to install research-ingestion?

minihellboy/factorminer/research-ingestion

$npx -y skills add minihellboy/factorminer --skill research-ingestion

Installs into the current project.

Summary

Absorb external research reports/papers into structured, retrievable hypothesis cues via FactorMiner's Report-to-Memory Absorption (RMA) service — an OHLCV-eligibility gate, a mechanism-family classifier, and research-path cues that feed factor generation prompts. Use to turn a r

Files · 1

View on GitHub

SKILL.md

1	# Research Ingestion
2
3	This skill runs FactorMiner's Report-to-Memory Absorption (RMA) service — a scoped-down implementation of the RMA layer from XAlpha (arXiv:2607.08332): it screens external research fragments for OHLCV-representability, classifies the survivors into a broad mechanism family, and extracts reusable research-path hypothesis cues. It does not mine, generate, or backtest factors; it turns raw research text into structured input for `factor-mining`'s generation prompts.
4
5	## Why absorption instead of raw text
6
7	Feeding report text directly into a generation prompt lets ungrounded or OHLCV-infeasible claims (analyst EPS revisions, order-book microstructure, fundamentals) leak into hypothesis generation. RMA gates every fragment first, so only price/volume-representable mechanisms reach the mining loop.
8
9	## The A/B/C pipeline
10
11	\| Layer \| Question \| Output \|
12	\|---\|---\|---\|
13	\| A (eligibility) \| Can this mechanism be observed, inferred, or proxied from daily OHLCV bars alone? \| KEEP/DROP + reason \|
14	\| B (mechanism family) \| Which broad mechanism bucket does it belong to? \| One of `factorminer.architecture.families.MECHANISM_FAMILIES` \|
15	\| C (archetype) \| What's the reusable research cue? \| A `ResearchArchetype` record with `research_paths` \|
16
17	DROPped fragments (fundamentals, analyst estimates, order-book state, news/sentiment, macro releases) are discarded — they are not representable under the daily OHLCV factor contract.
18
19	## Workflow
20
21	### 1. Ingest a research note
22
23	```bash
24	factorminer ingest-research path/to/report_fragment.txt
25	```
26
27	Add `--mock` to run offline with the deterministic mock LLM provider (no API calls) — useful for smoke tests, never as a research result.
28
29	### 2. Read the classification
30
31	The command prints the KEEP/DROP verdict and reason. For a KEPT fragment it also prints the assigned mechanism family, fine-grained family, mechanism role, and research-path cues.
32
33	### 3. Hand off to mining
34
35	The resulting `ResearchArchetype` records are meant to be threaded into `PromptContextBuilder.build(..., research_archetypes=[...])` so `factor-mining`'s generation prompts carry the research-path text alongside memory and family context. Absorption itself never calls the mining loop — invoke `factor-mining` separately once you have archetypes worth exploring.
36
37	## Guardrails
38
39	- A-layer eligibility is a feasibility gate, not a quality signal — a KEPT fragment is not yet a validated hypothesis, only one that could become an OHLCV factor.
40	- Never present a `ResearchArchetype`'s `research_paths` as a factor formula; it is a hypothesis cue for `factor-mining`, not executable code.
41	- `--mock` classification is deterministic keyword heuristics, not real research judgment — never cite mock output as evidence.
42
43	## MCP alternative
44
45	When the FactorMiner MCP server is connected, `ingest_research_note` exposes the same workflow as a tool, returning the KEEP/DROP verdict and (for KEEP) the archetype record directly.

People who liked this also viewed

6 picks

Type

research-ingestion

95stars28forks

View on GitHub

How to install research-ingestion?

minihellboy/factorminer/research-ingestion

$npx -y skills add minihellboy/factorminer --skill research-ingestion

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Research Ingestion
2
3	This skill runs FactorMiner's Report-to-Memory Absorption (RMA) service — a scoped-down implementation of the RMA layer from XAlpha (arXiv:2607.08332): it screens external research fragments for OHLCV-representability, classifies the survivors into a broad mechanism family, and extracts reusable research-path hypothesis cues. It does not mine, generate, or backtest factors; it turns raw research text into structured input for `factor-mining`'s generation prompts.
4
5	## Why absorption instead of raw text
6
7	Feeding report text directly into a generation prompt lets ungrounded or OHLCV-infeasible claims (analyst EPS revisions, order-book microstructure, fundamentals) leak into hypothesis generation. RMA gates every fragment first, so only price/volume-representable mechanisms reach the mining loop.
8
9	## The A/B/C pipeline
10
11	\| Layer \| Question \| Output \|
12	\|---\|---\|---\|
13	\| A (eligibility) \| Can this mechanism be observed, inferred, or proxied from daily OHLCV bars alone? \| KEEP/DROP + reason \|
14	\| B (mechanism family) \| Which broad mechanism bucket does it belong to? \| One of `factorminer.architecture.families.MECHANISM_FAMILIES` \|
15	\| C (archetype) \| What's the reusable research cue? \| A `ResearchArchetype` record with `research_paths` \|
16
17	DROPped fragments (fundamentals, analyst estimates, order-book state, news/sentiment, macro releases) are discarded — they are not representable under the daily OHLCV factor contract.
18
19	## Workflow
20
21	### 1. Ingest a research note
22
23	```bash
24	factorminer ingest-research path/to/report_fragment.txt
25	```
26
27	Add `--mock` to run offline with the deterministic mock LLM provider (no API calls) — useful for smoke tests, never as a research result.
28
29	### 2. Read the classification
30
31	The command prints the KEEP/DROP verdict and reason. For a KEPT fragment it also prints the assigned mechanism family, fine-grained family, mechanism role, and research-path cues.
32
33	### 3. Hand off to mining
34
35	The resulting `ResearchArchetype` records are meant to be threaded into `PromptContextBuilder.build(..., research_archetypes=[...])` so `factor-mining`'s generation prompts carry the research-path text alongside memory and family context. Absorption itself never calls the mining loop — invoke `factor-mining` separately once you have archetypes worth exploring.
36
37	## Guardrails
38
39	- A-layer eligibility is a feasibility gate, not a quality signal — a KEPT fragment is not yet a validated hypothesis, only one that could become an OHLCV factor.
40	- Never present a `ResearchArchetype`'s `research_paths` as a factor formula; it is a hypothesis cue for `factor-mining`, not executable code.
41	- `--mock` classification is deterministic keyword heuristics, not real research judgment — never cite mock output as evidence.
42
43	## MCP alternative
44
45	When the FactorMiner MCP server is connected, `ingest_research_note` exposes the same workflow as a tool, returning the KEEP/DROP verdict and (for KEEP) the archetype record directly.

People who liked this also viewed

6 picks

Type