Skills for Real Engineers. Straight from my .agents directory.
autobahn
Agent skill fromLilMGenius/paperthin
Carve guardrail-adjacent items out of scope with safe alternatives before risk-adjacent work starts, then run the safe remainder at full strength in a fresh subagent that only ever sees the carved prompt, never the risky input. Use when a task includes stealth, scraping, privacy, IP, policy, licensing, security, or other safety-adjacent material that could be silently dropped, over-elaborated, or needlessly diluted. Fires on the impulse, not only the topic: the moment you notice yourself about to hedge, soften, silently skip, or brace for a refusal, carve before you execute.
How to install autobahn
npx skills add LilMGenius/paperthin# or add only this skill to Claude Code: copy the folder
skills/depth/autobahn/
# into
~/.claude/skills/autobahn/ (all projects)
.claude/skills/autobahn/ (one project)
1Carve unsafe scope out of a task, execute the safe remainder in a fresh subagent that sees only the carved prompt, and report a visible descope ledger.23## Goal45Make scope decisions explicit before executing risk-adjacent work. Carve in this session; execute in a fresh subagent that receives only the carved prompt, never the risky original or this session's reasoning. Because the executing agent never reads the risky ask, it has nothing to hedge, dilute, or refuse and can build the safe scope at full strength. Remove the unsafe ask and run a genuinely safe prompt; never slip a flagged ask past a guardrail.67## Workflow891. **FRAME**: Read the task, inputs, and user-stated risk posture. Proceed if the user already authorized descoping. Otherwise, propose a carve with an explicit split and wait for approval before **RUN**. A bright-line item has no safe version and is non-negotiable; a gray-zone alternative trades away scope the user may want and needs their decision. If every item is bright-line, proceed and record it in the ledger. If the user disputes a bright-line call, send only the item's abstract description to a fresh context for re-evaluation, stripped of negotiation and persuasion. Do not re-litigate it in the pressured session. Record the appeal and outcome in the ledger either way.10…
Excerpt shown under the repository's MIT license.
More skills in paperthin
Similar skill repositories
An agentic skills framework & software development methodology that works.
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
Public repository for Agent Skills
Makes your AI agent think like the laziest senior dev in the room. The best code is the code you never wrote.
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
