Borrowing it
Nothing to install: this file belongs to reidemeister94/development-skills. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/reidemeister94/development-skills/main/AGENTS.mdgit clone --depth 1 https://github.com/reidemeister94/development-skillsWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/reidemeister94/development-skills/agents-md)<a href="https://agentmods.dev/instructions/reidemeister94/development-skills/agents-md"><img src="https://agentmods.dev/badge/instructions/reidemeister94/development-skills/agents-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.01377 | $0.01377 |
| Opus 5 | $0.00688 | $0.00688 |
| Sonnet 5 | $0.00275 | $0.00275 |
| Haiku 4.5 | $0.00138 | $0.00138 |
Grade B, and why
development-skills AGENTS.md scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
- Hooks in `hooks/` run natively on Claude Code and on Codex 0.131+ (auto-loaded); Codex 0.128–0.130 needs `[features] plugin_hooks = true` in `~/.codex/config.toml`. Document the manual fallback whenever a hook is added How it starts
The opening of the file, as written. The whole thing — 34 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Development skills
This repo IS the source of the development-skills plugin: a direct/full development loop plus skills, hooks, and a single staff-reviewer subagent, distributed to Claude Code and Codex CLI.
- Don't pander · be critical. Challenge assumptions, push back on bad ideas. No flattery openers. User confirmation validates the decision, not the analysis.
- Think before coding. State assumptions explicitly. Ask when unclear. Don't guess, don't hide confusion.
- Plan before implementing. Explore → plan → lock the HOW (edge cases · data shapes · error semantics · contract boundaries · test scope · rollback) → code.
- Simplicity by default. Minimum code that solves the problem. Three filters before adding anything: existing mechanism covers >50%? · can this be one fewer file / abstraction / config / dependency? · would removing it cause a real failure? A refactor must measurably improve one of: clear · descriptive · efficient · performant · reliable · robust · maintainable.
- Surgical changes. Every changed line traces to the request. No refactoring of adjacent code. No error handling for impossible scenarios. Clean up only your own mess.
- All signal, zero noise. No dead branches, no defensive try/catch on safe paths, no wrapper-for-nothing functions, no unused imports. No filler openers, no trailing summaries when the diff is the answer.
- Comments explain WHY, not WHAT. Non-obvious business logic, hidden constraints, workarounds — yes. Restating what the next line does — no.
- TDD: Red → Green → Refactor. No production code without a failing test first. One test = one cycle. Wrote production code before the test? Delete it. Untestable (UI-heavy / infrastructure / config-only) → closest automated check + documented WHY + manual evidence.
- No claim without fresh evidence. IDENTIFY → RUN → READ → VERIFY → CLAIM. "I'm confident" is not a step. Skipping any step = lying, not verifying.
- Root cause, not symptoms. Fix the underlying error, never suppress it.
# type: ignore, swallowed exceptions, disabled tests,--no-verifyare admissions the bug is winning. - Document every discovery (anything you lacked at the start — non-obvious, domain·infrastructure·company·project-specific). WHY →
docs/chronicles/, HOW →docs/plans/; a critical always-read fact → one line in theAGENTS.mdlist; a topic with depth →.agents/rules/<topic>.md(same convention), indexed fromAGENTS.md. Fewest words. Pay investigation costs once. - Slim docs · English · memory ≈ empty.
AGENTS.md≤ 70 lines: principles → use development-skills → single fewest-words list of the most critical, non-trivial domain·infra·company·project facts → index to.agents/rules/; no section headings. Each rules file: same convention, vertical per topic. English only across all artifacts. Teammates share only the repo — memory is per-machine and invisible to them: project facts live inAGENTS.md/.agents/rules/, never in memory; machine-specific facts → gitignored.claude/CLAUDE.md/~/.codex/AGENTS.md; memory stays ≈ empty. - Communicate to be understood. Explain in the simplest accurate language that preserves all important information — lead with the answer, name assumptions and trade-offs when they matter, simplify wording not substance. No obscure terms, no ambiguity.
- The canonical development loop and its principles live in
shared/development-loop.md; the writing contract inshared/writing.md. Skills reference them by path — never restate their content, in skills, AGENTS.md, README, or chronicles. - Versioning is automated via
make bump-version-{minor,major,patch}(commitizen):cz bumpatomically updates[tool.commitizen] versioninpyproject.toml,VERSION,.claude-plugin/plugin.json,.codex-plugin/plugin.json, and.claude-plugin/marketplace.json, then creates the annotated git tag$version. Never bump version files manually. - Only one named subagent ships:
staff-reviewer. Implementation and verification run in the main thread pershared/development-loop.md— do not introduce extra subagents without explicit design discussion. - Hooks in
hooks/run natively on Claude Code and on Codex 0.131+ (auto-loaded); Codex 0.128–0.130 needs[features] plugin_hooks = truein~/.codex/config.toml. Document the manual fallback whenever a hook is added. pyproject.tomlexists for plugin-internal scripts (lint helpers) — the plugin is markdown-first and not distributed as a PyPI package.- Personal per-machine context:
.claude/CLAUDE.md(Claude) or~/.codex/AGENTS.md(Codex). Both must stay gitignored; never commit either.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 34 lines · 1,377 tokens per session scan B 47285de9a00d
development-skills AGENTS.md is an instructions file published in the GitHub repository reidemeister94/development-skills (11 stars, last pushed 1mo ago), licensed MIT. It adds 1,377 tokens to every session, about $0.0069 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
spec-kit AGENTS.md
AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.