Loading Iris Code...

Connect your agent.
Bring your rules.

Resolved rules, real findings and checked changes, delivered to your coding agent over MCP.

iris_schemaFree structure; saved project evidence included: Call before a query, migration or model change to get real table and column names. Whole-schema structure, focused-table usage and exports are Free. Aggregate schema usage and warnings are available from earned project evidence; fresh project inspection uses applicable project access. Respects repository handoff policy. For repositories with supported backend schema declarations; no live database connection.
iris_checkFree file checks; monthly project allowance: Findings for a path, each with a rule id, a line, a real severity, and a stable id the agent can pass back. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_configFree: The project's resolved thresholds, scoring weights, and gate limits, exactly as the pre-push hook and CI would apply them.
iris_fix_safeFree file fixes; approved plan fixes included: Preview, then apply, only the changes Iris Code can prove safe. Refuses per finding with a reason. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_refactor_planFree file plans; monthly project allowance: A refactor plan from the repository's own conventions: safe fixes, ranked structural suggestions with evidence and risk, and a session baseline to verify against. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_verify_refactorIncluded with your captured plan: Compares the code with the captured plan's baseline and reports improved, not improved or regressed, catching packed lines, broken imports, removed exports and new findings, with a correction. Verification of the same approved captured plan is included, including its planned files; a new plan is fresh work.
iris_explainFree for saved reviews; Pro for fresh file explanations: Findings plus the rule detail behind them, so an agent can act on the reason rather than the label. Reading an owned saved workspace scan is included. Fresh file explanations through this tool require Pro.
iris_gateFree saved gate preview; Pro enforcement: Whether the configured gate would pass, and which files fail which rule. The same answer the hook gives.
iris_check_dependencyFree package lookup; project refresh uses allowance: OSV advisories for a dependency at its installed version, with severity, confidence, and the nearest fixed version.
iris_check_packageFree: Whether one package is safe to add, with a suggested install command when a safer version exists. Keeps working when the agent handoff is turned off.

Single-file entry points stay Free. Sign in for fresh project work within your monthly allowance; opening your saved results and verifying an approved captured plan do not spend another action. Pro adds local project work without the Free monthly cap and paid enforcement. Access is checked when each tool runs.

Compare Free and Pro

Before the first edit.
After every change.

Project rules tell the agent your limits before it starts. MCP answers questions while it works. Both are free.

iris rules writes the project's limits, naming convention and gate rules into the files agents read, including AGENTS.md and CLAUDE.md.

Set up project rules

Iris Code CLI · in your project

iris rules

Connect the agent you already use.

Iris Code gives your coding agent real project rules and checks over MCP. Choose your client to set it up.

JetBrains IDEs

Open a project, then use Tools | Iris Code | Set Up MCP Server. Paste the copied configuration into your assistant's MCP settings.

Use the plugin's copied configuration.

JetBrains needs the project-specific configuration supplied by Iris Code. Open the project before using the setup action.

Get the JetBrains plugin

Read JetBrains IDEs setup

Local server. Explicit access.

The MCP server runs locally alongside your agent. Analysis stays on your machine. Free tools work without Pro activation; Pro calls require a valid Pro entitlement. Dependency and advisory lookups can use the network, with permission. Cloud repository scanning is a separate opt-in, not this connection.

Activation, permissions and full MCP setup

From a guess to a checked answer.

Over MCP, give coding agents the same project limits, findings and gate result as your hook and CI.

It reads your rules first.

Before it edits anything, the agent calls iris_config: the project's resolved thresholds, scoring weights and gate limits, exactly as the pre-push hook and CI would apply them.

72/100 · illustrative file score

Connect your agent

cancel_order.ts · agent conversationillustrative

You
Is cancel_order.ts ready to push?
Tool call · Free
iris_config
Tool result
minHealthScore 75 · gateMaxSecuritySmells 0
Agent
Your gate is 75, with no security smells allowed. Checking the file against it.

A worked example, not a live agent. Iris Code removes the provably safe debug line; the agent handles the secret and query changes. A score alone does not prove code is safe or correct.

Permission to help.
Boundaries on how.

These checks run in the MCP server. They are not instructions the model can choose to ignore.

A repository can turn off agent handoff with "agentHandoff": false, on every plan. iris_check_package keeps working.

Preview first: iris_fix_safe returns the exact changes it would make plus a single-use token. Applying requires that token, so Iris Code only ever writes a change set it has already shown you. A confirmation flag alone would not give you that, because the model sets the flag.
Refusals with reasons: A hardcoded secret is never auto-fixed. A debug print sharing its line with real code is left alone, because deleting the line would take the code with it. One refusal never blocks the fixes that are provable.
Root-confined paths: Every requested path must resolve inside the project root, checked after resolving symlinks, so a link pointing outside your workspace is refused rather than followed. There is no shell-execution tool and no command passthrough.
Masked secrets only: Secret values are never returned to the agent, only the masked form every other Iris Code surface shows. Analysis runs on your machine; only dependency and advisory lookups reach the network, and those ask first.
Disclosed truncation: Large results drop the lowest severities first and report what was withheld, so a critical finding is never cut to make room for a magic number and a partial answer cannot look complete.
The enforced numbers: An agent can count lines and guess at a limit, and its guess changes between runs. Iris Code answers with the thresholds your pre-push hook actually enforces, so work that passes here passes there.

Start with a file
you know well.

Connect Iris Code, then ask your agent whether that file is ready to push.

Choose your client Read the MCP guide

Connect your agent.
Bring your rules.

