primary-logic

Original：🇺🇸 English

Translated

Real-time investment context from Primary Logic — LLM-ranked relevance and impact signals from podcasts, articles, X/Twitter, Kalshi, Polymarket, earnings calls, filings, and other monitored sources across public and private companies.

2installs

Sourceprimarylogic/agent-skills

Added on2026-02-27

NPX Install

npx skill4agent add primarylogic/agent-skills primary-logic

SKILL.md Content

View Translation Comparison →

Primary Logic Investment Intelligence

Use the Primary Logic MCP tools to retrieve real-time, continuously refreshed investment context with LLM-ranked relevance and impact signals for public and private companies.

Setup

Claude Code / Cowork (recommended)

bash

claude plugins add PrimaryLogic/agent-skills

Codex

bash

codex mcp add primary-logic --url https://primarylogic--pulse-backend-external-api-app.modal.run/mcp

Other MCP-capable agents

Add to your agent's MCP config:

URL:

https://primarylogic--pulse-backend-external-api-app.modal.run/mcp

Auth: OAuth (automatic) or Bearer token with API key

npx (universal)

bash

npx skills add PrimaryLogic/agent-skills

Activation Cues

Activate this skill when the user asks for any of:

ticker-specific bullish or bearish evidence
recent catalysts or risk signals from content
per-content relevance or impact details by ticker
source coverage or content visibility checks
API key usage diagnostics
setup help for agentic decision support or user-controlled trading workflows

What This Data Represents

Source-normalized investment context: top podcasts, articles/news, X/Twitter, Kalshi, Polymarket, earnings calls, filings, and other monitored channels normalized into one feed
LLM-heavy signal extraction: per-ticker relevance and impact scores attached to each content item to prioritize material developments
Public + private company coverage: source visibility and ticker coverage for the requesting organization

Connection

MCP server:
```
primary-logic
```
connector

Base URL:

https://primarylogic--pulse-backend-external-api-app.modal.run/mcp

Auth: Bearer token via
```
PRIMARYLOGIC_API_KEY
```
environment variable

Available MCP Tools

Use the

primary-logic

connector tools:

Tool	Purpose
`health_check`	Validate connectivity and auth
`search_content`	Broad content discovery with filters (tickers, sources, sentiment, time)
`get_content`	Fetch a single content item by ID
`get_content_ticker_signals`	Per-item ticker attribution with relevance and impact
`get_ticker_content`	Signal-ranked content for a specific ticker
`list_sources`	Org source visibility and coverage
`list_tickers`	Search available tickers
`get_ticker_detail`	Ticker summary with optional signal stats
`get_usage`	API usage telemetry

Hard Rules

Never fabricate data; all claims must map to tool responses.
Access is user-entitlement scoped to the API key creator; if calls fail with billing errors, key-owner subscription status is usually the cause.
Data visibility is org-scoped; if records are missing, org source visibility may be the cause.
If a tool call fails, report the error and suggest a concrete next step.
Use absolute timestamps in outputs when the user asks about recent windows.
Do not claim market prices, positions, or execution events unless explicitly present in the data.
Do not present outputs as guaranteed returns or autonomous execution instructions.
Frame outputs as context for decision support and user-approved actions.

Input Contract

Interpret each user request into this query plan:

objective: thesis_support, counter_thesis, catalyst_scan, sentiment_shift, coverage_check
scope:
- tickers: list of uppercase ticker symbols
- time window: since and until in ISO datetime format
- source_types: optional list
signal filters:
- min_relevance: 0..1
- min_abs_impact: 0..10
- sentiment: positive | negative | neutral
- include_reasoning: true when the user asks why
retrieval:
- limit (default 50)
- sort mode: date | abs_impact | relevance

If ticker or time window is missing for an investment query, ask one concise clarification.

Query Defaults

Default content limit: 50 unless user asks otherwise.
Apply ticker filters whenever the user names tickers.
For larger pulls, continue pagination while next_cursor is present.
For signal-heavy tasks, start with min_relevance >= 0.6 and min_abs_impact >= 5.

Data Shape

See response contracts for canonical payload examples.

Decision Workflow

Validate connectivity once per session with
```
health_check
```
.
Use
```
search_content
```
for broad discovery pulls.
Use
```
get_ticker_content
```
for signal-ranked ticker analysis.
Use
```
get_content_ticker_signals
```
for per-item attribution detail.

Use

get_ticker_detail

with

include_signal_stats=true

for summary context.

Use
```
list_sources
```
when visibility or source scope is ambiguous.
Use
```
get_usage
```
for rate-limit troubleshooting.

Output Contract

Return structured investment output with:

key_findings: 3 to 7 concise bullets
thesis_view: one short paragraph
supporting_evidence: list of {content_id, ticker, impact_score, relevance_score}
contrary_evidence: same schema as supporting_evidence
catalysts: list
risks: list
tool_trace:
- tools used
- filters applied
- time window
- pagination coverage

If results are empty, return "no qualifying records" and suggest exactly which filter to relax first.

Billing Troubleshooting

If
```
health_check
```
returns a 402 error, check key-owner subscription entitlement first (user-level access), then confirm the key is active and not revoked.
Use
```
get_usage
```
after a successful health check to verify rate-limit posture for the current key.

References

Use cases
API recipes
Response contracts

primary-logic

NPX Install

Tags

SKILL.md Content

Primary Logic Investment Intelligence

Setup

Claude Code / Cowork (recommended)

Codex

Other MCP-capable agents

npx (universal)

Activation Cues

What This Data Represents

Connection

Available MCP Tools

Hard Rules

Input Contract

Query Defaults

Data Shape

Decision Workflow

Output Contract

Billing Troubleshooting

References