codi-cli

codi-cli / schemas

schemas

Type Aliases

AgentFrontmatterInput

type AgentFrontmatterInput = object;

Defined in: schemas/agent.ts:102

Validates the YAML frontmatter of a .codi/agents/<name>.md file.

Agent definitions are sub-agents that can be invoked during skill execution. Each agent has a system prompt (the Markdown body) and frontmatter that controls its behavior, tool access, and model settings.

Type Declaration

background?

optional background?: boolean;

color?

optional color?: string;

description?

optional description?: string;

disallowedTools?

optional disallowedTools?: string[];

effort?

optional effort?: "high" | "medium" | "low" | "max";

isolation?

optional isolation?: string;

managed_by?

optional managed_by?: "codi" | "user";

maxTurns?

optional maxTurns?: number;

mcpServers?

optional mcpServers?: string[];

memory?

optional memory?: "user" | "project" | "none";

model?

optional model?: string;

name

name: string;

permissionMode?

optional permissionMode?: "unrestricted" | "readonly" | "limited";

skills?

optional skills?: string[];

tools?

optional tools?: string[];

version?

optional version?: number;

AgentFrontmatterOutput

type AgentFrontmatterOutput = object;

Defined in: schemas/agent.ts:103

Validates the YAML frontmatter of a .codi/agents/<name>.md file.

Agent definitions are sub-agents that can be invoked during skill execution. Each agent has a system prompt (the Markdown body) and frontmatter that controls its behavior, tool access, and model settings.

Type Declaration

background?

optional background?: boolean;

color?

optional color?: string;

description

description: string;

disallowedTools?

optional disallowedTools?: string[];

effort?

optional effort?: "high" | "medium" | "low" | "max";

isolation?

optional isolation?: string;

managed_by

managed_by: "codi" | "user";

maxTurns?

optional maxTurns?: number;

mcpServers?

optional mcpServers?: string[];

memory?

optional memory?: "user" | "project" | "none";

model?

optional model?: string;

name

name: string;

permissionMode?

optional permissionMode?: "unrestricted" | "readonly" | "limited";

skills?

optional skills?: string[];

tools?

optional tools?: string[];

version

version: number;

EvalCase

type EvalCase = object;

Defined in: schemas/evals.ts:49

Validates a single evaluation test case for a skill.

Eval cases define prompts and expected behaviors used to assess skill quality during development and evolution.

Type Declaration

description

description: string;

expectations

expectations: string[];

files

files: string[];

id

id: string;

lastRunAt?

optional lastRunAt?: string;

passed?

optional passed?: boolean;

passRate?

optional passRate?: number;

prompt

prompt: string;

EvalsData

type EvalsData = object;

Defined in: schemas/evals.ts:50

Validates the evals.json file for a skill.

Each skill can have an evals.json file in its directory containing test cases used to evaluate the skill’s behavior and guide evolution.

Type Declaration

cases

cases: object[];

lastUpdated?

optional lastUpdated?: string;

skillName

skillName: string;

FeedbackAgent

type FeedbackAgent = typeof FEEDBACK_AGENTS[number];

Defined in: schemas/feedback.ts:65

FeedbackEntry

type FeedbackEntry = object;

Defined in: schemas/feedback.ts:64

Validates a complete skill feedback entry stored in .codi/feedback/.

Each entry records a single skill execution outcome with optional issues and improvement suggestions. Entries are used by codi skill evolve to propose targeted skill improvements.

Type Declaration

agent

agent: "claude-code" | "cursor" | "codex" | "windsurf" | "cline" | "copilot";

id

id: string;

issues

issues: object[];

outcome

outcome: "success" | "partial" | "failure";

skillName

skillName: string;

suggestions

suggestions: string[];

taskSummary

taskSummary: string;

timestamp

timestamp: string;

FeedbackIssue

type FeedbackIssue = object;

Defined in: schemas/feedback.ts:63

Validates a single issue reported in skill feedback.

Issues are structured observations about problems encountered during skill execution, used to drive skill evolution and improvement.

Type Declaration

category

category: 
  | "trigger-miss"
  | "trigger-false"
  | "unclear-step"
  | "missing-step"
  | "wrong-output"
  | "context-overflow"
  | "other";

description

description: string;

severity

severity: "high" | "medium" | "low";

FeedbackOutcome

type FeedbackOutcome = typeof FEEDBACK_OUTCOMES[number];

Defined in: schemas/feedback.ts:66

FlagConditionsOutput

type FlagConditionsOutput = object;

Defined in: schemas/flag.ts:101

Conditions that gate a “conditional” mode flag.

At least one field must be present. When multiple fields are set, all must match.

Type Declaration

agent?

optional agent?: string[];

file_pattern?

optional file_pattern?: string[];

framework?

optional framework?: string[];

lang?

optional lang?: string[];

FlagDefinitionOutput

type FlagDefinitionOutput = object;

Defined in: schemas/flag.ts:102

Validates a single flag entry in .codi/flags.yaml.

Flags control feature toggles and configuration values that are injected into agent instruction files during generation.

Type Declaration

conditions?

optional conditions?: object;

conditions.agent?

optional agent?: string[];

conditions.file_pattern?

optional file_pattern?: string[];

conditions.framework?

optional framework?: string[];

conditions.lang?

optional lang?: string[];

locked

locked: boolean;

mode

mode: 
  | "enforced"
  | "enabled"
  | "disabled"
  | "inherited"
  | "delegated_to_agent_default"
  | "conditional";

value?

optional value?: string | number | boolean | string[];

FlagModeOutput

type FlagModeOutput = 
  | "enforced"
  | "enabled"
  | "disabled"
  | "inherited"
  | "delegated_to_agent_default"
  | "conditional";

Defined in: schemas/flag.ts:100

Valid values for the mode field in a flag definition.

Controls how Codi manages the flag’s value in generated agent instruction files.

HookDefinitionOutput

type HookDefinitionOutput = object;

Defined in: schemas/hooks.ts:39

Validates a single hook definition entry in a hooks configuration file.

Type Declaration

command

command: string;

condition

condition: string;

name

name: string;

staged_filter?

optional staged_filter?: string;

HooksConfigOutput

type HooksConfigOutput = object;

Defined in: schemas/hooks.ts:40

Validates the generated .codi/hooks.yaml configuration file.

This file is produced by codi generate (not hand-authored) and describes how hooks are installed and managed across supported hook runners.

Type Declaration

custom

custom: Record<string, object[]>;

hooks

hooks: Record<string, Record<string, object[]>>;

install_method

install_method: "git-hooks" | "husky-append" | "pre-commit-append" | "manual";

runner

runner: "codi" | "none" | "husky" | "pre-commit";

version

version: "1";

IssueCategory

type IssueCategory = typeof ISSUE_CATEGORIES[number];

Defined in: schemas/feedback.ts:67

McpConfigInput

type McpConfigInput = object;

Defined in: schemas/mcp.ts:46

Validates the contents of .codi/mcp.yaml.

Defines MCP (Model Context Protocol) servers available to agents in this project. Server names defined here can be referenced in agent frontmatter via mcpServers.

Type Declaration

servers?

optional servers?: Record<string, {
  args?: string[];
  command?: string;
  enabled?: boolean;
  env?: Record<string, string>;
  headers?: Record<string, string>;
  type?: "stdio" | "http";
  url?: string;
}>;

McpConfigOutput

type McpConfigOutput = object;

Defined in: schemas/mcp.ts:47

Validates the contents of .codi/mcp.yaml.

Defines MCP (Model Context Protocol) servers available to agents in this project. Server names defined here can be referenced in agent frontmatter via mcpServers.

Type Declaration

servers

servers: Record<string, {
  args?: string[];
  command?: string;
  enabled?: boolean;
  env?: Record<string, string>;
  headers?: Record<string, string>;
  type?: "stdio" | "http";
  url?: string;
}>;

ProjectManifestInput

type ProjectManifestInput = object;

Defined in: schemas/manifest.ts:74

Validates the contents of a project’s codi.yaml manifest file.

The manifest is the primary project configuration document at .codi/codi.yaml. It declares the project name, which agents to target, and which presets are installed.

Type Declaration

agents?

optional agents?: string[];

description?

optional description?: string;

engine?

optional engine?: object;

engine.requiredVersion?

optional requiredVersion?: string;

layers?

optional layers?: object;

layers.agents?

optional agents?: boolean;

layers.context?

optional context?: boolean;

layers.rules?

optional rules?: boolean;

layers.skills?

optional skills?: boolean;

name

name: string;

presetRegistry?

optional presetRegistry?: object;

presetRegistry.branch?

optional branch?: string;

presetRegistry.url

url: string;

presets?

optional presets?: string[];

version

version: "1";

ProjectManifestOutput

type ProjectManifestOutput = object;

Defined in: schemas/manifest.ts:75

Validates the contents of a project’s codi.yaml manifest file.

The manifest is the primary project configuration document at .codi/codi.yaml. It declares the project name, which agents to target, and which presets are installed.

Type Declaration

agents?

optional agents?: string[];

description?

optional description?: string;

engine?

optional engine?: object;

engine.requiredVersion?

optional requiredVersion?: string;

layers?

optional layers?: object;

layers.agents

agents: boolean;

layers.context

context: boolean;

layers.rules

rules: boolean;

layers.skills

skills: boolean;

name

name: string;

presetRegistry?

optional presetRegistry?: object;

presetRegistry.branch

branch: string;

presetRegistry.url

url: string;

presets?

optional presets?: string[];

version

version: "1";

RuleFrontmatterInput

type RuleFrontmatterInput = object;

Defined in: schemas/rule.ts:64

Validates the YAML frontmatter of a .codi/rules/<name>.md file.

Rules are Markdown documents with frontmatter that define coding standards, conventions, and guidelines injected into agent instruction files.

Type Declaration

alwaysApply?

optional alwaysApply?: boolean;

description

description: string;

language?

optional language?: string;

managed_by?

optional managed_by?: "codi" | "user";

name

name: string;

priority?

optional priority?: "high" | "medium" | "low";

scope?

optional scope?: string[];

type?

optional type?: "rule";

version?

optional version?: number;

RuleFrontmatterOutput

type RuleFrontmatterOutput = object;

Defined in: schemas/rule.ts:65

Validates the YAML frontmatter of a .codi/rules/<name>.md file.

Rules are Markdown documents with frontmatter that define coding standards, conventions, and guidelines injected into agent instruction files.

Type Declaration

alwaysApply

alwaysApply: boolean;

description

description: string;

language?

optional language?: string;

managed_by

managed_by: "codi" | "user";

name

name: string;

priority

priority: "high" | "medium" | "low";

scope?

optional scope?: string[];

type

type: "rule";

version

version: number;

SkillFrontmatterInput

type SkillFrontmatterInput = object;

Defined in: schemas/skill.ts:135

Validates the YAML frontmatter of a .codi/skills/<name>/SKILL.md file.

Skills are reusable agent workflows with structured frontmatter that controls how they are invoked, which platforms they target, and how they run.

Type Declaration

agent?

optional agent?: string;

allowedTools?

optional allowedTools?: string[];

argumentHint?

optional argumentHint?: string;

category?

optional category?: string;

compatibility?

optional compatibility?: ("claude-code" | "cursor" | "codex" | "windsurf" | "cline" | "copilot")[];

context?

optional context?: "fork";

description

description: string;

disableModelInvocation?

optional disableModelInvocation?: boolean;

effort?

optional effort?: "high" | "medium" | "low" | "max";

hooks?

optional hooks?: Record<string, string | string[]>;

license?

optional license?: string;

managed_by?

optional managed_by?: "codi" | "user";

metadata?

optional metadata?: Record<string, string>;

model?

optional model?: string;

name

name: string;

paths?

optional paths?: string | string[];

shell?

optional shell?: "bash" | "powershell";

tools?

optional tools?: string[];

type?

optional type?: "skill";

user-invocable?

optional user-invocable?: boolean;

version?

optional version?: number;

SkillFrontmatterOutput

type SkillFrontmatterOutput = object;

Defined in: schemas/skill.ts:136

Validates the YAML frontmatter of a .codi/skills/<name>/SKILL.md file.

Skills are reusable agent workflows with structured frontmatter that controls how they are invoked, which platforms they target, and how they run.

Type Declaration

agent?

optional agent?: string;

allowedTools?

optional allowedTools?: string[];

argumentHint?

optional argumentHint?: string;

category?

optional category?: string;

compatibility?

optional compatibility?: ("claude-code" | "cursor" | "codex" | "windsurf" | "cline" | "copilot")[];

context?

optional context?: "fork";

description

description: string;

disableModelInvocation?

optional disableModelInvocation?: boolean;

effort?

optional effort?: "high" | "medium" | "low" | "max";

hooks?

optional hooks?: Record<string, string | string[]>;

license?

optional license?: string;

managed_by

managed_by: "codi" | "user";

metadata?

optional metadata?: Record<string, string>;

model?

optional model?: string;

name

name: string;

paths?

optional paths?: string | string[];

shell?

optional shell?: "bash" | "powershell";

tools?

optional tools?: string[];

type

type: "skill";

user-invocable?

optional user-invocable?: boolean;

version

version: number;

SkillTestManifest

type SkillTestManifest = object;

Defined in: schemas/skill-test.ts:27

Type Declaration

skill

skill: string;

tiers

tiers: object;

tiers.behavior?

optional behavior?: object;

tiers.behavior.port?

optional port?: number;

tiers.behavior.server

server: string;

tiers.behavior.startScript

startScript: string;

tiers.behavior.tests

tests: string;

tiers.contract

contract: boolean;

tiers.logic?

optional logic?: object;

tiers.logic.lib

lib: string;

tiers.logic.tests

tests: string;

Variables

AgentFrontmatterSchema

const AgentFrontmatterSchema: ZodObject<AgentFrontmatterOutput>;

Defined in: schemas/agent.ts:16

Validates the YAML frontmatter of a .codi/agents/<name>.md file.

Agent definitions are sub-agents that can be invoked during skill execution. Each agent has a system prompt (the Markdown body) and frontmatter that controls its behavior, tool access, and model settings.

EvalCaseSchema

const EvalCaseSchema: ZodObject<EvalCase>;

Defined in: schemas/evals.ts:9

Validates a single evaluation test case for a skill.

Eval cases define prompts and expected behaviors used to assess skill quality during development and evolution.

EvalsDataSchema

const EvalsDataSchema: ZodObject<EvalsData>;

Defined in: schemas/evals.ts:39

Validates the evals.json file for a skill.

Each skill can have an evals.json file in its directory containing test cases used to evaluate the skill’s behavior and guide evolution.

FEEDBACK_AGENTS

const FEEDBACK_AGENTS: readonly ["claude-code", "codex", "cursor", "windsurf", "cline", "copilot"];

Defined in: schemas/feedback.ts:5

Agent ids that can be the source of skill feedback entries.

FEEDBACK_OUTCOMES

const FEEDBACK_OUTCOMES: readonly ["success", "partial", "failure"];

Defined in: schemas/feedback.ts:8

Possible outcomes of a skill execution: ‘success’, ‘partial’, or ‘failure’.

FeedbackEntrySchema

const FeedbackEntrySchema: ZodObject<FeedbackEntry>;

Defined in: schemas/feedback.ts:52

Validates a complete skill feedback entry stored in .codi/feedback/.

Each entry records a single skill execution outcome with optional issues and improvement suggestions. Entries are used by codi skill evolve to propose targeted skill improvements.

FeedbackIssueSchema

const FeedbackIssueSchema: ZodObject<FeedbackIssue>;

Defined in: schemas/feedback.ts:30

Validates a single issue reported in skill feedback.

Issues are structured observations about problems encountered during skill execution, used to drive skill evolution and improvement.

FlagConditionsSchema

const FlagConditionsSchema: ZodObject<FlagConditionsOutput>;

Defined in: schemas/flag.ts:26

Conditions that gate a “conditional” mode flag.

At least one field must be present. When multiple fields are set, all must match.

FlagDefinitionSchema

const FlagDefinitionSchema: ZodObject<FlagDefinitionOutput>;

Defined in: schemas/flag.ts:65

Validates a single flag entry in .codi/flags.yaml.

Flags control feature toggles and configuration values that are injected into agent instruction files during generation.

FlagModeSchema

const FlagModeSchema: ZodEnum<FlagModeOutput>;

Defined in: schemas/flag.ts:8

Valid values for the mode field in a flag definition.

Controls how Codi manages the flag’s value in generated agent instruction files.

FlagValueSchema

const FlagValueSchema: ZodUnion<readonly [ZodBoolean, ZodNumber, ZodString, ZodArray<ZodString>]>;

Defined in: schemas/flag.ts:55

The valid value types for a flag.

HookDefinitionSchema

const HookDefinitionSchema: ZodObject<HookDefinitionOutput>;

Defined in: schemas/hooks.ts:5

Validates a single hook definition entry in a hooks configuration file.

HooksConfigSchema

const HooksConfigSchema: ZodObject<HooksConfigOutput>;

Defined in: schemas/hooks.ts:18

Validates the generated .codi/hooks.yaml configuration file.

This file is produced by codi generate (not hand-authored) and describes how hooks are installed and managed across supported hook runners.

ISSUE_CATEGORIES

const ISSUE_CATEGORIES: readonly ["trigger-miss", "trigger-false", "unclear-step", "missing-step", "wrong-output", "context-overflow", "other"];

Defined in: schemas/feedback.ts:11

Categories for classifying issues found during skill feedback collection.

McpConfigSchema

const McpConfigSchema: ZodObject<McpConfigOutput>;

Defined in: schemas/mcp.ts:37

Validates the contents of .codi/mcp.yaml.

Defines MCP (Model Context Protocol) servers available to agents in this project. Server names defined here can be referenced in agent frontmatter via mcpServers.

ProjectManifestSchema

const ProjectManifestSchema: ZodObject<ProjectManifestOutput>;

Defined in: schemas/manifest.ts:10

Validates the contents of a project’s codi.yaml manifest file.

The manifest is the primary project configuration document at .codi/codi.yaml. It declares the project name, which agents to target, and which presets are installed.

RuleFrontmatterSchema

const RuleFrontmatterSchema: ZodObject<RuleFrontmatterOutput>;

Defined in: schemas/rule.ts:15

Validates the YAML frontmatter of a .codi/rules/<name>.md file.

Rules are Markdown documents with frontmatter that define coding standards, conventions, and guidelines injected into agent instruction files.

SkillFrontmatterSchema

const SkillFrontmatterSchema: ZodObject<SkillFrontmatterOutput>;

Defined in: schemas/skill.ts:19

Validates the YAML frontmatter of a .codi/skills/<name>/SKILL.md file.

Skills are reusable agent workflows with structured frontmatter that controls how they are invoked, which platforms they target, and how they run.

SkillTestManifestSchema

const SkillTestManifestSchema: ZodObject<SkillTestManifest>;

Defined in: schemas/skill-test.ts:3