skvm-general
Drive the skvm CLI on behalf of a user to profile models, AOT-compile skills, run skill-assisted tasks, run benchmarks, and manage compiled proposals. Trigger when the user asks to "profile", "aot-compile", "bench", "run a single ad-hoc task with a skill", or asks about skvm proposals. Do NOT trigger for `jit-optimize` or when the user wants to optimize/improve a skill — use the sibling `skvm-jit` skill instead.
SJTU-IPADS · 已有 AI 审核记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add SJTU-IPADS/SkVM --skill skvm-general技能说明
name: skvm-general
description: Drive the skvm CLI on behalf of a user to profile models, AOT-compile skills, run skill-assisted tasks, run benchmarks, and manage compiled proposals. Trigger when the user asks to "profile", "aot-compile", "bench", "run a single ad-hoc task with a skill", or asks about skvm proposals. Do NOT trigger for jit-optimize or when the user wants to optimize/improve a skill — use the sibling skvm-jit skill instead.
SkVM General Usage
You are driving skvm, a CLI that AOT-compiles and runs LLM agent skills across heterogeneous models. Use this skill when the user wants to use skvm — profile a model, AOT-compile a skill, run a task with a skill, run a benchmark, or manage optimization proposals. Do not invent flags — every example below uses the real flag set from the installed skvm binary.
Step 1: Prerequisite self-check
Split the check in two — the binary must always be present, but the API key is only required for commands that call an LLM.
Always required — skvm is on PATH:
skvm --help >/dev/null 2>&1 || { echo "skvm not installed — tell the user to run: curl -fsSL https://skillvm.ai/install.sh | sh"; exit 1; }
Required only before LLM-calling commands — profile (without --list), aot-compile, pipeline, run, bench, jit-optimize. Local filesystem commands (profile --list, proposals list|show|reject, logs, clean-jit) do not need the API key — run them even if the key is unset.
# Before running profile/aot-compile/run/bench/jit-optimize:
test -n "${OPENROUTER_API_KEY:-}" || { echo "OPENROUTER_API_KEY is not set — ask the user for their key"; exit 1; }
If a required prerequisite is missing, stop and tell the user what is missing. Do not install anything yourself.
Step 2: Profile a model
A profile (TCP — Target Capability Profile) records what an LLM can do across 26 primitive capabilities. It is the input for AOT compilation and is cached so subsequent compile calls reuse it.
skvm profile --model=<id> # profile one model
skvm profile --model=<id1>,<id2> --concurrency=4 # profile several in parallel
skvm profile --model=<id> --adapter=opencode # non-default adapter
skvm profile --model=<id> --force # ignore cache, re-run
skvm profile --list # list cached profiles
Notes:
- Default adapter is
bare-agent. Other valid adapters:opencode,openclaw,hermes,jiuwenclaw. - Cache lives at
$SKVM_PROFILES_DIR(default.skvm/profiles/). - Profiling is expensive — confirm with the user before running on several models, and prefer
--concurrencyover sequential runs.
Step 3: AOT-compile a skill
AOT compilation rewrites a skill's SKILL.md (and optionally bundle files) so it fits a specific target model's capability profile. The three-pass AOT compiler runs by default.
skvm aot-compile --skill=<path> --model=<id> # all three passes
skvm aot-compile --skill=<path> --model=<id> --pass=1,2,3 # explicit
skvm aot-compile --skill=<path> --model=<id> --pass=1 # only pass 1 (SCR + gap analysis)
skvm aot-compile --skill=<path> --model=<id> --dry-run # no write
skvm pipeline --skill=<path> --model=<id> # profile-if-needed → aot-compile
Pass semantics:
--pass=1— SCR extraction, gap analysis, capability substitution/compensation--pass=2— dependency manifest + env-binding script generation--pass=3— workflow decomposition + DAG parallelism extraction
Compiled variants land under the proposals tree (proposals/aot-compile/...). Multiple passes can be combined in any subset: --pass=1,3 runs passes 1 and 3, skipping 2.
Step 4: Run a single task with a skill
For ad-hoc debugging of one skill on one task:
skvm run --task=<path/to/task.json> --model=<id> # no skill
skvm run --task=<path> --model=<id> --skill=<path/to/SKILL.md> # with skill
skvm run --task=<path> --model=<id> --adapter=opencode --verbose # explicit adapter + debug
Use this to reproduce a single failing task or validate a skill edit. Do not use it for benchmarking — use skvm bench instead.
Step 5: Bench a skill
Benchmarking runs a skill across many tasks and condition variants. It can get expensive fast — always confirm with the user before running across many models or tasks, and use --concurrency for parallelism.
skvm bench --model=<id> # all conditions, all tasks
skvm bench --model=<id> --conditions=original,aot-compiled # baseline + compiled
skvm bench --model=<id> --conditions=jit-optimized # use latest jit-optimize best round
skvm bench --model=<id> --conditions=jit-boost --jit-runs=5 # 5 warmup runs for solidification
skvm bench --model=<id1>,<id2> --concurrency=4 # multi-model in parallel
skvm bench --model=<id> --tasks=task_01,task_02 --runs-per-task=3 # specific tasks, 3 reps each
skvm bench --model=<id> --async-judge # defer LLM-judge to post-run batch
skvm bench --resume=latest # resume an interrupted session
skvm bench --list-sessions # list past sessions
Valid --conditions strings:
no-skill— run the task with no skill injected (baseline floor)original— the skill as-written (baseline ceiling)aot-compiled— full 3-pass AOT compiled variantaot-compiled-p1,-p2,-p3,-p12,-p13,-p23— single or partial AOT passesjit-optimized— the latest best-round variant fromskvm jit-optimizeproposalsjit-boost— code-solidification runtime optimization
Bench logs land at .skvm/log/bench/<sessionId>/.
Step 6: Manage jit-optimize proposals
Proposals a
阅读完整来源 ↗