| 1 | package tool |
| 2 | |
| 3 | import ( |
| 4 | "bytes" |
| 5 | "crypto/sha256" |
| 6 | "encoding/hex" |
| 7 | "encoding/json" |
| 8 | "errors" |
| 9 | "fmt" |
| 10 | "io" |
| 11 | "strconv" |
| 12 | "strings" |
| 13 | "sync" |
| 14 | |
| 15 | jsonschema "github.com/santhosh-tekuri/jsonschema/v6" |
| 16 | "github.com/santhosh-tekuri/jsonschema/v6/kind" |
| 17 | ) |
| 18 | |
| 19 | const maxArgumentViolations = 8 |
| 20 | |
| 21 | // ArgumentValidator adds call-dependent checks that cannot be represented by |
| 22 | // one provider-visible JSON Schema. JSON Schema validation always runs first. |
| 23 | type ArgumentValidator interface { |
| 24 | ValidateArguments(json.RawMessage) []ArgumentViolation |
| 25 | } |
| 26 | |
| 27 | // CapabilityArgumentContract is the effective inner contract exposed when a |
| 28 | // stable proxy injects target-specific fields such as a skill name. |
| 29 | type CapabilityArgumentContract struct { |
| 30 | Schema json.RawMessage `json:"input_schema"` |
| 31 | Example json.RawMessage `json:"call_example,omitempty"` |
| 32 | } |
| 33 | |
| 34 | // CapabilityArgumentProvider lets an inspect action show the actual inner |
| 35 | // contract instead of the proxy's generic arguments object. |
| 36 | type CapabilityArgumentProvider interface { |
| 37 | CapabilityArguments(capabilityID string) (CapabilityArgumentContract, bool) |
| 38 | } |
| 39 | |
| 40 | // ArgumentViolation is a value-free description of one invalid argument. It |
| 41 | // intentionally contains schema expectations, never the supplied value. |
| 42 | type ArgumentViolation struct { |
| 43 | Path string `json:"path"` |
| 44 | Keyword string `json:"keyword"` |
| 45 | Expected string `json:"expected"` |
| 46 | } |
| 47 | |
| 48 | // ArgumentValidationResult is the host-side result for one concrete target. |
| 49 | // Skipped is only safe for third-party MCP schemas that cannot be compiled; |
| 50 | // built-in schema failures are returned in CompileErr. |
| 51 | type ArgumentValidationResult struct { |
| 52 | Fingerprint string |
| 53 | Violations []ArgumentViolation |
| 54 | Skipped bool |
| 55 | CompileErr error |
| 56 | } |
| 57 | |
| 58 | type compiledArgumentSchema struct { |
| 59 | schema *jsonschema.Schema |
| 60 | err error |
| 61 | } |
| 62 | |
| 63 | var argumentSchemaCache sync.Map // map[string]compiledArgumentSchema |
| 64 | |
| 65 | // NormalizeArguments preserves the historical empty/null-to-object |
| 66 | // compatibility without guessing fields, coercing types, or rewriting values. |
| 67 | func NormalizeArguments(raw json.RawMessage) json.RawMessage { |
| 68 | trimmed := bytes.TrimSpace(raw) |
| 69 | if len(trimmed) == 0 || bytes.Equal(trimmed, []byte("null")) { |
| 70 | return json.RawMessage(`{}`) |
| 71 | } |
| 72 | return append(json.RawMessage(nil), trimmed...) |
| 73 | } |
| 74 | |
| 75 | // ValidateArguments validates args against the concrete tool's real schema and |
| 76 | // then runs its optional conditional validator. Compiled validators are shared |
| 77 | // by schema fingerprint and never resolve filesystem or network references. |
| 78 | func ValidateArguments(target Tool, raw json.RawMessage) ArgumentValidationResult { |
| 79 | if target == nil { |
| 80 | return ArgumentValidationResult{CompileErr: fmt.Errorf("argument validation target is nil")} |
| 81 | } |
| 82 | result := ValidateJSONSchemaValue(target.Schema(), NormalizeArguments(raw)) |
| 83 | if result.CompileErr != nil { |
| 84 | if _, thirdParty := target.(MCPMetadata); thirdParty { |
| 85 | result.Skipped = true |
| 86 | result.CompileErr = nil |
| 87 | return result |
| 88 | } |
| 89 | return result |
| 90 | } |
| 91 | |
| 92 | if conditional, ok := target.(ArgumentValidator); ok && len(result.Violations) < maxArgumentViolations { |
| 93 | remaining := maxArgumentViolations - len(result.Violations) |
| 94 | extra := conditional.ValidateArguments(NormalizeArguments(raw)) |
| 95 | if len(extra) > remaining { |
| 96 | extra = extra[:remaining] |
| 97 | } |
| 98 | result.Violations = append(result.Violations, extra...) |
| 99 | } |
| 100 | return result |
| 101 | } |
| 102 | |
