Tool tree
Release Evidence Room pairs a public fictional checkout QA target with a privacy-safe WebMCP review room where agents inspect bounded runtime evidence and humans decide.
Implementation
gradeNot graded
gradeNot graded
get_release_snapshotAnswer
Read the active synthetic release, checks, evidence gaps, every proposal with its ID and status, human decision, and current state version. Use the returned stateVersion for the next proposal. Appends a local audit entry but makes no release change.
View tool JSON
{
"name": "get_release_snapshot",
"kind": "answer",
"impl": "imperative",
"description": "Read the active synthetic release, checks, evidence gaps, every proposal with its ID and status, human decision, and current state version. Use the returned stateVersion for the next proposal. Appends a local audit entry but makes no release change.",
"inputSchema": {
"type": "object",
"properties": {},
"required": [],
"additionalProperties": false
},
"page": "/"
}query_network_evidenceAnswer
Filter only the bounded, redacted network evidence already present in this page. Updates local focus and audit state, but never fetches a URL, changes source evidence, or returns raw traffic.
View tool JSON
{
"name": "query_network_evidence",
"kind": "answer",
"impl": "imperative",
"description": "Filter only the bounded, redacted network evidence already present in this page. Updates local focus and audit state, but never fetches a URL, changes source evidence, or returns raw traffic.",
"inputSchema": {
"type": "object",
"properties": {
"riskType": {
"type": "string",
"enum": [
"duplicate_side_effect",
"response_loss"
],
"description": "Optional risk category already present in this room."
},
"severity": {
"type": "string",
"enum": [
"high",
"medium",
"low"
],
"description": "Optional severity filter for page-owned evidence."
},
"limit": {
"type": "integer",
"minimum": 1,
"maximum": 20,
"default": 20,
"description": "Maximum evidence items to return."
}
},
"required": [],
"additionalProperties": false
},
"page": "/"
}propose_test_caseAction
Create an evidence-linked pending test proposal for human review. Does not approve or execute the test.
View tool JSON
{
"name": "propose_test_case",
"kind": "act",
"impl": "imperative",
"description": "Create an evidence-linked pending test proposal for human review. Does not approve or execute the test.",
"inputSchema": {
"type": "object",
"properties": {
"expectedStateVersion": {
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991,
"description": "Copy stateVersion from the latest get_release_snapshot result."
},
"clientRequestId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": "^[A-Za-z0-9._:-]+$",
"description": "Use a unique ID per logical proposal; reuse it only to retry identical content."
},
"title": {
"type": "string",
"minLength": 1,
"maxLength": 120,
"pattern": ".*\\S.*",
"description": "Concise name for the missing regression test."
},
"given": {
"type": "string",
"minLength": 1,
"maxLength": 500,
"pattern": ".*\\S.*",
"description": "Test precondition grounded in room evidence."
},
"when": {
"type": "string",
"minLength": 1,
"maxLength": 500,
"pattern": ".*\\S.*",
"description": "Action or failure condition exercised by the test."
},
"then": {
"type": "string",
"minLength": 1,
"maxLength": 500,
"pattern": ".*\\S.*",
"description": "Observable safety outcome the test must assert."
},
"evidenceIds": {
"type": "array",
"minItems": 1,
"maxItems": 10,
"uniqueItems": true,
"items": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": ".*\\S.*"
},
"description": "Use evidence IDs returned by query_network_evidence."
}
},
"required": [
"expectedStateVersion",
"clientRequestId",
"title",
"given",
"when",
"then",
"evidenceIds"
],
"additionalProperties": false
},
"page": "/"
}run_approved_verificationAction
Execute an approved test proposal only inside this page's bounded synthetic sandbox. Supports an exact retry replay or a seeded state-machine monkey run with at most 100 steps. Returns risk_confirmed, not_reproduced, or inconclusive; never calls a live service or runs arbitrary code.
