1. 'Close diwan session' now uses TODO/DONE (assign_diwan_id inserts
'** TODO Close diwan session'), matching Reports. No headline anywhere
in diwan's own docs uses a [ ]/[X] checkbox now -- only :chat: list
items do, which is correct org usage (checklist items, not headline
states) and unaffected.
2. Rewrote PROTOCOL.md/TICK.md/DO.md/DESIGN.md to read as a single
coherent design rather than a changelog of how they got here. Removed:
contrastive 'not because X, but because Y' phrasing that only makes
sense if you know the prior rule, references to old tag names/formats
that no longer exist, a Claude-Code-vs-pi rationale that belongs in
DESIGN.md (if anywhere) not the agent-facing PROTOCOL.md, and 'no
:LINK: property' (never describe a convention by naming something it
doesn't have).
3. DESIGN.md got the heaviest pass since it had drifted from reality,
not just tone: it still said 'no flake.nix yet' (there is one, in
production use via samfelag), claimed several things 'not yet
re-verified live' that have since been verified multiple times
(tick's no-op gate, triage, the tag-consistency/escalation decision,
pi against real heading content via tick), and didn't mention
'diwan remote' or the actual flake.nix packaging shape anywhere.
Rewrote the verified-live claims and Known Gaps list to match current
reality, added flake.nix to the doc map and architecture section.
Live content migrated to match: diwan-dev-notes' decorate-home heading
('Close diwan session' checkbox -> TODO).
22 KiB
DESIGN.md — diwan
Audience: whoever (human or agent) picks up development of diwan itself next. If you're an agent dispatched by diwan to triage notes or do delegated work, you want PROTOCOL.md (given to you directly in your dispatch prompt) and the content repo's own AGENTS.md instead — this doc is one level up, about the tool, not about using the tool.
What this is
diwan is an AI layer over a plain org-mode "second brain" (notes, tasks, calendar, projects — normally edited by hand in Emacs/beorg, synced via Nextcloud). It periodically triages a capture inbox, keeps files tidy, and runs delegated research/work when asked, while treating the org files as the only real source of truth.
This repo is diwan the tool, and nothing else. It contains the CLI, a nix flake packaging it, and the generic procedure docs (PROTOCOL.md, TICK.md, DO.md) — no personal notes, no content-repo-specific convention docs. It's installed independently (on PATH, via this repo's own flake) and pointed at any notes repo. Marc's actual notes live in ~/nc/org-agenda (the real, continuously-synced data) and, for development/testing, in a throwaway sibling repo diwan-dev-notes (a git-tracked fixture copy, safe to break). Neither of those repos contains any of diwan's own code or docs.
The core loop (tick → triage/dispatch, do → delegated work, both via real herdr sessions, pi as the sole harness) is built, packaged as a flake, and deployed: samfelag (Marc's NixOS fleet config) has a module that installs the package and runs diwan tick on a systemd timer, already live on thingvellir. See "Known gaps" for what's still open.
Principles (why things are shaped this way)
These were distilled from how Marc already works and what he explicitly asked for; they're the standard to hold any future change to:
- Org files are ground truth, not a cache. If diwan vanished, the notes still work in Emacs/beorg. Nothing diwan needs lives outside
.orgfiles except the content repo's own.diwan/state.json, which is pure bookkeeping, not content. - Act first, review passively. No per-edit approval gate — every action is reported and trivially revertable instead.
- Mechanism over magic. Small, inspectable, deterministic steps over one opaque "agent did something" blob — Marc wants to be able to understand a failure well enough to jump in personally.
- Grow by accretion. Ship the smallest loop that works; add complexity only once the simpler version has been lived with. (Concretely: v0 has no custom CLI-level org-tag manipulation, no scheduler beyond a plain systemd timer, no kill-on-timeout logic — these are deliberately not yet built.)
- One harness, chosen deliberately:
pi. diwan dispatchespiexclusively, not a pluggable choice. It has no per-tool-call approval gate and no interactive first-run trust dialog — it runs unattended from the start, which is exactly diwan's own model. The procedure docs (TICK.md/DO.md) don't hardcode anything pi-specific into their wording anyway, as a matter of good practice. - Two symmetric channels, one substrate. Talk to diwan via the notes (
:chat:) or via the live herdr session — both write to the same git-tracked truth. - Mobile stays dumb. Nothing new to install on the phone.
- Tool and content are separate repos. diwan the tool (this repo) ships independent of any one person's notes; a content repo (e.g.
diwan-dev-notes, or Marc's real~/nc/org-agenda) only ever contains notes,AGENTS.md,diwan.org, and diwan's own tiny.diwan/runtime-bookkeeping directory (state.json,.lock) — never the CLI or the procedure docs. This is what makes the content repo distributable (synced via Nextcloud to a server, edited from beorg) without dragging a whole dev tool along with it, and what keeps diwan's own generic contract reviewable/versionable independent of any one person's data.
