Docs

Short, practical definitions for the terms openPrompting uses — plus the evidence model, freshness bands, commands, and links into the repository docs.

§ 01 — Terminology

The words that matter.

Agentic coding
Using a coding agent — a model paired with tools, repository access, and an instruction file — to perform work inside a codebase.
Model
The AI performing reasoning and generation.
Harness
The coding tool around that model: repository access, instruction files, tools, permissions, and context management.
Setup
One model paired with one harness.
Profile
A user-defined role or name pointing to a setup, such as builder or reviewer.
Task
The type of work being performed — feature, bug, review, refactor, and so on.
Prompting
Supplying goals, context, constraints, and acceptance criteria so an agent can act reliably.

Prompting is not magic words.

Good prompting is generally less about phrasing and more about supplying an agent with useful material:

  • Goals — what outcome is wanted.
  • Context — files, symbols, and prior decisions that matter.
  • Requirements — what must be true when the work is done.
  • Constraints — what must not change.
  • Acceptance criteria — how the outcome will be verified.
  • Validation — the command or test that confirms success.

Evidence model

Each knowledge entry carries structured source metadata and a verification date. Every claim is classified into one of four evidence classes; entries also record last verified and per-source last checked dates.

official

Official

Supported directly by first-party provider or harness documentation.

tested

Tested

Reproduced by maintainers with a procedure recorded in the entry.

community

Community

Useful practice without first-party confirmation.

legacy

Legacy

Historical or migration advice, not treated as the current default.

Freshness bands

The CLI’s doctor command reads last verified as a maintenance signal — a prompt to review an entry, not a correctness verdict:

  • 0–60 days: current.
  • 61–120 days: review suggested.
  • 121+ days: stale.

This site shows the same band on every entry. The catalog currently holds 5 models, 7 harnesses, and 9 tasks.

Commands

openprompting help
Show the command list and common examples. No config required.openprompting help
openprompting setup
Create or update .openprompting/config.yml; --check validates without writing.openprompting setup
openprompting guide
Display resolved model and harness guidance: recommendations, cautions, evidence, sources.openprompting guide --profile reviewer
openprompting new
Render a deterministic prompt skeleton for a task. No model call, no invented facts.openprompting new planning
openprompting doctor
Run read-only local diagnostics. Never executes project scripts.openprompting doctor
openprompting compare
Compare two profiles, setups, models, or harnesses from documented evidence. No winner inferred.openprompting compare builder reviewer

Successful commands exit 0; failures exit nonzero. Only setup writes project state; doctor is read-only.

Repository docs

The reference lives with the code. These links open the current files on GitHub: