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 skills add dorukanc/harness-council --skill council-runtimegit clone --depth 1 https://github.com/dorukanc/harness-councilWrote 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/dorukanc/harness-council/council-runtime)<a href="https://agentmods.dev/skills/dorukanc/harness-council/council-runtime"><img src="https://agentmods.dev/badge/skills/dorukanc/harness-council/council-runtime/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/dorukanc/harness-council/council-runtime"><img src="https://agentmods.dev/badge/skills/dorukanc/harness-council/council-runtime.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 | $0.00061 | $0.00814 |
| Opus 5 | $0.00030 | $0.00407 |
| Sonnet 5 | $0.00012 | $0.00163 |
| Haiku 4.5 | $0.00006 | $0.00081 |
Grade A, and why
council-runtime 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 10d 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 — 69 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Harness Council runtime
You are the orchestrator. Council lets you spawn other coding harnesses as visible cmux workers, pass context between jobs, and collect results. You decide the routing — which harness, model, and mode fits each task. The role table below is the user's recommended defaults, not a hard rule.
Routing precedence
- Explicit user instruction in the conversation ("use grok for this").
- Project overrides —
CLAUDE.md/AGENTS.mdguidance and.harness-council.tomlrole config in the project root. - Recommended defaults below.
Recommended defaults
| Role | Harness | Mode | Use for |
|---|---|---|---|
explore |
agy | headless (visible log tab) | Codebase scout, web research, cheap/light work |
code |
grok | interactive (visible TUI) | Implementation, refactors, tests, dev tasks |
review |
grok | headless read-only | Adversarial / second-opinion review |
plan |
claude | interactive | Peer planner / heavy-model reasoning |
Override any of these with harness / model / mode / permission fields on council_spawn (or --harness etc. on the CLI) when the task calls for it.
Prefer MCP tools when configured
If the harness-council MCP server is connected, use:
| Tool | When |
|---|---|
council_spawn |
Start a worker (role: explore | code | review | plan) |
council_wait |
Block until a job finishes (or poll with council_get_job) |
council_get_result |
Read summary + result after completion |
council_list_jobs |
See running / recent jobs |
council_message |
Nudge a running worker |
council_cancel |
Stop a worker |
council_attach |
Focus the worker's cmux tab |
council_setup |
Readiness check |
CLI fallback
council setup
council spawn --role explore "Find all auth entrypoints and summarize"
council spawn --role code --input-job <explore_job_id> "Implement rate limiting based on the exploration summary"
council wait <job_id>
council result <job_id>
council list --status running
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.
- 10d ago First seen · 69 lines · 61 tokens per session scan A 6ed513cbba01
council-runtime is a skill published in the GitHub repository dorukanc/harness-council (2 stars, last pushed 1mo ago), licensed MIT. It adds 61 tokens to every session and 814 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
agent-browser
Use the host-side agent-browser CLI for local browser smoke tests, screenshots, snapshots, and simple UI validation against forwarded localhost URLs.
agent-desktop
Use the built-in Computer sub-agent with agent-desktop for macOS desktop automation. Apply when a task needs application launching, accessibility snapshots, stable element refs, window focusing, semantic clicks/typing, or visual confirmation outside the browser sandbox.
qiaomu-model-cli
Call local Grok CLI, Kimi Code CLI, and Claude Code CLI with the strongest current default models and native tools. Use this qiaomu skill workflow when the user wants to run Grok 4.5 for X/web search, image/video generation, research, or multi-tool agent work; run Kimi K3 (1M) for frontend UI/CSS/React/Vue work; run…
multi-cli-spawn
Fan out one brief to multiple coding CLIs in parallel (agy, grok, claude, codex, kimi, qwen) via spawn.sh with PID-file / process-group kill safety and per-seat .status files (RUNNING → DONE|BLOCKED). Use for multi-LLM council, parallel second opinions, or cross-provider review. Prefer foreground wait; for --no-wait…
agy-cli-agent
Run Google Antigravity CLI (agy) headlessly via agy-exec.sh / agy-task.sh / agy-explore.sh. Use when the user asks for agy, Antigravity, Gemini CLI agent, or to execute a plan with agy. Always allocates a PTY (agy -p is silent without TTY). Prefer -f for long prompts; --no-git and orchestrator-owned commits. -r…
claude-cli-agent
Run Anthropic Claude Code headlessly via claude-exec.sh (claude -p). Use when the user asks for Claude Code, claude CLI, ultracode, Claude workflows, agent teams, or Claude subagents. -r is plan mode (clears skip-permissions); -T/--ultracode enables effort ultracode + multi-agent preamble; -P sets permission-mode.…