aide-inference-control SKILL.md skillA
aide-inference-control from anonymousnomad/aide-sovereign-workbench is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add anonymousnomad/aide-sovereign-workbench/aide-inference-control@git:20260908.e9ac98dPins git:20260908.e9ac98d in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- anonymousnomad/aide-sovereign-workbench
- Path
skills/packs/aide-inference-control/SKILL.md- Versions recorded
- 2
- Latest version
- git:20260908.e9ac98d, committed Sep 8, 2026
- First version recorded
- Aug 26, 2026
- Size
- 7,877 bytes
- Repository stars
- 53
- License
- Apache-2.0
In-app inference control + tuning SOP for AIDE — per-model sampler profiles (temperature/top-k/top-p/min-p/mirostat/repeat-penalty/seed), runtime knobs (ctx/threads/parallel/ngl), start-time CLI injection AND per-request overrides, presets, live tok/s/TTFT surfaces, benchmark-gated changes. Use whenever building the Model Hub tuning UI, touching model start arguments or chat sampler parameters, adding presets, displaying performance metrics, or deciding how a downloaded GGUF gets configured.…
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Inference Control — Every Model, Tuned Inside AIDE Born 2026-08-24 from operator directive: users must be able to "really tweak whatever model they're using right inside AIDE", referencing ToolNeuron's in-app download + benchmark + configuration pattern. ## Research base (verified sources) - **Pascal/GTX1060 backend split (llama.cpp issue #19817, Feb 2026 — our exact card)**: Vulkan is SLOWER at prompt-processing but NOTICEABLY FASTER at token generation vs CUDA on GTX 1060 (granite-1B tg128: 90.63 vs 61.65 t/s = +47%; falcon-7B +11%). Root context: Pascal CUDA k-quants historically weak (old PR #1930 needed LLAMA_CUDA_KQUANTS_ITER tweaks; one 1060 tester got best TG with zero offload). => backend is a PER-MODEL RUNTIME CHOICE in AIDE, never hardcoded: profile.runtime.backend selects binary via AIDE_LLAMA_SERVER_<BACKEND> env; profile.runtime.ngl passes -ngl (999 = full offload); flash_attn -> -fa. - **KV-cache law (discussion #23109)**: mixed K/V types tank both pp and tg hard; f16/f16 best for chat-style tg; quantized KV on old cards requires -fa. GGML_VK_PREFER_HOST_MEMORY=1 adds 10-15% tg on Vulkan. …
Read the whole file at its exact version.
How to install
mdr add anonymousnomad/aide-sovereign-workbench/aide-inference-control@sha256:75dfa7ea4a61b78dThe line at the top of this page pins the label git:20260908.e9ac98d, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_62vk6tj5y53bowqo)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 2 versions of this file, between Aug 26, 2026 and Sep 8, 2026, about one every 14 days.
| version | committed | commit | size | audit | |
|---|---|---|---|---|---|
| git:20260908.e9ac98d latest | 2026-09-08 | e9ac98d | 7,877 B | A | view · diff |
| git:20260826.8da8b03 | 2026-08-26 | 8da8b03 | 7,063 B | A | view |
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (7877 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
anonymousnomad/aide-sovereign-workbench · 53 stars · license Apache-2.0 · pushed 2026-10-26 · branch covert-production
API
GET https://markdownregistry.com/api/v1/artifacts/art_62vk6tj5y53bowqo GET https://markdownregistry.com/api/v1/resolve?ref=anonymousnomad/aide-sovereign-workbench/aide-inference-control GET https://markdownregistry.com/api/v1/blob/75dfa7ea4a61b78d0b7f87386f721d36748f42d0f4e9fb86c3c6188be2f067d8
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.