Gemini Agent
- 1 installs
- Updated June 12, 2026
- ahsan361/crisis-intelligence-response-orchestrator
gemini-agent is a Claude Code skill that guides implementing Python AI pipeline agents using the Google Gemini API with structured JSON output and shared client key rotation.
About
gemini-agent guides implementation of AI pipeline agents that use Google Gemini. Each agent is a Python function in backend/agents/ that accepts a state dict and returns a dict, instantiating the GenAI client through a shared client manager for API-key rotation. It enforces structured application/json output, appends outcomes to state trace with fields like decision and confidence, and requires JSON-parse fallbacks so a failed parse does not crash the pipeline.
- Implements AI pipeline agents in Python using the Gemini GenAI client
- Enforces API-key rotation via a shared client manager
- Requires structured application/json output and trace-state mutation
Gemini Agent by the numbers
- 1 all-time installs (skills.sh)
- Ranked #14,102 of 16,546 AI & Agent Building skills by installs in the Skillselion catalog
- Data as of Jul 7, 2026 (Skillselion catalog sync)
gemini-agent capabilities & compatibility
Requires a Google Gemini API key managed via the client manager.
- Capabilities
- gemini integration · agent implementation · structured output · state tracing
- Use cases
- api development · orchestration
- Pricing
- Bring your own API key
What gemini-agent says it does
Implement AI pipeline agents using Gemini.
Always instantiate the GenAI client using `get_client()` from `agents.client_manager`. This guarantees API key rotation.
If `json.loads(response.text)` fails, the agent must catch the exception and return a default fallback state dict to prevent the pipeline from crashing.
npx skills add https://github.com/ahsan361/crisis-intelligence-response-orchestrator --skill gemini-agentAdd your badge
Show developers this skill is listed on Skillselion. Paste this into your README.
| Installs | 1 |
|---|---|
| Last updated | June 12, 2026 |
| Repository | ahsan361/crisis-intelligence-response-orchestrator ↗ |
What it does
Implement a Python Gemini pipeline agent with structured JSON output, key rotation, and trace-state mutation.
Who is it for?
Adding agents to a FastAPI/Supabase pipeline that uses a client manager and an agent_trace JSONB column.
Skip if: Directly instantiating google.genai.Client with an environment variable - the docs forbid it in favor of ClientManager.
When should I use this skill?
You are adding a new Gemini-backed pipeline agent to a Python backend that follows the client-manager and trace-state conventions.
What you get
A deterministic Python agent function that returns structured JSON, rotates API keys, and updates the trace state safely.
- a Python agent function under backend/agents/ returning structured JSON
By the numbers
- trace entry captures 5 fields (agent, timestamp, decision, confidence, input/output summary)
Files
Gemini Agent Skill
Instructions
1. Agent Definition: Create a new python file in backend/agents/. Define a function accepting a state: dict and returning a dict. 2. Client Management: Always instantiate the GenAI client using get_client() from agents.client_manager. This guarantees API key rotation. 3. Structured Output: Ensure the prompt asks for application/json output, and set response_mime_type="application/json" in GenerateContentConfig. 4. State Mutation: Append the agent's outcome to state["trace"] as a dictionary containing agent, timestamp, decision, confidence, and input_summary/output_summary.
Constraints
- Do not initialize the
google.genai.Clientdirectly with an environment variable. TheClientManagermust handle it. - Agents must be deterministic in output schema. Always provide a clear JSON format block in the prompt.
Common Pitfalls
- Missing Fallbacks: If
json.loads(response.text)fails, the agent must catch the exception and return a default fallback state dict to prevent the pipeline from crashing. - Trace Formatting: Forgetting to update
state["trace"]breaks the mobile app's Agent Trace screen, which relies on theagent_traceJSONB column.
Related skills
Setup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196k
Lark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402k
CavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7k
Lark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16k
Running Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72
Codex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10FAQ
How should the Gemini client be created?
Always via get_client() from agents.client_manager, which guarantees API key rotation; never instantiate google.genai.Client directly with an env variable.
What happens if JSON parsing fails?
The agent must catch the exception and return a default fallback state dict to prevent the pipeline from crashing.