| name | olwb-benefits |
| description | Create, configure, and troubleshoot olwb's benefits features: destinations (/dest), sending (/send), and the notes → agent-work issues pipeline (/issues). USE FOR: adding or wiring a destination or agent preset; diagnosing stale CLI sessions, missing inbox responses, tui terminal problems, or rejected issue drafts; setting up the issues pipeline for a target repo. DO NOT USE FOR: core capture/liner/session usage (see README.md) or modifying olwb's Lua source. |
| compatibility | Designed for Claude Code and similar coding agents helping a user of the olwb micro plugin. |
| metadata | {"argument-hint":"What to set up or the symptom, e.g. \"add an opencode review dest\" or \"responses not landing in inbox\".","user-invocable":"true"} |
olwb-benefits
Helps a user create, configure, and debug olwb's sending and issues
features. Authoritative references, in order: docs/dest.md,
docs/send.md, docs/issues.md, then help/olwb.md for exact command
syntax. Read the relevant doc before answering; don't work from memory.
All commands below are typed in olwb's one line. Every one is also
reachable as > olwb <verb> [args] without the leading /.
Where state lives
Nearly every question here is answerable from two files under the data
directory (default $XDG_DATA_HOME/olwb, overridable via olwb.datadir):
state.json — destinations (command, into liner, kind), stored
dest×liner CLI session ids, unread counts, active liner/filter.
issues/ — issue drafts: <id>.sh (the filing script), <id>.json
(manifest with sources/repo/status), <id>.raw.txt (only when a model
response was rejected).
When troubleshooting, read state.json first; it is the ground truth the
overlays (/dest, /issues list) render from.
Creation recipes
Destination piping to a plain command:
/dest add jrnl tee -a ~/journal.md
Agent preset (adapter kind inferred from the first word; responses to a
dedicated liner, conversation resumed per liner):
/dest add oc-review opencode run --agent review
/dest into oc-review reviews
Issues pipeline for a repo (always set the local path when one exists —
it enriches drafts with the repo's AGENTS.md):
/issues repo add olwb TGPSKI/olwb-micro-plugin ~/git/TGPSKI/olwb-micro-plugin
Troubleshooting
| Symptom | Likely cause | Fix |
|---|
| Response never lands anywhere | destination's into is -, or the command failed silently | /dest overlay to check into; run the command manually with a small stdin payload |
| Response lands but no badge | it landed in the active liner — badges only fire for non-active liners | working as designed; Alt-i is only for the inbox case |
| Raw JSON appears as the message | adapter kind missing (wrapper script, renamed binary) | /dest kind <name> claude|codex|opencode |
| Conversation doesn't continue across sends | session stored per destination and liner — a different liner is a different conversation | /dest session list to see mappings |
| Send hangs or repeats fresh | stale session id (CLI purged it); olwb retries fresh exactly once per send | /dest session clear <name> from the affected liner |
tui opens nothing | no terminal auto-detected | /set termcmd "<terminal cmd>" (stored in state.json, not micro settings) |
| Draft rejected | model response wasn't strict JSON per the template | read issues/<id>.raw.txt; if the model fenced or chatted, tighten <datadir>/issues-prompt.md or switch /issues model |
/issues file refuses | draft already filed — refiling is deliberately blocked | check /issues list status; re-draft if the sources changed |
| Drafts cite no repo context | repo registered without a local path | /issues repo rm + re-add with the path |
Guidance defaults
- Prefer editing seeded presets over adding near-duplicates; they never
re-seed, so changes stick.
- For a new agent workflow, wire the destination and its
into liner
before the first send — retroactively moving responses means manual
copying.
- The
<id>.sh review step in the issues pipeline is load-bearing: never
advise skipping the read-before-file step, and never file on the user's
behalf without showing the script.