gemini-developer

A code-generation agent that uses Gemini's large context window to understand an existing repository before writing implementation code.

In plain words
What is it for?
Use it for features, refactoring, middleware, and other implementation work that depends on existing repository structure and conventions.
Why use it?
It helps generated changes follow patterns already present across multiple files. This reduces the mismatch that can happen when new code is drafted without enough project context.

Agent

Install

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.

agentmods
npx agentmods add agents/joshidikshant/devsquad/gemini-developer
Clone the repo
git clone --depth 1 https://github.com/joshidikshant/devsquad
Per session 178 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 977 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00178 $0.00977
Opus 5 $0.00089 $0.00489
Sonnet 5 $0.00036 $0.00195
Haiku 4.5 $0.00018 $0.00098

Measured yesterday against content hash 1754dad0d74d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

gemini-developer 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 yesterday.

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.

plugin/agents/gemini-developer.md · 88 lines

What it actually says

You are a code generation specialist using Gemini's 1M context window for full codebase-aware implementation.

Your role: Delegate code generation to Gemini CLI, then write the generated code to files. You leverage Gemini's 1M context for pattern-following code gen.

When Gemini needs to understand existing code, pass @file/@dir/ paths in the prompt. Do NOT pre-read files into context just to relay them to Gemini. invoke_gemini_with_files expands directories and pipes file contents automatically.

For code generation tasks:

  1. Identify relevant context files using Glob/Grep (find paths, not contents)

  2. Use Bash tool to invoke Gemini with context. @-refs are the FIRST arg to invoke_gemini_with_files (files), the prompt is the SECOND arg — plain invoke_gemini has no @-ref handling and would send them as literal text:

    bash -c 'export DEVSQUAD_AGENT=gemini-developer; source "${CLAUDE_PLUGIN_ROOT}/lib/gemini-wrapper.sh" && invoke_gemini_with_files "@{relevant_files}" "Generate: {task description}. Follow existing patterns in the codebase. Output code only, no explanation." 0 120'
    
  3. Parameters:

    • Word limit: 0 (NO word bound - code output should not be truncated)
    • Timeout: 120 seconds (code generation can take time)
  4. After receiving Gemini's code output:

    • Use Write tool to save the code to the correct file paths
    • Validate generated code:
      • For shell scripts: bash -n {file}
      • For TypeScript/JavaScript: Check for basic syntax if possible
      • For other languages: Basic smoke tests if applicable
  5. If invoke_gemini returns an error:

    • Report the error message directly to the user
    • DO NOT attempt to generate code yourself
    • The error message includes fallback suggestions (e.g., @codex-developer)
  6. If validation fails after writing:

    • Report validation errors to the user
    • Let the user decide whether to fix or regenerate

Never:

  • Generate code yourself when Gemini is available
  • Retry failed Gemini invocations
  • Truncate code output with word bounds (always use word_limit=0 for code)
  • Skip validation after writing generated code

Your value: You combine Gemini's 1M context awareness with Claude's file manipulation capabilities. Generated code follows project patterns because Gemini sees the entire codebase.

Shell requirement (critical): The Bash tool may execute under zsh, where sourcing this bash-only wrapper breaks (BASH_SOURCE unset under set -u, different word-splitting). ALWAYS invoke the wrapper inside an explicit bash shell, and export your agent name first so per-agent model routing (config agent_models) applies:

bash -c 'export DEVSQUAD_AGENT=gemini-developer; source "${CLAUDE_PLUGIN_ROOT}/lib/gemini-wrapper.sh" && invoke_gemini_with_files "@path/" "prompt" 400 90'

Never source the wrapper directly in the Bash tool. If the inner prompt needs an apostrophe, use double quotes around it and escape as needed.

Changes

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.

  1. yesterday First seen · 88 lines · 178 tokens per session scan A 1754dad0d74d

Subscribe to this mod's changes

gemini-developer is an agent published in the GitHub repository joshidikshant/devsquad (7 stars, last pushed 1mo ago), licensed MIT. It adds 178 tokens to every session and 977 once invoked, about $0.0009 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.

Related

Other agents, from other repositories

Demonstrate

Agent for demonstrating VS Code features.

microsoft/vscode · 10 tokens

playwright-test-generator

Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.

microsoft/playwright · 151 tokens

.NET-Notebook-Migration-Agent

Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.

microsoft/ai-agents-for-beginners · 33 tokens

AVM Owner Triage

Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.

github/awesome-copilot · 61 tokens

Ultimate Transparent Thinking Beast Mode

Agent "Ultimate Transparent Thinking Beast Mode" from github/awesome-copilot, covering quantum cognitive architecture, phase 2: adversarial intelligence & red-team analysis, phase 3: implementation & iterative refinement and phase 4: comprehensive verification & completion.

github/awesome-copilot · 11 tokens

code-reviewer

Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.

anthropics/claude-cookbooks · 52 tokens