View tool JSON
{
"name": "run_approved_verification",
"kind": "act",
"impl": "imperative",
"description": "Execute an approved test proposal only inside this page's bounded synthetic sandbox. Supports an exact retry replay or a seeded state-machine monkey run with at most 100 steps. Returns risk_confirmed, not_reproduced, or inconclusive; never calls a live service or runs arbitrary code.",
"inputSchema": {
"type": "object",
"properties": {
"expectedStateVersion": {
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991,
"description": "Copy stateVersion from the latest get_release_snapshot result."
},
"clientRequestId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": "^[A-Za-z0-9._:-]+$",
"description": "Use a unique ID per logical run; reuse it only to retry identical input."
},
"testProposalId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": ".*\\S.*",
"description": "An approved test proposal to execute in the bounded synthetic sandbox."
},
"strategy": {
"type": "string",
"enum": [
"targeted_retry",
"seeded_monkey"
],
"description": "Run the exact retry replay or a reproducible bounded state-machine exploration."
},
"seed": {
"type": "integer",
"minimum": 0,
"maximum": 2147483647,
"description": "Required only for seeded_monkey so the same trace can be reproduced."
},
"maxSteps": {
"type": "integer",
"minimum": 1,
"maximum": 100,
"description": "Required only for seeded_monkey; hard bounds the explored transitions."
}
},
"required": [
"expectedStateVersion",
"clientRequestId",
"testProposalId",
"strategy"
],
"oneOf": [
{
"properties": {
"strategy": {
"const": "targeted_retry"
}
},
"not": {
"anyOf": [
{
"required": [
"seed"
]
},
{
"required": [
"maxSteps"
]
}
]
}
},
{
"properties": {
"strategy": {
"const": "seeded_monkey"
}
},
"required": [
"seed",
"maxSteps"
]
}
],
"additionalProperties": false
},
"page": "/"
}propose_release_decisionAction
Create a non-binding READY or HOLD proposal grounded in room evidence. Requires an approved test proposal and its matching verification result; READY additionally requires a not_reproduced verdict. Cannot confirm, deploy, or release.
View tool JSON
{
"name": "propose_release_decision",
"kind": "act",
"impl": "imperative",
"description": "Create a non-binding READY or HOLD proposal grounded in room evidence. Requires an approved test proposal and its matching verification result; READY additionally requires a not_reproduced verdict. Cannot confirm, deploy, or release.",
"inputSchema": {
"type": "object",
"properties": {
"expectedStateVersion": {
"type": "integer",
"minimum": 0,
"maximum": 9007199254740991,
"description": "Copy stateVersion from the latest get_release_snapshot result."
},
"clientRequestId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": "^[A-Za-z0-9._:-]+$",
"description": "Use a unique ID per logical proposal; reuse it only to retry identical content."
},
"recommendation": {
"type": "string",
"enum": [
"ready",
"hold"
],
"description": "Non-binding recommendation; only a human can confirm it."
},
"rationale": {
"type": "string",
"minLength": 1,
"maxLength": 1000,
"pattern": ".*\\S.*",
"description": "Evidence-grounded explanation for the recommendation."
},
"evidenceIds": {
"type": "array",
"minItems": 1,
"maxItems": 10,
"uniqueItems": true,
"items": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": ".*\\S.*"
},
"description": "Use evidence IDs returned by query_network_evidence."
},
"testProposalId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": ".*\\S.*",
"description": "ID of the approved test proposal that this decision depends on."
},
"verificationResultId": {
"type": "string",
"minLength": 1,
"maxLength": 80,
"pattern": ".*\\S.*",
"description": "Verification result ID returned by run_approved_verification for the approved test."
}
},
"required": [
"expectedStateVersion",
"clientRequestId",
"recommendation",
"rationale",
"evidenceIds",
"testProposalId",
"verificationResultId"
],
"additionalProperties": false
},
"page": "/"
}