codeburn/docs/providers/vscode-cline-parser.md

vscode-cline-parser (Shared Helper)

Shared discovery and parsing for Cline and VS Code extensions descended from Cline.

• Source: src/providers/vscode-cline-parser.ts
• Loading: not a provider; imported by cline.ts, ibm-bob.ts, kilo-code.ts, and roo-code.ts.
• Test: none directly. Coverage comes from tests/providers/cline.test.ts, tests/providers/ibm-bob.test.ts, tests/providers/kilo-code.test.ts, and tests/providers/roo-code.test.ts.

What it does

Two responsibilities:

1. discoverClineTasks(extensionId) walks a base directory's tasks/ child and returns one source per task that has a ui_messages.json file (vscode-cline-parser.ts:25-50). Without an override directory it uses VS Code's globalStorage/<extensionId>/ path.
2. discoverClineTasksInBaseDirs(baseDirs) does the same for non-VS Code apps with compatible task storage, such as IBM Bob.
3. createClineParser reads each task's ui_messages.json and api_conversation_history.json, extracts model, tools, and token counts, and yields ParsedProviderCall objects.

Storage layout

Per task directory:

<baseDir>/tasks/<taskId>/
  ui_messages.json                # event stream
  api_conversation_history.json   # full prompt history with model tags

Model resolution

The model is extracted from api_conversation_history.json by searching user message content blocks for a <model>...</model> tag. Falls back to the provider-supplied auto model (cline-auto by default) if no tag is found.

Token extraction

From api_req_started entries inside ui_messages.json. Each such entry's text field is JSON-parsed; the parsed object holds tokensIn, tokensOut, cacheReads, cacheWrites, and (optionally) cost.

If cost is present, it is used directly. If not, calculateCost from src/models.ts computes it from tokens.

Deduplication

Per <providerName>:<taskId>:<index> where index is the position of the api_req_started entry within ui_messages.json.

Quirks

• Only the first user message is emitted as userMessage in the ParsedProviderCall. Subsequent user turns are accounted but not surfaced.
• The model regex looks inside content blocks, not at top-level fields. Some Cline-derivative extensions emit the model elsewhere; if you add support for one, branch on extension ID rather than rewriting the regex.

When fixing a bug here

1. A change here ripples to Cline, IBM Bob, KiloCode, and Roo Code. Run all four provider test files before opening a PR.
2. If you find that one of the extensions emits a different shape, branch on the extension ID parameter that the discovery function already takes; do not duplicate the parser.
3. If you add support for another Cline-family task store, register it as a thin wrapper file in the same shape as cline.ts, ibm-bob.ts, kilo-code.ts, and roo-code.ts.