| 103 | // ValidateJSONSchemaValue validates an arbitrary JSON value against a schema. |
| 104 | // It is used for third-party MCP outputSchema telemetry as well as argument |
| 105 | // contracts; callers decide whether a compile failure is fatal or advisory. |
| 106 | func ValidateJSONSchemaValue(schemaRaw, raw json.RawMessage) ArgumentValidationResult { |
| 107 | schemaRaw = bytes.TrimSpace(schemaRaw) |
| 108 | fingerprint := schemaFingerprint(schemaRaw) |
| 109 | result := ArgumentValidationResult{Fingerprint: fingerprint} |
| 110 | compiled := loadCompiledArgumentSchema(fingerprint, schemaRaw) |
| 111 | if compiled.err != nil { |
| 112 | result.CompileErr = compiled.err |
| 113 | return result |
| 114 | } |
| 115 | var value any |
| 116 | decoder := json.NewDecoder(bytes.NewReader(raw)) |
| 117 | decoder.UseNumber() |
| 118 | if err := decoder.Decode(&value); err != nil { |
| 119 | result.Violations = []ArgumentViolation{{Path: "", Keyword: "json", Expected: "one valid JSON object"}} |
| 120 | return result |
| 121 | } |
| 122 | if err := decoder.Decode(&struct{}{}); err != io.EOF { |
| 123 | result.Violations = []ArgumentViolation{{Path: "", Keyword: "json", Expected: "one valid JSON object"}} |
| 124 | return result |
| 125 | } |
| 126 | if err := compiled.schema.Validate(value); err != nil { |
| 127 | result.Violations = validationViolations(err) |
| 128 | } |
| 129 | return result |
| 130 | } |
| 131 | |
| 132 | func schemaFingerprint(raw []byte) string { |
| 133 | sum := sha256.Sum256(raw) |
| 134 | return hex.EncodeToString(sum[:]) |
| 135 | } |
| 136 | |
| 137 | // SchemaFingerprint returns a deterministic digest for diagnostics and cache |
| 138 | // invalidation without returning the schema itself. |
| 139 | func SchemaFingerprint(raw json.RawMessage) string { |
| 140 | return schemaFingerprint(bytes.TrimSpace(raw)) |
| 141 | } |
| 142 | |
| 143 | // InvalidateArgumentSchemas drops compiled validators after a catalog change. |
| 144 | func InvalidateArgumentSchemas(fingerprints []string) { |
| 145 | for _, fingerprint := range fingerprints { |
| 146 | if fingerprint != "" { |
| 147 | argumentSchemaCache.Delete(fingerprint) |
| 148 | } |
| 149 | } |
| 150 | } |
| 151 | |
| 152 | func loadCompiledArgumentSchema(fingerprint string, raw []byte) compiledArgumentSchema { |
| 153 | if cached, ok := argumentSchemaCache.Load(fingerprint); ok { |
| 154 | return cached.(compiledArgumentSchema) |
| 155 | } |
| 156 | compiled := compileArgumentSchema(fingerprint, raw) |
| 157 | actual, _ := argumentSchemaCache.LoadOrStore(fingerprint, compiled) |
| 158 | return actual.(compiledArgumentSchema) |
| 159 | } |
| 160 | |
| 161 | func compileArgumentSchema(fingerprint string, raw []byte) compiledArgumentSchema { |
| 162 | var doc any |
| 163 | decoder := json.NewDecoder(bytes.NewReader(raw)) |
| 164 | decoder.UseNumber() |
| 165 | if err := decoder.Decode(&doc); err != nil { |
| 166 | return compiledArgumentSchema{err: fmt.Errorf("invalid JSON schema: %w", err)} |
| 167 | } |
| 168 | if err := decoder.Decode(&struct{}{}); err != io.EOF { |
| 169 | return compiledArgumentSchema{err: fmt.Errorf("invalid JSON schema: multiple values")} |
| 170 | } |
| 171 | obj, ok := doc.(map[string]any) |
| 172 | if !ok { |
| 173 | return compiledArgumentSchema{err: fmt.Errorf("JSON schema root must be an object")} |
| 174 | } |
| 175 | _, explicitDialect := obj["$schema"] |
| 176 | compile := func(draft *jsonschema.Draft) (*jsonschema.Schema, error) { |
| 177 | compiler := jsonschema.NewCompiler() |
| 178 | compiler.UseLoader(nil) |
| 179 | compiler.DefaultDraft(draft) |
| 180 | resource := "urn:reasonix:argument-schema:" + fingerprint |
| 181 | if err := compiler.AddResource(resource, doc); err != nil { |
| 182 | return nil, err |
| 183 | } |
| 184 | return compiler.Compile(resource) |
| 185 | } |
| 186 | compiled, err := compile(jsonschema.Draft2020) |
| 187 | if err != nil && !explicitDialect { |
| 188 | compiled, err = compile(jsonschema.Draft7) |
| 189 | } |
| 190 | if err != nil { |
| 191 | return compiledArgumentSchema{err: fmt.Errorf("compile JSON schema: %w", err)} |
| 192 | } |
| 193 | return compiledArgumentSchema{schema: compiled} |
| 194 | } |