Resolved rules, real findings and checked changes, delivered to your coding agent over MCP.

iris_schemaFree structure; saved project evidence included: Call before a query, migration or model change to get real table and column names. Whole-schema structure, focused-table usage and exports are Free. Aggregate schema usage and warnings are available from earned project evidence; fresh project inspection uses applicable project access. Respects repository handoff policy. For repositories with supported backend schema declarations; no live database connection.
iris_checkFree file checks; monthly project allowance: Findings for a path, each with a rule id, a line, a real severity, and a stable id the agent can pass back. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_configFree: The project's resolved thresholds, scoring weights, and gate limits, exactly as the pre-push hook and CI would apply them.
iris_fix_safeFree file fixes; approved plan fixes included: Preview, then apply, only the changes Iris Code can prove safe. Refuses per finding with a reason. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_refactor_planFree file plans; monthly project allowance: A refactor plan from the repository's own conventions: safe fixes, ranked structural suggestions with evidence and risk, and a session baseline to verify against. One-file work is Free. Fresh project inspection and new multi-file plans use applicable project access; reopening owned evidence does not spend another action.
iris_verify_refactorIncluded with your captured plan: Compares the code with the captured plan's baseline and reports improved, not improved or regressed, catching packed lines, broken imports, removed exports and new findings, with a correction. Verification of the same approved captured plan is included, including its planned files; a new plan is fresh work.
iris_explainFree for saved reviews; Pro for fresh file explanations: Findings plus the rule detail behind them, so an agent can act on the reason rather than the label. Reading an owned saved workspace scan is included. Fresh file explanations through this tool require Pro.
iris_gateFree saved gate preview; Pro enforcement: Whether the configured gate would pass, and which files fail which rule. The same answer the hook gives.
iris_check_dependencyFree package lookup; project refresh uses allowance: OSV advisories for a dependency at its installed version, with severity, confidence, and the nearest fixed version.
iris_check_packageFree: Whether one package is safe to add, with a suggested install command when a safer version exists. Keeps working when the agent handoff is turned off.

Compare Free and Pro

Before the first edit.
After every change.

Project rules tell the agent your limits before it starts. MCP answers questions while it works. Both are free.

iris rules writes the project's limits, naming convention and gate rules into the files agents read, including AGENTS.md and CLAUDE.md.

Set up project rules

Iris Code CLI · in your project

iris rules

Connect the agent you already use.

Iris Code gives your coding agent real project rules and checks over MCP. Choose your client to set it up.

JetBrains IDEs

Open a project, then use Tools | Iris Code | Set Up MCP Server. Paste the copied configuration into your assistant's MCP settings.

Use the plugin's copied configuration.

JetBrains needs the project-specific configuration supplied by Iris Code. Open the project before using the setup action.

Get the JetBrains plugin

Read JetBrains IDEs setup

Local server. Explicit access.

Activation, permissions and full MCP setup

From a guess to a checked answer.

Over MCP, give coding agents the same project limits, findings and gate result as your hook and CI.

It reads your rules first.

Before it edits anything, the agent calls iris_config: the project's resolved thresholds, scoring weights and gate limits, exactly as the pre-push hook and CI would apply them.

72/100 · illustrative file score

Connect your agent

cancel_order.ts · agent conversationillustrative

You
Is cancel_order.ts ready to push?
Tool call · Free
iris_config
Tool result
minHealthScore 75 · gateMaxSecuritySmells 0
Agent
Your gate is 75, with no security smells allowed. Checking the file against it.

A worked example, not a live agent. Iris Code removes the provably safe debug line; the agent handles the secret and query changes. A score alone does not prove code is safe or correct.

Permission to help.
Boundaries on how.

These checks run in the MCP server. They are not instructions the model can choose to ignore.

A repository can turn off agent handoff with "agentHandoff": false, on every plan. iris_check_package keeps working.

Preview first: iris_fix_safe returns the exact changes it would make plus a single-use token. Applying requires that token, so Iris Code only ever writes a change set it has already shown you. A confirmation flag alone would not give you that, because the model sets the flag.
Refusals with reasons: A hardcoded secret is never auto-fixed. A debug print sharing its line with real code is left alone, because deleting the line would take the code with it. One refusal never blocks the fixes that are provable.
Root-confined paths: Every requested path must resolve inside the project root, checked after resolving symlinks, so a link pointing outside your workspace is refused rather than followed. There is no shell-execution tool and no command passthrough.
Masked secrets only: Secret values are never returned to the agent, only the masked form every other Iris Code surface shows. Analysis runs on your machine; only dependency and advisory lookups reach the network, and those ask first.
Disclosed truncation: Large results drop the lowest severities first and report what was withheld, so a critical finding is never cut to make room for a magic number and a partial answer cannot look complete.
The enforced numbers: An agent can count lines and guess at a limit, and its guess changes between runs. Iris Code answers with the thresholds your pre-push hook actually enforces, so work that passes here passes there.

Start with a file
you know well.

Connect Iris Code, then ask your agent whether that file is ready to push.

Choose your client Read the MCP guide

Before the first edit.After every change.

JetBrains IDEs

Use the plugin's copied configuration.

Local server. Explicit access.

It reads your rules first.

Start with a fileyou know well.

Before the first edit.After every change.

JetBrains IDEs

Use the plugin's copied configuration.

Local server. Explicit access.

It reads your rules first.

Start with a fileyou know well.

Before the first edit.
After every change.

Start with a file
you know well.

Before the first edit.
After every change.

Start with a file
you know well.