$npx skillfedfor your agent

hydra

Hydra is a Lead-driven orchestration toolkit that lets you coordinate multiple agents through structured workflows. The Lead makes strategic decisions while workers report semantic outcomes (done, needs_rework, replan) via a typed result contract, enabling parallel task execution with explicit decision points and intervention capabilities.

Hydra orchestrates multi-agent workflows where a Lead agent makes strategic decisions while workers handle operational tasks through a structured result contract.

AI-generated summary based on this skill's SKILL.md

★ 374  29 MITupdated by blueberrycongee

Decision gist · record as of 2026-05-31

Hydra orchestrates multi-agent workflows where a Lead agent makes strategic decisions while workers handle operational tasks through a structured result contract. Hydra is a Lead-driven orchestration toolkit that lets you coordinate multiple agents through structured workflows. The Lead makes strategic decisions while workers report semantic outcomes (done, needs_rework, replan) via a typed result contract, enabling parallel task execution with explicit decision points and intervention capabilities.

manual: git clone https://github.com/blueberrycongee/termcanvas → cp -r termcanvas/skills/skills/hydra ~/.claude/skills/hydra
skills/skills/hydra/SKILL.md · version 4fdd9dc5

Use it when

  • Hydra routes agent outcomes—completed, stuck, or error—and manages retries through decision points.
  • Yes.

Verify before relying

Read SKILL.md below before installing (1 file). Open directory: indexed for reading, not audited.

Same gist for agents: .md · .json

Install

blueberrycongee/termcanvas/hydra · repository language: TypeScript

Open directory. Skills are indexed for reading, not audited. Review a skill's body before installing it.

Frequently asked questions

AI-generated answers based on this skill's SKILL.md and metadata

What is hydra multi-agent orchestration?

Hydra is a Lead-driven orchestration toolkit that coordinates multiple agents through structured workflows. The Lead makes strategic decisions while workers report semantic outcomes (done, needs_rework, replan) via a typed result contract, enabling parallel task execution with explicit decision points and intervention capabilities.

How does hydra dispatch and retry agents?

Hydra routes agent outcomes—completed, stuck, or error—and manages retries through decision points. When agents report needs_rework or error states, the Lead can intervene with feedback and redispatch tasks. Parallel worktrees remain isolated during execution, then merge results once agents complete or reach terminal states.

Can hydra run parallel agent tasks with isolated worktrees?

Yes. Hydra dispatches parallel agent tasks, each with isolated worktrees to prevent interference. Once agents finish or reach terminal outcomes, hydra merges their results according to the Lead's coordination logic, enabling safe concurrent execution and result aggregation.

How do I monitor agent progress and intervene at decision points?

Hydra lets you watch decision points and inspect workflow state, telemetry, and event logs for debugging. You can monitor agent progress in real time and intervene with feedback at decision points, allowing the Lead to redirect stuck or underperforming agents before they consume more resources.

What does the agent result contract schema define?

Hydra's typed result contract defines semantic intent result parsing: agents report outcomes as done, needs_rework, or replan. This contract ensures consistent communication between workers and the Lead, enabling predictable routing and decision-making across your multi-agent workflow.

How does hydra handle workflow state and event logs?

Hydra maintains a workflow ledger that tracks all events, state transitions, and agent telemetry. You can inspect this ledger to debug failures, audit decisions, and understand how the Lead coordinated agents. State management captures each decision point and outcome for full workflow visibility.

SKILL.md

Rendered from the published skill. Quoted content, verbatim.

Hydra Orchestration Toolkit

Hydra is a Lead-driven orchestration toolkit. You (the Lead agent) make strategic decisions; Hydra handles operational management (dispatch, retry, health checks, result collection).

Sub-agents output semantic intent (done/needs_rework/replan), not routing information. Hydra manages the lifecycle; you decide what happens next.

Why this design (vs. other coding-agent products)

  • SWF decider pattern, specialized for LLM deciders. Hydra is the AWS SWF / Cadence / Temporal decider pattern. hydra watch is PollForDecisionTask; the Lead is the decider; lead_terminal_id enforces single-decider semantics.
  • Parallel-first, not bolted on. dispatch + depends_on + worktree + merge are first-class. Other products (Factory.ai's Droid, Amp, Claude Code subagents) treat parallelism as open research; Hydra makes it the default.
  • Typed result contract. Workers publish a schema-validated result.json (outcome: completed | stuck | error,

(truncated - see the full file via the links below)

File tree — 1 file
skills/skills/hydra/SKILL.md

Let your AI agent find skills like this

Example. Real query, live index.

You found this page by searching. An agent finds it by wishing: SkillFed indexes 56,283 agent skills by what they can do, searchable in plain language.

wish › “Orchestrate multiple agents through a lead-driven workflow with decision points”

Give your agent the search over MCP, or paste the wish link into any chat. No install? Search from any chat →

Related skills

qa
by blueberrycongee · blueberrycongee/termcanvas

QA skill uses real browser automation to test websites end-to-end, capturing visual evidence and validating user flows. It audits pages systematically, scores site health across console errors and interactivity, then fixes issues atomically with commit-per-fix discipline.

MITupdated May 2026
★ 374repo stars
investigate
by blueberrycongee · blueberrycongee/termcanvas

Investigate guides you through a disciplined five-phase debugging process: collect evidence, form ranked hypotheses, verify each one methodically, apply a minimal fix only after confirming root cause, then verify resolution. The workflow prevents guess-and-fix cycles by enforcing evidence gathering before hypothesis formation and a three-strike rule that forces re-collection rather than endless speculation.

MITupdated May 2026
★ 374repo stars
code-review
by blueberrycongee · blueberrycongee/termcanvas

Code Review executes a systematic multi-phase examination of pull requests and code diffs, starting with scope and intent identification before conducting a critical pass for logic errors, injection vulnerabilities, and resource leaks. Specialist lenses then apply targeted checks based on change type—tests, data access, auth, or UI—with each finding assigned a confidence level to guide merge decisions.

MITupdated May 2026
★ 374repo stars
Upstash Workflow
by lobehub · lobehub/lobehub

This skill guides you through implementing Upstash Workflow and QStash async patterns in LobeHub. It covers three core architectural layers—entry point validation, cursor-based pagination with fan-out chunking, and single-item task execution—plus dry-run mode to preview work before committing. Use it to orchestrate large-scale batch processing while respecting rate limits and step constraints.

no license declared → metadata onlyupdated Jul 2026
★ 80,894repo stars
google-workspace
by swarmclawai · swarmclawai/swarmclaw

Google Workspace CLI (`gws`) brings Google Workspace API automation to the command line, handling Drive files, Docs, Sheets, Gmail messages, Calendar events, and Chat spaces with structured JSON responses. Use it to read, list, and mutate Workspace resources while confirming IDs and respecting API limits.

MITupdated Jun 2026
★ 628repo stars
upstash-workflow-js
by upstash · upstash/skills

Build reliable serverless workflows with Upstash Workflow SDK for TypeScript/JavaScript. Define workflow endpoints, execute steps sequentially, and trigger runs from your backend using the Workflow client. Includes patterns for retries, webhooks, cross-workflow invocation, and local development.

MITupdated Jul 2026
★ 15repo stars
Tags
multi-agent-coordinationworkflow-orchestrationdecision-point-routingagent-lifecycle-managementparallel-executionresult-contracttelemetry-drivenintervention-pointssemantic-outcomesdecider-pattern