Tool tree
gradeNot graded
codex_scan_workspaceAnswer
Reads an allowlisted set of Codex, MCP, package, instruction, and skill files inside the folder the person already granted. It never executes downloaded code or opens a picker. If no browser workspace is connected, it returns a bounded Codex host-handoff fallback rather than claiming access.
View tool JSON
{
"name": "codex_scan_workspace",
"kind": "answer",
"impl": "imperative",
"description": "Reads an allowlisted set of Codex, MCP, package, instruction, and skill files inside the folder the person already granted. It never executes downloaded code or opens a picker. If no browser workspace is connected, it returns a bounded Codex host-handoff fallback rather than claiming access.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_get_tool_inventoryAnswer
Returns configured MCP servers and static MCP-related package signals. It distinguishes declarations from live runtime tools and never returns secrets or raw files.
View tool JSON
{
"name": "codex_get_tool_inventory",
"kind": "answer",
"impl": "imperative",
"description": "Returns configured MCP servers and static MCP-related package signals. It distinguishes declarations from live runtime tools and never returns secrets or raw files.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_get_findingsAnswer
Returns evidence-based configuration, instruction-chain, package-wiring, and readiness findings from the latest scan.
View tool JSON
{
"name": "codex_get_findings",
"kind": "answer",
"impl": "imperative",
"description": "Returns evidence-based configuration, instruction-chain, package-wiring, and readiness findings from the latest scan.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_get_instruction_chainAnswer
Returns the ordered AGENTS and SKILL metadata found in scope without returning instruction text.
View tool JSON
{
"name": "codex_get_instruction_chain",
"kind": "answer",
"impl": "imperative",
"description": "Returns the ordered AGENTS and SKILL metadata found in scope without returning instruction text.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_get_workspace_graphAnswer
Returns a sanitized graph linking allowlisted configuration and package artifacts to their declared MCP signals and current non-healthy findings. It describes static evidence only and never executes code or returns file contents.
View tool JSON
{
"name": "codex_get_workspace_graph",
"kind": "answer",
"impl": "imperative",
"description": "Returns a sanitized graph linking allowlisted configuration and package artifacts to their declared MCP signals and current non-healthy findings. It describes static evidence only and never executes code or returns file contents.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_explain_findingAction
Returns one finding by id so Codex and the user can discuss the evidence and next step without exposing file contents.
View tool JSON
{
"name": "codex_explain_finding",
"kind": "act",
"impl": "imperative",
"description": "Returns one finding by id so Codex and the user can discuss the evidence and next step without exposing file contents.",
"inputSchema": {
"type": "object",
"properties": {
"findingId": {
"type": "string"
}
},
"required": [
"findingId"
],
"additionalProperties": false
}
}codex_plan_hardeningAction
Creates a review-only plan from the latest scan. Ambiguous commands, TOML, policy, and instruction changes remain manual. The person must use the visible browser control to grant write access to the selected folder; an agent tool call cannot open that permission dialog.
View tool JSON
{
"name": "codex_plan_hardening",
"kind": "act",
"impl": "imperative",
"description": "Creates a review-only plan from the latest scan. Ambiguous commands, TOML, policy, and instruction changes remain manual. The person must use the visible browser control to grant write access to the selected folder; an agent tool call cannot open that permission dialog.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_request_host_handoffAction
Returns a bounded handshake for Codex to request native host filesystem permission, use fs/readDirectory/fs/readFile or fs/copy/fs/writeFile, and return only an allowlisted snapshot. This page never invokes host permissions itself.
View tool JSON
{
"name": "codex_request_host_handoff",
"kind": "act",
"impl": "imperative",
"description": "Returns a bounded handshake for Codex to request native host filesystem permission, use fs/readDirectory/fs/readFile or fs/copy/fs/writeFile, and return only an allowlisted snapshot. This page never invokes host permissions itself.",
"inputSchema": {
"type": "object",
"properties": {
"operation": {
"type": "string",
"enum": [
"scan",
"apply"
]
},
"actionIds": {
"type": "array",
"items": {
"type": "string"
}
}
},
"required": [
"operation"
],
"additionalProperties": false
}
}codex_submit_host_snapshotSensitive Action
Accepts only relative, allowlisted files read by Codex after native host approval and updates the shared readiness view. Raw files are analyzed in the page and are never returned by this tool.
View tool JSON
{
"name": "codex_submit_host_snapshot",
"kind": "transact",
"impl": "imperative",
"description": "Accepts only relative, allowlisted files read by Codex after native host approval and updates the shared readiness view. Raw files are analyzed in the page and are never returned by this tool.",
"inputSchema": {
"type": "object",
"properties": {
"files": {
"type": "array",
"minItems": 1,
"maxItems": 240,
"items": {
"type": "object",
"properties": {
"path": {
"type": "string"
},
"text": {
"type": "string"
}
},
"required": [
"path",
"text"
],
"additionalProperties": false
}
}
},
"required": [
"files"
],
"additionalProperties": false
}
}codex_apply_hardeningAction
After exact action ids are reviewed, applies only the in-memory demo action. In a real workspace, the person uses MCPation’s visible browser control to grant write access, create a sibling backup, apply the listed JSON cleanup, and rescan; this tool returns a bounded Codex host-handoff fallback because a tool call cannot open the required user-gesture permission dialog. It never invents commands, URLs, policies, or instruction text.
View tool JSON
{
"name": "codex_apply_hardening",
"kind": "act",
"impl": "imperative",
"description": "After exact action ids are reviewed, applies only the in-memory demo action. In a real workspace, the person uses MCPation’s visible browser control to grant write access, create a sibling backup, apply the listed JSON cleanup, and rescan; this tool returns a bounded Codex host-handoff fallback because a tool call cannot open the required user-gesture permission dialog. It never invents commands, URLs, policies, or instruction text.",
"inputSchema": {
"type": "object",
"properties": {
"actionIds": {
"type": "array",
"items": {
"type": "string"
},
"minItems": 1
},
"confirm": {
"type": "boolean",
"const": true
}
},
"required": [
"actionIds",
"confirm"
],
"additionalProperties": false
}
}codex_verify_workspaceAction
Rescans the granted workspace and returns the new readiness score, findings, and declared tool surface. The visible audit trail marks the verification only after this rescan completes.
View tool JSON
{
"name": "codex_verify_workspace",
"kind": "act",
"impl": "imperative",
"description": "Rescans the granted workspace and returns the new readiness score, findings, and declared tool surface. The visible audit trail marks the verification only after this rescan completes.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}codex_get_access_scopeAction
Returns the active folder boundary, access mode, and privacy guarantees. It never claims whole-system access. A read-only import includes the bounded Codex host-handoff fallback; a direct browser grant can be rescanned in place.
View tool JSON
{
"name": "codex_get_access_scope",
"kind": "act",
"impl": "imperative",
"description": "Returns the active folder boundary, access mode, and privacy guarantees. It never claims whole-system access. A read-only import includes the bounded Codex host-handoff fallback; a direct browser grant can be rescanned in place.",
"inputSchema": {
"type": "object",
"properties": {},
"additionalProperties": false
}
}