Ingest entire markdown documentation files verbatim #3

New issue

Closed

opened 2026-05-23 20:55:21 +00:00 by coilysiren · 1 comment

coilysiren commented

2026-05-23 20:55:21 +00:00

Owner

Originally filed by @coilysiren on 2026-05-22T06:46:20Z - https://github.com/coilysiren/repo-recall/issues/250

Idea

repo-recall should ingest every markdown documentation file in a repo verbatim. The whole file, every markdown file, read end to end. No summarization, no meta-analysis. Eat the whole thing.

Why now

Kai is doing a documentation rewrite that consolidates all docs into one strict, predictable structure (named sections, flat docs/*.md, size caps, validators). That strictness is exactly what makes verbatim markdown ingestion tractable. The docs are short, structured, and bounded, so storing them whole is cheap and high-signal.

Contrast with code

Existing repo-recall issues cover code meta-analysis, where the code body does not need to be stored verbatim. Documentation is the opposite case. Docs are the human-authored intent of the repo, and they are small enough to keep whole. Pull markdown verbatim, keep code as meta-analysis.

Proposed work

Walk every *.md in an ingested repo and store the full file text.
Make that text searchable alongside session transcripts.
Optional gating: only ingest once docs live in the consolidated structure, so repo-recall is not indexing scattered drafts. Decide whether to gate on a marker or just ingest all markdown and let structure improve over time.

_Originally filed by @coilysiren on 2026-05-22T06:46:20Z - [https://github.com/coilysiren/repo-recall/issues/250](https://github.com/coilysiren/repo-recall/issues/250)_ **Idea** repo-recall should ingest every markdown documentation file in a repo verbatim. The whole file, every markdown file, read end to end. No summarization, no meta-analysis. Eat the whole thing. **Why now** Kai is doing a documentation rewrite that consolidates all docs into one strict, predictable structure (named sections, flat `docs/*.md`, size caps, validators). That strictness is exactly what makes verbatim markdown ingestion tractable. The docs are short, structured, and bounded, so storing them whole is cheap and high-signal. **Contrast with code** Existing repo-recall issues cover code meta-analysis, where the code body does not need to be stored verbatim. Documentation is the opposite case. Docs are the human-authored intent of the repo, and they are small enough to keep whole. Pull markdown verbatim, keep code as meta-analysis. **Proposed work** - Walk every `*.md` in an ingested repo and store the full file text. - Make that text searchable alongside session transcripts. - Optional gating: only ingest once docs live in the consolidated structure, so repo-recall is not indexing scattered drafts. Decide whether to gate on a marker or just ingest all markdown and let structure improve over time.

coilysiren referenced this issue

2026-05-23 20:55:28 +00:00

recall_action_required: define an ordering contract #60

coilysiren referenced this issue

2026-05-23 20:55:30 +00:00

Native pattern for scheduled auto-mode agents driving the action-required queue #72

coilysiren commented

2026-05-30 05:44:01 +00:00

Author

Owner

Merged into #5 in the 2026-05-29 backlog burn-down. Verbatim markdown ingest overlaps docs-ingest issue Reopen if it should stand alone.

coilysiren closed this issue