Implementation Prompt
# Implementation Prompt
*Version:* v0.5
*Date:* 2026-03-24
*Last reviewed:* 2026-03-24
Goal: implement a scoped change without drifting from repo contracts.
Instructions:
- read `AGENTS.md`, `docs/CONTEXT_ENGINEERING.md`, `docs/CODEX_PROMPTING.md`,
`docs/TASK_MANAGEMENT.md`, and only the most relevant files first
- note assumptions before changing behavior
- edit the smallest useful set of files
- retrieve additional docs or code only when the current change needs them
- update docs, prompts, evals, and tests together when behavior changes
- update `work/` with status, verification state, blockers, and next action before
ending non-trivial work
- if AI behavior changes, review `docs/MODEL_POLICY.md` and `docs/GUARDRAILS.md`
- run the required verification and report anything you could not verifywhen to use it
Community prompt sourced from the open-source GitHub repo toniacprado/aidev-repo-template (MIT). A "Implementation Prompt" style prompt — adapt the placeholders and specifics to your task. Imported as-is and not independently retested here, so check the output before relying on it.
tags
codingcommunitydeveloper
source
toniacprado/aidev-repo-template · MIT
more in Coding
Coding✓ tested
Senior code review (strict mode)
senior staff engineer running a merciless but fair review
Coding✓ tested
Debug by hypothesis, not by guessing
debugging partner who forms theories before touching code
Coding✓ tested
Generate tests from described behavior
test engineer who writes tests that would actually catch regressions