Context Package

Session / task name: [short descriptive title]
Created: [YYYY-MM-DD]
Token estimate: [~X,000 tokens]
Quality score: [X / 10]
Model target: [claude-3-sonnet / gpt-4o / other]

TASK

[What you need the AI to do. Be specific: what to analyse, what output you need, any hard constraints.]

Example:

Analyse the churn data in prod.subscriptions and identify the top 3 factors that distinguish customers who churned within 90 days of signup from those who stayed. Produce a ranked list of factors with supporting evidence. Use only data from 2024-01-01 onwards. Output as a Markdown table followed by a short narrative summary.

BUSINESS CONTEXT

[Company/product context, how key metrics are defined, relevant business rules.]

Key metric definitions:

Churned: [definition — e.g. subscription cancelled or not renewed within 90 days of expiry]
Active: [definition]
MRR: [definition]

Business context: [2–3 sentences about the product, customer base, and anything relevant to interpreting the data.]

DATA SCHEMA

[Tables and columns relevant to the task. Include only what's needed.]

-- prod.subscriptions
CREATE TABLE prod.subscriptions (
    subscription_id   VARCHAR,
    customer_id       VARCHAR,
    plan_tier         VARCHAR,   -- free, starter, pro, enterprise
    status            VARCHAR,   -- active, cancelled, expired
    created_at        TIMESTAMP,
    cancelled_at      TIMESTAMP, -- null if not cancelled
    mrr               DECIMAL
);

-- prod.customers
CREATE TABLE prod.customers (
    customer_id       VARCHAR,
    industry          VARCHAR,
    company_size      VARCHAR,   -- SMB, Mid-Market, Enterprise
    signup_source     VARCHAR,
    created_at        TIMESTAMP
);

PRIOR FINDINGS

[Results from previous analyses relevant to this task. Summarise — don't paste full reports.]

[e.g. Last quarter's churn analysis found SMB had 2× the churn rate of Enterprise — focus there]
[e.g. "Onboarding completion" was identified as a leading indicator of 90-day retention]

CONSTRAINTS

[e.g. Only use data from 2024-01-01 onwards — earlier data has schema inconsistencies]
[e.g. Exclude free-tier accounts — they are not revenue-bearing]
[e.g. Do not make causal claims — this is observational data]

OUTPUT FORMAT

[How you want the response structured.]

Format: [Markdown table + narrative / SQL query / bullet points / slide outline]
Length: [Concise — under 500 words / detailed — as needed]
Terminology: [Use "customer" not "user"; use "$" not "USD"]

Bundle built with scripts/context_bundler.py. Token count verified with scripts/token_counter.py.

/context-packager

assetscontext_package_template.md