Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add StChiotis/Library-First-Engineering --skill lfe-buildergit clone --depth 1 https://github.com/StChiotis/Library-First-EngineeringWrote 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/skills/stchiotis/library-first-engineering/lfe-builder)<a href="https://agentmods.dev/skills/stchiotis/library-first-engineering/lfe-builder"><img src="https://agentmods.dev/badge/skills/stchiotis/library-first-engineering/lfe-builder/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/stchiotis/library-first-engineering/lfe-builder"><img src="https://agentmods.dev/badge/skills/stchiotis/library-first-engineering/lfe-builder.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00034 | $0.01723 |
| Opus 5.5 | $0.00014 | $0.00689 |
| Sonnet 5.5 | $0.00007 | $0.00345 |
| Haiku 4.5 | $0.00003 | $0.00172 |
Grade A, and why
lfe-builder scanned grade A with 0 findings 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 1mo 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
LFE Builder
Mission
Execute the approved plan in .plans/active_plan.md into production-ready code. You are the "Maker" who adheres to strict engineering standards and logic sovereignty.
Sub-Pipeline (execute in this order)
/lfe-builder(this skill) → Implement the plan/lfe-tdd→ Red-green-refactor quality pass on what you just built → writes.plans/tdd_report.md
Toolbox
/lfe-tdd: Mandatory Step 2. Run after implementation to validate and refactor.
Hard Rules
- Plan Adherence: Implement only the features defined in the active plan.
- Logic Sovereignty: Core business/domain logic must be centralized in the designated "Engine" or core modules.
- No Magic Numbers: All configurable values must be extracted into constants files.
- Domain/Boundary Separation: Place stable domain logic in core modules (per Logic Sovereignty); keep volatile boundary code (I/O, UI, framework wiring, external integrations) in separate adapter layers. The project's
engineering-standards.mddefines what "core" and "boundary" mean for this project type. - File-Based Input: Read
active_plan.mdas the source of truth, not conversation context.
Workflow
-
Check plan-critique gate (machine-checkable, file-based): Before reading anything else, open
.plans/plan_critique.mdand parse its YAML frontmatter. The gate opens only when one of these is true:verdict: PASS(anybrain_confirmationvalue), orverdict: WARNANDbrain_confirmationis a non-null ISO-8601 timestamp.
Halt and refuse to write to
src/in any other case — including: file absent,verdict: BLOCK,verdict: WARNwithbrain_confirmation: null, or unparseable frontmatter. Infer nothing about Brain confirmation from the conversation; the typed frontmatter field is the only valid signal. If halting, tell the Brain: "Plan-critique gate is closed (reason: <missing file | verdict: BLOCK | WARN not confirmed in file>). Re-run /lfe-plan-critique or updatebrain_confirmationbefore /lfe-builder can proceed." -
Review: Read
.plans/active_plan.mdandengineering-standards.md, then resolve the re-entry mode in this order:- Rework re-entry (highest precedence) — if
.plans/rework_directive.mdexists and itsslice:matchesactive_plan.md'sslice:: the Brain rejected this slice at the finalization gate. Read the## Rework Directiveand re-implement the named defect in the slice's files — run Implement → Refactor → Mark Done normally (the rework path re-implements; unlike the diagnose path below, it does not skip Implement). This branch wins over anydiagnosis_report.mdpresent. The plan-critique gate (Step 1) keys onplan_critique.mdverdict, which holds across rework rounds, so the already-open gate stays open — no re-critique. - Diagnosis retry — else if
.plans/diagnosis_report.mdexists, check itsslice:field againstactive_plan.md'sslice:field:- Match → this is a legitimate retry path after a failed inspection.
/lfe-diagnosehas already applied the fix tosrc/and recorded the Root Cause and Fix Summary. Read those sections so you understand the post-diagnosis state, but leave the slice as/lfe-diagnosefixed it — skip Steps 3–4 (Implement/Refactor) and proceed directly to Step 5 (Mark Done) with a freshbuilder_done.mdwhose## Files Touchedreflects the diagnose-applied fix and whose## Notes for TDDflags the regression-test added by diagnose. - Mismatch → the diagnosis report is stale (e.g., from a previous slice whose cleanup didn't complete). Ignore it and proceed with implementation normally from Step 3. (Hygiene will flag the stale file as orphaned on its next sweep.)
- Match → this is a legitimate retry path after a failed inspection.
- Fresh implementation — else proceed normally from Step 3.
- Rework re-entry (highest precedence) — if
-
Implement: Use vertical slices (one test -> one implementation) to build the feature.
-
Refactor: Clean up the implementation once the tests pass, without changing behavior.
-
Mark Done: Before handing off to TDD, write
.plans/builder_done.mdso the implementation phase has a physical checkpoint for crash recovery (so a session that dies between coding and TDD resumes without re-implementing). Schema below. -
TDD Pass: Run
/lfe-tddfor the red-green-refactor quality pass. -
Automated Testing: Automatically run the full test suite after dev work to verify the new additions and catch regressions.
-
Handoff: Once code is written,
builder_done.mdexists, TDD pass complete, and all automated tests are passing, signal the transition to Inspector.
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.
- 1mo ago First seen · 102 lines · 34 tokens per session scan A 4b82f0c1cbcb
lfe-builder is a skill published in the GitHub repository StChiotis/Library-First-Engineering (46 stars, last pushed 3mo ago), licensed MIT. It adds 34 tokens to every session and 1,723 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
ai-engineering-toolkit
Overview of six AI engineering workflows: prompt evaluation (8-dimension scoring), context budget planning, RAG pipeline design, agent security audit (65-point checklist), eval harness building, and product sense coaching.
promptfoo-provider-setup
Connect Promptfoo to a model, live HTTP API, local Python/JavaScript provider, or app code. Use for request/auth mapping, response parsing, OpenAPI setup, and connection smoke tests. Use promptfoo-evals for broader eval coverage and promptfoo-redteam-setup for attack selection.
avoid-ai-writing
Audit and rewrite content to remove AI writing patterns ("AI-isms"). Use this skill when asked to "remove AI-isms," "clean up AI writing," "edit writing for AI patterns," "audit writing for AI tells," or "make this sound less like AI." Supports a detect-only mode, an edit-in-place mode for files, an optional voice…
prompt-master
Generates optimized prompts for AI tools. Activates only when the user explicitly asks to write, fix, improve, or adapt a prompt for a specific AI tool (LLM, Cursor, Midjourney, image AI, video AI, coding agents, etc.). Does not activate for general conversation, coding tasks, document writing, or other…
voice-preserving-rewriter
Use when the user asks to rewrite, humanize, clean up, or remove AI-isms from text while preserving the writer's voice, facts, intent, structure, register, and protected material.
flux2-lora-training
Plan or review LoRA and edit-training work specifically for FLUX.2 Klein or Qwen-Image-Edit, including paired datasets, trainer-version contracts, and held-out fidelity checks. Do not use for generic Stable Diffusion/DiT training, prompt authoring, or model serving; route those tasks to their specialized skill.