A secondary source is second-hand information: a blog post, a forum answer, a tutorial, a summary, or the model's own recollection from training. It is someone's account of the thing rather than the thing itself. Handy for getting oriented, not safe to build on unchecked.
The trap
Secondary sources feel authoritative and are easy to reach, so it is tempting to stop there. But they go stale, they simplify, and they are sometimes just wrong. The model's own memory is the sneakiest one of all, because it presents a half-remembered API with the same confidence as a fact. That is a short hop from a hallucination: plausible, fluent, and wrong.
Where secondary sources earn their keep:
- Getting the lay of the land on an unfamiliar topic.
- Finding the name of the thing you should then go and verify.
- Sanity-checking your own understanding before you commit.
Always resolve to the original
Use a secondary source to point yourself at the primary source, then confirm. Reading the real docs turns a vague recollection into contextual knowledge the agent can actually trust. Orientation is fine; verification is mandatory.
Related terms
Primary source
A primary source is the authoritative original: the actual code, the real types, the official docs. Point agents at primary sources so they read reality instead of guessing from memory.
Read definition →Hallucination
A hallucination is a confident, plausible-sounding output that is simply wrong: an invented API, a fabricated file path, a made-up citation. It is not the model lying. It is the model doing exactly what it always does, predicting plausible text, with no built-in sense of truth.
Read definition →Contextual knowledge
Contextual knowledge is what a model knows because it is in the context right now: the files, docs, and output you gave it. It is current and grounded, and it is the main lever against hallucination.
Read definition →