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 agentmods add commands/marcosd4h/deepextractruntime/lift-classgit clone --depth 1 https://github.com/marcosd4h/DeepExtractRuntimeWrote 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/commands/marcosd4h/deepextractruntime/lift-class)<a href="https://agentmods.dev/commands/marcosd4h/deepextractruntime/lift-class"><img src="https://agentmods.dev/badge/commands/marcosd4h/deepextractruntime/lift-class.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 | $0.00000 | $0.01651 |
| Opus 5 | $0.00000 | $0.00826 |
| Sonnet 5 | $0.00000 | $0.00330 |
| Haiku 4.5 | $0.00000 | $0.00165 |
Grade A, and why
lift-class 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 5d 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 — 134 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Lift Class
Overview
Batch-lift all methods of a C++ class with shared type context, producing a single cohesive .cpp output file with struct definitions, constants, and all methods in dependency order. Uses assembly as ground truth to ensure 100% functional equivalence.
Primary mechanism: code-lifter subagent. This command delegates to the code-lifter subagent, which maintains shared struct definitions, naming conventions, accumulated constants, and already-lifted code in its context across all methods. The grind loop is a fallback for edge cases (see below).
The text after /lift-class specifies the class name and optionally the module:
/lift-class CSecurityDescriptor-- searches all modules/lift-class appinfo.dll CSecurityDescriptor-- targets specific module/lift-class cmd.exe --list-- list all detected classes first
IMPORTANT: Execution Model
This is an execute-immediately command. Do NOT present anything for user confirmation. Run and write the final lifted code and summary straight to the chat as your response. The user expects to see the completed output.
Execution Context
IMPORTANT: Any inline Python that imports
helpers.*must run withcd <workspace>/.claude(so the.claude/directory is onsys.path), not from the workspace root. Script invocations likepython .claude/skills/.../script.pycan be run from the workspace root because those scripts manage their own path setup.
Workspace Protocol
Apply filesystem handoff for all intermediate artifacts:
- Create
.claude/workspace/<module>_lift_class_<timestamp>/. - For any supporting skill/script invocations, pass:
--workspace-dir <run_dir>--workspace-step <step_name>
- Keep only summaries in context; read full intermediate data from
<run_dir>/<step_name>/results.jsonwhen needed. - For fallback grind-loop runs, use the session-scoped scratchpad
.claude/hooks/scratchpads/{session_id}.mdfor loop state, but store intermediate script outputs in the workspace run directory.
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.
- 5d ago First seen · 134 lines · 0 tokens per session scan A 9b07e563bc55
lift-class is a command published in the GitHub repository marcosd4h/DeepExtractRuntime (19 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 1,651 tokens. 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 commands, from other repositories
cpp-test
Enforce TDD workflow for C++. Write GoogleTest tests first, then implement. Verify coverage with gcov/lcov.
gen-gtests
Generate unit Googletests for a diff from trunk.
verify-claim
验证一条 C++ 断言 —— 编译最小例子 + 查 cppreference,给出成立/不成立的实证(金科玉律固化版).
update_ggml
This command updates the ggml library to the latest version and synchronizes all skill files with the updated API.
cpp-build
Fix C++ build errors, CMake issues, and linker problems incrementally. Invokes the cpp-build-resolver agent for minimal, surgical fixes.
cpp-build
Fix C++ build errors, CMake issues, and linker problems incrementally. Invokes the cpp-build-resolver agent for minimal, surgical fixes.