ahepi/deepreason/assembly
Claude Code and agent skill by ahepi/deepreason
- community
Checking that the SKILL.md file is still in the repository.
How an LLM agent selects, installs, and glues the machine's modules (gate, driver, checkers, skills) for a NEW task or repository. Use when setting up any workflow from these tools, deciding which pieces a task needs, or wiring a new artifact type into gated production. Enforces minimal installation and the standard glue patterns.
Install
npx vskill@latest install ahepi/deepreason/assemblySKILL.md
Assembly (gluing the machine's pieces)
<!-- PROMPT-CORE-BEGIN --> You are assembling a workflow from independent modules (see MODULES.md, provided as read context). Rules:
- MINIMAL INSTALL: list what the task actually requires before copying
anything. A module not named by the task's acceptance command or skill is not installed. Models (the driver, M2) come LAST and only if unattended generation is actually wanted -- most workflows are complete without them.
- Decide the three glue questions in order:
a. What is "done"? -> a deterministic COMMAND with exit codes. If no checker exists for the artifact type, the FIRST task is building one (single stdlib file + a FORMAT.md grammar, following battery_digest.py as the template) -- never a model judging doneness. b. Who must not collide? -> if more than one actor or session will write, install the gate (M1) and put every artifact in a cone. One actor, one session: the gate is optional; evidence still says commit early. c. Who generates? -> humans or a chat agent: give them the relevant M4 skill prompt-cores directly. Unattended: install the driver, write the stage in treadle.toml with the skill, the checker as acceptance, and any grammar files as context_files (a skill may never demand conformance to a file the model cannot read).
- STANDARD GLUE PATTERNS -- do not invent alternatives:
- acceptance = checker invocation, exit code is the verdict;
- grammars live in the repo (FORMAT.md files) and reach models as
READ-ONLY REFERENCE context, never inlined into skills;
- proposals have no authority: generated artifacts count only after
the checker passes; a model's failure claim is evidence about the model, not the domain;
- every new artifact type gets: a grammar, a checker, a skill, a
stage -- in that order.
- Report the assembly as a table (module -> why installed / why
skipped) before running anything. If the task fits no module and no checker can be defined, say so plainly: this machine only produces what it can deterministically accept. <!-- PROMPT-CORE-END -->
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingtools/treadle/repo-assets/skills/assembly/SKILL.md
Popularity
- 7-day trend score
- 5
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/ahepi/deepreason/assembly)