A

Source Resolution And Runtime

Source resolution turns raw ingest inputs into typed source briefs. Runtime inspection then asks an adapter to load readable content for a source when CodeAlmanac has one available [@source_service]. The concept page is Source Material; this page covers the service boundary that prepares material for ingest.

Resolution#

SourcesService.resolve(...) wraps each input string in SourceAddress and passes it to resolve_address(...) [@source_service]. The resolver recognizes GitHub shorthand, Git ranges, Git diffs, transcript references, HTTP and HTTPS URLs, and local paths [@address_resolution].

The resolved SourceBrief contains a SourceRef, title, provenance kind, and prompt hint [@source_models]. This gives lifecycle prompts typed source facts instead of raw strings.

Runtime Inspection#

Runtime inspection is adapter-based. SourcesService.inspect_runtime(...) asks each configured runtime adapter whether it supports the source reference and returns the first adapter result [@source_service]. If no adapter supports the source, the service returns a skipped runtime snapshot titled with the unsupported reference identity [@source_service].

Ingest uses this boundary before it renders the writing prompt. It resolves the requested inputs, records preparation events, inspects runtime snapshots, and passes both briefs and snapshots into the operation prompt [@ingest_workflow].

Transcript Discovery#

Transcript discovery is a separate source path used by sync. The default discovery set has two adapters: Claude and Codex [@transcript_adapters]. The source model has the same two transcript app values, claude and codex, so there is no separate app identity for Codex app, Claude Desktop, Claude web, or editor-specific surfaces [@source_models].

The Codex adapter scans .codex/sessions under the configured home directory, reads the first JSONL lines for session metadata, and skips transcripts whose metadata marks thread_source as subagent [@codex_transcripts]. The Claude adapter scans .claude/projects under the configured home directory and skips paths that contain subagents [@claude_transcripts].

Sync does not ingest every discovered transcript. It matches each transcript cwd to a registered repository root, skips unregistered working directories, and skips transcripts older than the active sync window as inactive [@sync_evaluation]. That means a transcript can be discovered correctly but still not become ingest input for the current sync run.

Accepted input forms are summarized in Source Addresses. Page evidence uses a different contract, Frontmatter And Sources.