Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/KaydenClark/LLM_Workbenchnpx agentmods add skills/kaydenclark/llm_workbench/directorWrote 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/skills/kaydenclark/llm_workbench/director)<a href="https://agentmods.dev/skills/kaydenclark/llm_workbench/director"><img src="https://agentmods.dev/badge/skills/kaydenclark/llm_workbench/director/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kaydenclark/llm_workbench/director"><img src="https://agentmods.dev/badge/skills/kaydenclark/llm_workbench/director.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00019 | $0.03300 |
| Opus 5.5 | $0.00008 | $0.01320 |
| Sonnet 5.5 | $0.00004 | $0.00660 |
| Haiku 4.5 | $0.00002 | $0.00330 |
Grade A, and why
director 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 12d 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 — 276 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Director
Purpose
Coordinate the whole project across Spec-bound Dispatchers and integrate independently reviewed results.
The Director is a role: its scope is the whole project and its integration
branch, coordinating Spec-bound Dispatchers and cross-Spec work. The owner
remains the human above the Director. Role scope composes with the assigned
stance. Each Dispatcher works under the dispatcher role entry and adopts the
spec-planner stance at flight launch; Spec Planner, Spec Manager, Reviewer
and Auditor are jobs a Dispatcher performs inside its own Spec, not extra
roles, and each Worker carries one assigned Task for one attempt and hands
back to its Dispatcher. A role never grants, removes, or transfers authority:
the owner request, Workbench controls, repository permissions and governing
context establish it first, and occupying the integration branch adds
nothing. Loading this skill never spawns an agent. Changing stance alone
creates no handoff and never makes a prior participant independent.
Method / Posture
Load the project controls and the current integration state first. Recover
accepted decisions, open gates and the assigned Specs from tracked owners:
each assigned SPEC.md resolved through workbench/manifest.json, the
rendered Taskboard, the ADR collection under workbench/docs/adr and the
Wiki router. Never depend on a local chat, private memory or an unmerged
branch for state another agent must continue from. Record each coordination
decision compactly in its existing owner as it is made, not at closeout.
A Spec's records keep their single writer:
route what belongs in a Spec to that Spec's writer, or record it in the
coordination owner the project names, and
never edit another writer's Spec state concurrently. Distinguish a documented
decision, an unmerged candidate and a delivered capability; only a reviewed
merge into the integration branch delivers.
Obligations
- Identify the assigned Specs and their open gates from the tracked owners above, without relying on a local chat or private memory.
- Assign one Spec and its branch to each Dispatcher. Name the single durable writer for every shared artifact (Spec records, projections, controls, routers) and record each cross-Spec dependency and the landing order in a tracked owner on the integration branch, so independent lanes proceed in parallel without colliding.
- Monitor Dispatcher reports and resolve coordination issues inside the project assignment. Record a permission refusal or an unsupported host capability in the owning Spec rather than routing around it or acting on a Worker's behalf. Escalate only a genuine owner choice, phrased as options, a recommendation and its cost, and recorded in the owning Spec; never re-ask a question the owner settled.
- Arrange separate-context review of each immutable assembled candidate and coordinate its merge request into the integration branch. A rebased or re-merged tip is a new candidate and needs a fresh review before it merges. Neither a Dispatcher nor an implementing Worker supplies independent approval of its own candidate, and the Director never approves a candidate it built. While a room's Task-PR exemption holds, a Task may land as its own PR, judged by its merge answers rather than a separate-context review; read the release owner for that route rather than assuming it.
- Keep accepted decisions, progress, branch and candidate references and remaining gates in the tracked owners on the integration branch through reviewed changes. Allocate IDs only when the record that uses them is committed. Owner Human QA and integration-to-main promotion remain owner acts.
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.
- 12d ago First seen · 276 lines · 19 tokens per session scan A 3a113da9d09f
director is a skill published in the GitHub repository KaydenClark/LLM_Workbench (2 stars, last pushed 2d ago), licensed MIT. It adds 19 tokens to every session and 3,300 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-10-07.
Other skills, from other repositories
mcore-split-pr
Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.
shiplog
Recap of everything shipped since the last run - cross-repo PRs, security fixes, star deltas, and X traction, synthesized into a digest article and a ready-to-post shiplog in your voice.
connect
Choose which wiki this session uses, the team project's pages on the Nimbalyst server or the project's local wiki of files, and connect a repository to its team project. Use at the start of a task before any wiki tool, when a wiki tool returns reponotbound, ambiguousproject, pinmismatch, projectnotaccessible…
llxprt-issue-workflow
Use this skill when asked to address, fix, or work on a GitHub issue in the llxprt-code repository. Covers the complete issue lifecycle - branch setup, gh CLI usage, test-first planning, subagent delegation and review loops, the full verification cycle (including the zai-glm-flash smoke test), open code review (ocr)…
cuga-github-issues
Create GitHub issues for cuga-agent (bugs, features, epics, designs, and related work) against origin using gh, with epic → feature → issue hierarchy and GraphQL sub-issue linking. Use when the user says create new issues, report a bug, file a bug, feature request, new feature, create an epic, GitHub issue, sub-issue…
implement-issue
Implement a Linear brief in a named GitHub repo. Use when the brief skill is implement-issue or the user wants a new branch and draft PR.