Rename CONTEXT.md to GLOSSARY.md

AGENTS.md

@@ -2,7 +2,8 @@
 
 Working on this repository rather than using the tool? See
 [CONTRIBUTING.md](CONTRIBUTING.md) for the tests, the link check and the
-conventions a change follows.
+conventions a change follows. The domain vocabulary is in
+[GLOSSARY.md](GLOSSARY.md).
 
 Instructions for any AI agent or automation tool driving logseq-cli from a
 shell. Nothing here is specific to one assistant: the CLI is a plain Python

docs/commands.md

@@ -101,7 +101,7 @@ graph.
 | `suggest-connections [--min-confidence N] [--min-shared N] [--max-suggestions N] [--focus PAGE]` | Topic-based connection suggestions. `--min-shared` (default 3) is the real filter — it sets how many topics two pages must share before the pair counts at all; `--min-confidence` (default 0.3) then scores it. `--max-suggestions` (default 10) caps the list and the remainder is reported as withheld; `total_found` counts what the graph held, not what survived the cap. `--focus` restricts to one page |
 | `get-page-stats --page NAME` | Page statistics (blocks, words, in/outbound links) |
 
-**analyze-journal-patterns output.** `--json` carries `tasks`: `{"open": N, "done": N, "cancelled": N, "done_rate": X}`, and every entry in `entries` carries `tasks: {"open": N, "done": N, "cancelled": N}`. A task is a block with a marker Logseq reads (the states are in `CONTEXT.md`, under Task State); text that looks like one, such as `TODO:`, a lower-case `todo` or `- [ ]`, is not counted. `done_rate` is `done / (open + done)` in percent with one decimal, `0` without open and done tasks; cancelled tasks are not part of it. Tasks are dated by the page their block is on. `get-todos --from/--to` also counts the days a block ref carried a task to. Habit tracking counts the text pattern `- [ ]` / `- [x]`, not Logseq's checkboxes.
+**analyze-journal-patterns output.** `--json` carries `tasks`: `{"open": N, "done": N, "cancelled": N, "done_rate": X}`, and every entry in `entries` carries `tasks: {"open": N, "done": N, "cancelled": N}`. A task is a block with a marker Logseq reads (the states are in `GLOSSARY.md`, under Task State); text that looks like one, such as `TODO:`, a lower-case `todo` or `- [ ]`, is not counted. `done_rate` is `done / (open + done)` in percent with one decimal, `0` without open and done tasks; cancelled tasks are not part of it. Tasks are dated by the page their block is on. `get-todos --from/--to` also counts the days a block ref carried a task to. Habit tracking counts the text pattern `- [ ]` / `- [x]`, not Logseq's checkboxes.
 
 **analyze-graph output.** `--json` carries `tasks`: `{"open": N, "done": N, "cancelled": N, "open_by_marker": {"<MARKER>": N, ...}}`. A task is a block with a marker Logseq reads; `TODO:`, a lower-case `todo`, `[ ]` at the start of a block and `TODO` followed directly by a line break are not. `open` is the same number as the `count` of `get-todos --no-follow-refs`. The text output has the line `Tasks: 12 open (2 DOING, 10 TODO), 40 done, 1 cancelled`, open markers in the order of the list, and a note on stderr when blocks start with `[ ]`: `Note: N blocks start with "[ ]", which Logseq shows as text, not as a task or checkbox.`
 
@@ -328,7 +328,7 @@ logseq-cli smart-query --request "offene aufgaben"   # German → finds open tas
 logseq-cli smart-query --request "open tasks"         # English → same result
 ```
 
-The pattern for open tasks finds every open marker Logseq reads (the markers and their three states are in `CONTEXT.md`, under Task State); the pattern for done tasks finds `DONE` only.
+The pattern for open tasks finds every open marker Logseq reads (the markers and their three states are in `GLOSSARY.md`, under Task State); the pattern for done tasks finds `DONE` only.
 
 Date formatting is locale-independent — weekday and month names are always English (as Logseq expects), regardless of system locale.