runlist 0.99.0 → 0.99.2

package/README.md

@@ -71,8 +71,8 @@ controls are experimental; live retrieval depends on a compatible external
 provider. Local chat and push are deferred.
 
 <!-- runlist:release:start -->
-CLI and desktop release version: **0.99.0**.
-[Mac installer and release notes](https://github.com/reowens/runlist/releases/tag/desktop-v0.99.0).
+CLI and desktop release version: **0.99.2**.
+[Mac installer and release notes](https://github.com/reowens/runlist/releases/tag/desktop-v0.99.2).
 Normal product releases publish both components; an installed Mac app is updated with rollback after it has quit.
 <!-- runlist:release:end -->
 
@@ -1137,18 +1137,32 @@ are per machine in `~/.runlist/model.json`.
 `memoryMb` (what it holds, null when nothing is loaded), followed by the full
 detail.
 
-## One document's card
+## Read a plan or long document
 
 ```bash
-runlist show docs/plans/shelf.md          # title, status, next step, blockers, checklist, related, links
-runlist show shelf bins --json            # [{ path, title, status, nextStep, blockers, checklist,
-                                          #    related: [{ field, path, title, status }], links: [...] }]
+runlist show docs/plans/shelf.md          # card, outline and open boxes with line numbers
+runlist show shelf --section "Phase 2"    # one section (-s is the short form)
+runlist show shelf -s "Phase 2" --from 80 # continue at a source line
+runlist show shelf --all                 # every open box (default: first 30)
+runlist show shelf --card                # card only
+runlist show shelf --frontmatter         # frontmatter, with long values collapsed
+runlist show shelf bins --json           # cards with structured brief data in card.brief
 ```
 
 Read-only: nothing is claimed. `related` is what the frontmatter names
 (`related_plans`, `related_docs`, `supports_plans`, `parent_plan`, `runlist` and
 the configured reference fields), each with its title and status; `links` is
-every other document the body links to.
+every other document the body links to. The text brief caps each related/link
+list at eight entries; JSON retains the full lists.
+
+Phase headings show checklist progress as `[done/total]`; `→` marks the active
+phase. Contradictory phase markers are identified, and open boxes are grouped by
+phase. Headings and boxes inside code fences are ignored. Read the outline first,
+then one section at a time. Section reads stop at 500 lines or 60 KB and print
+the continuation command with `--from` when needed.
+
+`runlist use <doc>` also prints this brief for documents over 500 lines or 60 KB.
+Use `runlist use <doc> --full` when the whole body is needed.
 
 ## Failed commands
 

package/bin/runlist.mjs

@@ -1346,8 +1346,8 @@ Examples:
   runlist prompt list                    # singular alias for \`runlist prompts list\`
 
   runlist prompts show resume-foo        # peek without consuming (triage)
-  runlist prompts show --all             # peek the WHOLE pending queue in one call
-  runlist prompts show --all --limit 10  # ...capped
+  runlist prompts                       # survey names before selecting bodies
+  runlist prompts show --all --limit 10  # peek a capped batch
   runlist prompts show a b c             # peek several by name
   runlist prompts next --dry-run         # preview without consuming
   runlist prompts archive old-thing

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "runlist",
-  "version": "0.99.0",
+  "version": "0.99.2",
   "description": "CLI for managing markdown documents with YAML frontmatter — index, query, validate, graph, export, lifecycle, and AI summaries.",
   "type": "module",
   "license": "MIT",

package/plugins/runlist-codex/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "runlist-codex",
-  "version": "0.99.0",
+  "version": "0.99.2",
   "description": "Runlist workflow skill, session orientation, and tool guard for Codex",
   "author": {
     "name": "Reo Owens"

package/plugins/runlist-codex/skills/runlist/SKILL.md

@@ -20,7 +20,7 @@ Use this skill in repositories with `runlist.config.mjs` or `dotmd.config.mjs`.
 - **Subagent handoffs:** use `runlist baton --subagent <task-slug> @/tmp/draft.md` to save `subagent-<task-slug>`. Under a parent-owned plan, choose a distinct task slug that does not resolve to a plan; do not release the parent's claim. A child owning a dedicated plan can use `runlist baton --subagent <plan-file> @/tmp/draft.md`. Add `--parent resume-<session-name>` to link the child beneath its parent in `rl prompts`; the parent may be saved later. Parent-session batons keep the default `resume-*` naming. Matching prefixes are kept once; conflicting prefixes are refused. Existing prompts keep their names, and `--subagent --replace` refuses a parent-prefixed prompt.
 - **Draft outlines:** `runlist baton --template` prints the resume outline. The desktop Templates editor saves resume/subagent outlines in `runlist.templates.json`; repository baton instructions still apply.
 - **Child draft:** `runlist baton --subagent --template --parent resume-<name>` prints the saved child outline (built-in: Task, State, Next, Evidence and Boundaries). Fill the draft; its `parent_prompt` travels on save. A launcher can pass `RUNLIST_PARENT_PROMPT` into the child environment instead. Keep full records at durable repo-relative paths with commits and checked results; temporary scratch paths are supplementary. Completed work is a result for the parent to record, not another restart. Repair old links without replacing text: `runlist prompts link <child...> --parent resume-<name>`. Showing or consuming a parent surfaces its pending children without consuming them.
-- Consume a saved prompt with `runlist use <prompt-file>` (or bare `runlist use` for the oldest). This archives and claims before printing the body. Only an agent session consumes; a plain terminal gets the body plus the `runlist prompts use <path>` line, and the prompt stays pending. A prompt that ends with that line is still pending: run the command first. For read-only triage, use `runlist prompts show <file>` or `runlist prompts show --all`; never open pending prompts with a file reader or `cat`.
+- Consume a saved prompt with `runlist use <prompt-file>` (or bare `runlist use` for the oldest). This archives and claims before printing the body. Only an agent session consumes; a plain terminal gets the body plus the `runlist prompts use <path>` line, and the prompt stays pending. A prompt that ends with that line is still pending: run the command first. For read-only triage, survey names with `runlist prompts`, then peek selected bodies with `runlist prompts show a b c`; use `show --all` only with `--limit N`. Never open pending prompts with a file reader or `cat`.
 - Pending prompts are session-local. Commit the tracked plan or source files named by baton, not `docs/prompts/*.md`.
 - Create a document with `runlist new <type> <name> @<draft-file>`. A draft starting with frontmatter can supply scaffold fields; `type` is fixed by the command.
 - Record a decision with `runlist new decision <plan> --question "…" @<record-file>`; configured registers may also require `--answers`.