| 195 | |
| 196 | func validationViolations(err error) []ArgumentViolation { |
| 197 | var validationErr *jsonschema.ValidationError |
| 198 | if !errors.As(err, &validationErr) { |
| 199 | return []ArgumentViolation{{Path: "", Keyword: "schema", Expected: "arguments satisfying the tool schema"}} |
| 200 | } |
| 201 | leaves := make([]*jsonschema.ValidationError, 0, maxArgumentViolations) |
| 202 | collectValidationLeaves(validationErr, &leaves) |
| 203 | violations := make([]ArgumentViolation, 0, len(leaves)) |
| 204 | for _, leaf := range leaves { |
| 205 | keyword := "schema" |
| 206 | if path := leaf.ErrorKind.KeywordPath(); len(path) > 0 { |
| 207 | keyword = path[len(path)-1] |
| 208 | } |
| 209 | violations = append(violations, ArgumentViolation{ |
| 210 | Path: jsonPointer(leaf.InstanceLocation), |
| 211 | Keyword: keyword, |
| 212 | Expected: expectedForErrorKind(leaf.ErrorKind), |
| 213 | }) |
| 214 | } |
| 215 | if len(violations) == 0 { |
| 216 | violations = append(violations, ArgumentViolation{Path: "", Keyword: "schema", Expected: "arguments satisfying the tool schema"}) |
| 217 | } |
| 218 | return violations |
| 219 | } |
| 220 | |
| 221 | func collectValidationLeaves(err *jsonschema.ValidationError, out *[]*jsonschema.ValidationError) { |
| 222 | if err == nil || len(*out) >= maxArgumentViolations { |
| 223 | return |
| 224 | } |
| 225 | if len(err.Causes) == 0 { |
| 226 | *out = append(*out, err) |
| 227 | return |
| 228 | } |
| 229 | for _, cause := range err.Causes { |
| 230 | collectValidationLeaves(cause, out) |
| 231 | if len(*out) >= maxArgumentViolations { |
| 232 | return |
| 233 | } |
| 234 | } |
| 235 | } |
| 236 | |
| 237 | func expectedForErrorKind(errorKind jsonschema.ErrorKind) string { |
| 238 | switch k := errorKind.(type) { |
| 239 | case *kind.Type: |
| 240 | return strings.Join(k.Want, " or ") |
| 241 | case *kind.Required: |
| 242 | return "required properties: " + strings.Join(k.Missing, ", ") |
| 243 | case *kind.AdditionalProperties: |
| 244 | return "only declared properties; remove: " + strings.Join(k.Properties, ", ") |
| 245 | case *kind.Enum: |
| 246 | return "one of: " + boundedSchemaValues(k.Want) |
| 247 | case *kind.Const: |
| 248 | return "constant: " + boundedSchemaValue(k.Want) |
| 249 | case *kind.MinProperties: |
| 250 | return "at least " + strconv.Itoa(k.Want) + " properties" |
| 251 | case *kind.MaxProperties: |
| 252 | return "at most " + strconv.Itoa(k.Want) + " properties" |
| 253 | case *kind.MinItems: |
| 254 | return "at least " + strconv.Itoa(k.Want) + " items" |
| 255 | case *kind.MaxItems: |
| 256 | return "at most " + strconv.Itoa(k.Want) + " items" |
| 257 | default: |
| 258 | return "value satisfying " + lastKeyword(errorKind.KeywordPath()) |
| 259 | } |
| 260 | } |
| 261 | |
| 262 | func boundedSchemaValues(values []any) string { |
| 263 | parts := make([]string, 0, len(values)) |
| 264 | for _, value := range values { |
| 265 | parts = append(parts, boundedSchemaValue(value)) |
| 266 | if len(strings.Join(parts, ", ")) >= 512 { |
| 267 | break |
| 268 | } |
| 269 | } |
| 270 | return truncateASCII(strings.Join(parts, ", "), 512) |
| 271 | } |
| 272 | |
| 273 | func boundedSchemaValue(value any) string { |
| 274 | b, err := json.Marshal(value) |
| 275 | if err != nil { |
| 276 | return "declared schema value" |
| 277 | } |
| 278 | return truncateASCII(string(b), 256) |
| 279 | } |
| 280 | |
| 281 | func lastKeyword(path []string) string { |
| 282 | if len(path) == 0 { |
| 283 | return "the schema" |
| 284 | } |
| 285 | return path[len(path)-1] |
| 286 | } |
| 287 | |
| 288 | func jsonPointer(tokens []string) string { |
| 289 | var b strings.Builder |
| 290 | for _, token := range tokens { |
| 291 | b.WriteByte('/') |
| 292 | b.WriteString(strings.ReplaceAll(strings.ReplaceAll(token, "~", "~0"), "/", "~1")) |
| 293 | } |
| 294 | return b.String() |
| 295 | } |
| 296 | |
| 297 | func truncateASCII(value string, limit int) string { |
| 298 | if len(value) <= limit { |
| 299 | return value |
| 300 | } |
| 301 | if limit <= 3 { |
| 302 | return value[:limit] |
| 303 | } |
| 304 | return value[:limit-3] + "..." |
| 305 | } |
| 306 |