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 azixxxxx/Hydra --skill hydra-panelgit clone --depth 1 https://github.com/azixxxxx/HydraWrote 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/azixxxxx/hydra/hydra-panel)<a href="https://agentmods.dev/skills/azixxxxx/hydra/hydra-panel"><img src="https://agentmods.dev/badge/skills/azixxxxx/hydra/hydra-panel/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/azixxxxx/hydra/hydra-panel"><img src="https://agentmods.dev/badge/skills/azixxxxx/hydra/hydra-panel.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 | $0.00034 | $0.01041 |
| Opus 5 | $0.00017 | $0.00521 |
| Sonnet 5 | $0.00007 | $0.00208 |
| Haiku 4.5 | $0.00003 | $0.00104 |
Grade A, and why
hydra-panel 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 8d 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.
Hydra Panel
Principle
Use diversity to expose assumptions, then let Sol verify and decide. Never turn model agreement into proof.
Hard limits
- Use a maximum three Claude sessions.
- Use a maximum one Fable session.
- Use a maximum two debate rounds: one independent proposal round and one critique round.
- Respect the lower runtime concurrency limit returned by
hydra_route. - Use only top-level Claude sessions. Do not create nested teams or automatic fallback work.
- Keep each test or execute session in its own worktree.
Setup
- Call
hydra_preflight. Stop on any error or ambiguous billing boundary. - Call
hydra_budgetwith{"action":"status"}. Build the context capsule below. Callhydra_routewith its exact byte count andindependentJudgmentsRequired: true. UseuserMode: "maximum"only when the user explicitly requested maximum mode. - Follow the staged plan. Start two members immediately only when the task is parallelizable and independent judgments are the explicit product requirement. A third Fable perspective remains conditional on a named contradiction or missing long-horizon assumption.
- Define one output schema for all panel members with
proposal,claims,evidence,assumptions,uncertainty,tests, anddecisionCriteria. Cap each independent proposal at 8,192 UTF-8 bytes; it cites capsule evidence by reference and contains no copied capsule or transcript.
Round 1: independent proposals
Call hydra_start separately for each member. Route-required immediate panel members use budgetIntent.priority: "quality-floor". A conditional third member uses "opportunistic", valuableWork: true, and distinctInformationGain: true only when a named contradiction or missing assumption makes that statement true. Give every member only:
- the original objective and constraints;
- repository root and current Git SHA;
- its distinct role;
- access mode and expected output schema.
Do not reveal another member's identity, prompt, proposal, or intermediate event. Collect terminal results with hydra_status and hydra_result. A failed job does not authorize a replacement beyond the session cap.
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.
- 8d ago First seen · 69 lines · 34 tokens per session scan A a71a0b18784a
hydra-panel is a skill published in the GitHub repository azixxxxx/Hydra (0 stars, last pushed 6d ago), licensed MIT. It adds 34 tokens to every session and 1,041 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 2026-08-31.
Other skills, from other repositories
claudexor
Use the local Claudexor control plane for harness-agnostic coding work across Claude Code, Codex, Cursor, and OpenCode. Use when a task benefits from route-aware harness and account selection, quota-aware account rotation, shared thread context, read-only planning or research, best-of-N execution, or cross-harness…
remarc
Manage Remarc session comments and contextual feedback. Use when the user mentions Remarc sessions, comments, handoff, triage, review, addressing, resolving, status updates, or summaries.
coding-agents-farm
To orchestrate parallel coding-agent farms (Claude, Codex, Copilot, Gemini, etc.) on isolated git worktrees.
app-it
Turn a local project or a hosted web app (including a published Claude Artifact URL) into a macOS Dock-launchable .app. Use when the user wants a clickable Dock app, local app package, icon, App It install, hosted-URL or Artifact wrapper, or repeatable desktop launcher. Defaults to native Swift WebKit, shipped…
using-codegrid
How an AI coding agent should operate inside CodeGrid — the macOS canvas that runs Claude, Codex, Gemini, Cursor, Grok, and shells side by side and lets them collaborate. Use this skill whenever you are running in a CodeGrid pane, or a user asks you to spawn/list/control agents, open a project, drive the workspace, or…
app-it-windows
Create Windows beta desktop launchers for local web projects: WPF + WebView2 .exe, Start Menu shortcut, .ico, warm server, quit cleanup. Use when the user asks for app-it-windows, a Windows app, .exe launcher, Start Menu shortcut, Windows package, or desktop icon. CI-guarded but untested on Windows hardware; never…