forge
Developer not provided
Description: Adversarial code hardening for Claude Code. Forge spawns two agents — a Builder and a Breaker — that iterate against each other until the code is unbreakable. The Builder implements a...
From the developer
Works with
Claude Code
Description
From the marketplace listing
Description: Adversarial code hardening for Claude Code. Forge spawns two agents — a Builder and a Breaker — that iterate against each other until the code is unbreakable. The Builder implements a feature, the Breaker writes adversarial tests to prove bugs, and the Builder must fix the implementation without deleting or weakening the tests. The loop continues until the Breaker can't produce a failing test. The result is battle-tested code with a permanent adversarial test suite. Key features: - Evidence-based, not opinion-based. Every issue the Breaker raises must have a failing test that proves the bug. No test, no break. This eliminates false positives and subjective feedback. - Iterative adversarial loop. Builder and Breaker are separate agents that go back and forth for up to N rounds. The Builder cannot dismiss feedback; it must make the failing tests pass. - Adversarial tests are permanent. The Breaker's tests stay in the repo under tests/adversarial/ as a regression suite. Future changes are protected by the same tests that originally caught the bugs. - Automatic checkpoint. Creates a git tag before starting so the entire forge session can be rolled back with one command. - Configurable rounds. Default 5 rounds, adjustable with --rounds N. Stops early if the Breaker can't find issues. - Auto-detects project stack. Reads config files to determine language, test framework, and conventions. Works with Python, TypeScript, Rust, Go, Ruby, and anything with a test runner.
What changed
No changes recorded yet. Follow this plugin to keep track of future changes.
Similar plugins
Find competitors to follow as you develop your plugin.
-
A pipeline of skills that implements two feedback loops for Claude Code. The inner loop (plan-interview, intent-framed-agent, context-sur...
Shared terms: implement, two, agent, code
-
Multi-agent iterative code review until zero issues remain. Spawns up to 10 specialist reviewers (code quality, security, performance, te...
Shared terms: spawn, until, agent, code
-
Eight agents and eight slash commands turn a one-line feature request into a tracked implementation pipeline. Independent verification at...
Shared terms: until, feature, agent, code
-
Full SDLC pipeline with 7 agents: architecture, code review, QA, security. Two decisions per feature, the rest is automated.
Shared terms: two, feature, agent, code
-
Automate the TDD cycle with a collaborative agent team: two parallel coders and a devil's advocate, working on isolated files for rapid, ...
Shared terms: two, agent, failing, result
Files & skills
forge · catalog manifest
Saved declaration
{}
forge · catalog manifest
Saved declaration
{
"name": "forge",
"source": {
"sha": "95ef910b5528695bded466500f14515c4e14eeb1",
"url": "https://github.com/UtkarshJain98/forge.git",
"source": "url"
},
"homepage": "https://github.com/UtkarshJain98/forge",
"description": "Description:\n\nAdversarial code hardening for Claude Code. Forge spawns two agents — a Builder and a Breaker — that iterate against each other until the code is unbreakable. The Builder implements a feature, the Breaker writes adversarial tests to prove bugs, and the Builder must fix the implementation without deleting or weakening the tests. The loop continues until the Breaker can't produce a failing test. The result is battle-tested code with a permanent adversarial test suite.\n\nKey features:\n\n - Evidence-based, not opinion-based. Every issue the Breaker raises must have a failing test that proves the bug. No test, no break. This eliminates false positives and subjective feedback.\n - Iterative adversarial loop. Builder and Breaker are separate agents that go back and forth for up to N rounds. The Builder cannot dismiss feedback; it must make the failing tests pass.\n - Adversarial tests are permanent. The Breaker's tests stay in the repo under tests/adversarial/ as a regression suite. Future changes are protected by the same tests that originally caught the bugs.\n - Automatic checkpoint. Creates a git tag before starting so the entire forge session can be rolled back with one command.\n - Configurable rounds. Default 5 rounds, adjustable with --rounds N. Stops early if the Breaker can't find issues.\n - Auto-detects project stack. Reads config files to determine language, test framework, and conventions. Works with Python, TypeScript, Rust, Go, Ruby, and anything with a test runner."
}
Catalog declarations and available plugin.json metadata are saved. Referenced skills and executable source files are not downloaded or tested. Missing components in a manifest do not prove that the plugin has none.
Installation instructions
Community
/plugin marketplace add anthropics/claude-plugins-community
/plugin install forge@claude-communityTechnical details
- Sources
- Community
- First seen
- Oct 3, 2026 · 22:13 UTC
- Last seen
- Nov 7, 2026 · 00:08 UTC
- Latest observed change
- Oct 3, 2026 · 22:13 UTC
- Collection status
- Listed
- Reported installs
- Not provided
Installation counts are source-reported, not independently verified active users. Observation dates are not release dates.
Download plugin data (JSON)Plugin history: freshly prepared. Prepared . Refreshed when catalog data changes or after 6 hours. Your account and Follow status are live.