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 skills/shafthq/shaft_engine/local-coding-delegatenpx skills add ShaftHQ/SHAFT_ENGINE --skill local-coding-delegategit clone --depth 1 https://github.com/ShaftHQ/SHAFT_ENGINEWhat 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 | $0.00039 | $0.00482 |
| Opus 5 | $0.00019 | $0.00241 |
| Sonnet 5 | $0.00008 | $0.00096 |
| Haiku 4.5 | $0.00004 | $0.00048 |
Grade A, and why
local-coding-delegate 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 3d 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.
What it actually says
Local coding delegate
Optional. Not always-loaded. A user-configurable local coding loop that runs as a mechanical or default delegate only.
The most-intelligent or default capability model stays the decider. Load this skill when that decider judges a local loop would add value, then probe the host before configuring anything.
When it may run
- The adopter asked for a local loop, or the decider chose one for cheap, bounded, already-specified work.
- The hardware probe recommends a size class
other than
refuse. - The task is mechanical or default-capability labor, not review, not the GitHub playbook, and not a public-API change unless both the probe and the task say the local coder is enough.
Honest refuse is a valid completion. Do not download or configure a local
coder unless the adopter asked and the probe says the host can run a useful
one. Never invent a default vendor model name. Speak in capability levels and
size classes (small, medium, large, or refuse).
Probe
Run the stdlib probe. It reports OS, RAM, and GPU memory when present, then recommends a size class or refuses.
python3 chaos-engine/skills/local-coding-delegate/scripts/probe_hardware.py
refuse— do not install or invoke a local coder.small/medium/large— the host can run a useful coder in that class. The adopter chooses the vendor and the model. This skill does not.
Bounds
The local loop never replaces independent review, never owns merge or GitHub delivery, and never decides architecture. The host session may hand one mechanical batch to a writer. Close that writer after its PR exists. If the probe refuses, or the task outgrows the recommended class, keep the work on the decider model.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 3d ago First seen · 54 lines · 39 tokens per session scan A 9bb5740c1927
local-coding-delegate is a skill published in the GitHub repository ShaftHQ/SHAFT_ENGINE (407 stars, last pushed yesterday), licensed MIT. It adds 39 tokens to every session and 482 once invoked, about $0.0002 per session on Opus 5. 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 skills, from other repositories
debug-optimize-lcp
Guides debugging and optimizing Largest Contentful Paint (LCP) using Chrome DevTools MCP tools. Use this skill whenever the user asks about LCP performance, slow page loads, Core Web Vitals optimization, or wants to understand why their page's main content takes too long to appear. Also use when the user mentions…
xcodebuildmcp-structured-output-review
Use when reviewing XcodeBuildMCP structured output schema changes, schema versioning, manifest outputSchema metadata, and JSON fixture compatibility.
opik-diagnose
Surface the Opik traces worth a developer's attention, ranked by signal — errors, failed tool calls, latency, regressions, and low online-eval scores — plus Diagnostics issues. Reads live/production traces via the SDK (searchtraces and agentinsights) and works with no MCP; uses the MCP issue entity when connected.…
create-operator
MUST READ before creating or moving any operator (createop, copyop, setopposition, or executepython builds). Required verification, canonical positioning recipe, and error-checking steps.
cortex-automate
Set up automation — prospective memory triggers, neuro-symbolic rules, and CLAUDE.md sync. Use when the user says 'remind me when', 'trigger when', 'create a rule', 'auto-remember', 'sync to CLAUDE.md', 'push insights', 'set up trigger', 'when I open this file', 'when this keyword appears', or when you want to…
brand-design
Generate brand guidelines, design systems with design tokens (W3C format), creative direction (SCAMPER, Design Thinking, TRIZ), and design briefs with scope and success criteria.