README Prompt Template
Overview, Installation, Usage, Examples, Configuration — the README skeleton with required, runnable code examples.
The contract that stops AI documents from restructuring themselves: a pinned section skeleton, forced tables, and strict consistency rules.
Ask a model for "a comparison document" five times and you get five structures. The markdown output contract fixes that: a "# " title rule, named "## " sections in a pinned order, table rules that refuse prose-collapse, and a strict consistency block — every section used, exact heading text, nothing merged. This resource loads the structure-enforcement showcase: a feature comparison whose Comparison Table section MUST come back as a real markdown table, identical columns every run.
Generate once per document type
The prompt is the template — reuse it every time that document gets written.
Watch the table contract
Header row, separator row, one row per item, same columns, "—" for blanks — the rules that keep a table a table.
Keep Strict for parseable docs
Exact heading text is what lets scripts, diffs, and reviewers rely on the structure.
The TABLE RULES block forces a real table wherever content compares items. It requires a header row, a separator row, one row per item, the same columns every run, and '—' for empty cells, and states you must never collapse a table into prose. Generate this prompt once, paste it into your assistant, and the Comparison Table section comes back as an actual markdown table.
The DOCUMENT STRUCTURE pins four sections in order — ## Overview, ## Comparison Table, ## Strengths & Trade-offs, ## Recommendation — and the CONSISTENCY RULES demand the exact heading text with no rewording, no merging, and no omitting. That fixed skeleton is what lets a downstream diff, script, or reviewer rely on the structure across runs.
Use this for human-readable markdown documents; for a strict data contract reach for the JSON Output Prompt Builder instead. This template locks prose sections and tables with OUTPUT RULES like 'Do not wrap the document in code fences' and a MARKDOWN RULE of 'No HTML tags — pure markdown', which is document structure, not a typed data schema a program consumes directly.
Overview, Installation, Usage, Examples, Configuration — the README skeleton with required, runnable code examples.
Overview, Authentication, Endpoints, Error Handling, Rate Limits — endpoint docs in an identical structure, with parameter tables and runnable examples forced.
Added, Changed, Fixed, Removed, Security — changelog entries written as user-visible changes, not commit messages.
Summary, Why It Matters, What Happens Next — the executive summary contract for readers who will never open the source.
Stop getting 'Sure, here is the JSON…' — the output-contract pattern that forces models to return only parseable JSON: schema, example, and a strict rule block.
The JSON won't parse and you can't see why. Deterministic cause-sniffing — trailing commas, single quotes, unclosed brackets — and the repair prompt that fixes it.
Build prompts that produce documents in a fixed structure — headings, sections, and tables.
The full path to a content operation that runs, not a pile of posts — set the editorial strategy, research the topics, build a reusable template, then produce and QA structured pieces on repeat.
The AI nails your format once, then the next output adds a "Sure, here's…" wrapper, renames a field, turns the table into bullets, and drops a section. Here's how to write the format as a contract and catch the drift when it happens anyway.