Architecture at a glance
Three layers, each independently simple:
- Git, in the content repo. Every automated write to the content repo is a commit. This is the revert mechanism (principle #2) — not a bespoke undo log. Commits are authored as
diwan <diwan@local>when diwan itself is the actor, or the user's own git identity whendiwan ticksnapshots whatever manual edits were lying around at the start of a run (so no tick ever operates on a dirty/mixed workspace, and "who changed this" is answerable fromgit log --authoralone). This repo (the tool) has its own, unrelated git history — just ordinary source control for diwan's own development. - herdr, a terminal-multiplexer-for-agents with a JSON-over-socket CLI (
herdr workspace/tab/pane/agent ...). diwan doesn't invent a session store: one herdr workspace labeleddiwanholds every tick run and every opendosession as a separate, independently addressable, persistent tab. See PROTOCOL.md's herdr section for the exact calls. - The
diwanCLI, a single dependency-free Python script, packaged by this repo'sflake.nix(pname = "diwan"): the script andPROTOCOL.md/TICK.md/DO.mdget installed together under$out/share/diwan/, and$out/bin/diwanis amakeWrapper-generated wrapper aroundpython3that setsDIWAN_DIRexplicitly to thatshare/diwandirectory and putsgitonPATH. It does only the mechanical, deterministic parts: git bookkeeping, enough org parsing to read/write aDIWAN_IDproperty and insert aClose diwan sessionTODO item, and herdr dispatch. It never triages, writes prose, or makes judgment calls — that's the dispatched agent's job, guided byTICK.md/DO.md. It distinguishes two roots explicitly (see the script's own docstring):DIWAN_DIR(where it and its docs are installed, overridable via theDIWAN_DIRenv var) andDIWAN_NOTES_DIR(whatever notes repo it's invoked against, found viagit rev-parse --show-toplevelfrom the current directory by default, overridable via theDIWAN_NOTES_DIRenv var — this is how a systemd service that doesn'tcdanywhere points it at the right repo).diwanmust be run from inside (or below) a content repo, or haveDIWAN_NOTES_DIRset — it has no content of its own to act on.
Everything else (what counts as "trivial", where an inbox item belongs, how to answer a chat question) is intentionally not code — it's prose in TICK.md/DO.md that the dispatched pi agent has to interpret. The CLI's job ends at "here's a prompt, here's a terminal"; everything judgment-shaped is left to the model.
Doc map
| File | Lives in | Audience | Contents |
|---|---|---|---|
DESIGN.md (this file) |
tool repo | diwan developers | why, architecture, current implementation status, known gaps |
flake.nix |
tool repo | nix, and anything consuming this flake (e.g. samfelag) |
packages the CLI + docs together; see "Architecture at a glance" |
PROTOCOL.md |
tool repo | any dispatched diwan agent (tick or do) | diwan's generic contract: tags, diwan.org, concurrency contract, revert model, herdr topology — deployment-independent |
TICK.md |
tool repo | a diwan tick agent specifically |
the 5-step tick procedure |
DO.md |
tool repo | a diwan do agent specifically |
scope, chat protocol, commit discipline, how to stop |
AGENTS.md |
content repo | any dispatched diwan agent (tick or do) | that deployment's specifics: layout, org conventions already in use — read on top of PROTOCOL.md, never restates it |
herdr --skill (external) |
— | anyone scripting herdr | authoritative herdr CLI reference — re-check this if herdr's version changes, since diwan's calls were verified against 0.9.5 specifically |
If you change a convention, update it in exactly one of these (whichever owns it) and make sure the others still only point at it rather than re-stating it — the duplication between PROTOCOL.md and any content repo's AGENTS.md is deliberately kept to zero; don't let that drift. Anything true of diwan regardless of which notes repo it's pointed at belongs in PROTOCOL.md (this repo); anything specific to one person's notes belongs in that content repo's own AGENTS.md, never here.
What diwan tick actually does
Two halves: a deterministic CLI half (cmd_tick in diwan) and a judgment-driven agent half (TICK.md).
CLI half, in order:
- If the content repo's working tree is dirty, commit everything under the user's own git identity (
chore: snapshot manual edits before tick) — this is what makes "no manual commit requirement" (PROTOCOL.md) actually true. - Load the content repo's
.diwan/state.json'slast_tick_commit. Diff that commit against currentHEAD(in the content repo), excluding.diwan/itself (so diwan's own bookkeeping commits never count as "something changed"). No prior tick → treat as "always changed." - If nothing changed and
--forcewasn't passed: print and exit. This is the only gate on whether a tick does anything at all. - Otherwise, pick a tab name
tick-<UTC YYYYMMDD-HHMMSS>(lowercase + digits + hyphens only — see PROTOCOL.md's herdr section for why that's not negotiable) and dispatch a fresh agent there, in a pane whose working directory is the content repo (DIWAN_NOTES_DIR), with a prompt that says: read<DIWAN_DIR>/PROTOCOL.mdandAGENTS.md(the latter resolved relative to the agent's own cwd, i.e. the content repo), follow<DIWAN_DIR>/TICK.md, here's the commit range. This dispatch is fire-and-forget — the CLI call returns as soon as the prompt is submitted, it does not wait for the tick to finish. - Record the new tab name in the content repo's
state.json'stick_history(kept to the lastMAX_TICK_TABS, default 3, override viaDIWAN_MAX_TICK_TABS), commit that under the diwan identity, and — if this pushed any older tick tabs out of the retention window — close those tabs for real viaherdr tab close.
Note what's not here: last_tick_commit is deliberately not updated by the CLI at dispatch time. It's updated by the dispatched agent itself, as the literal last step of TICK.md, only if everything else succeeded. If a tick agent crashes or is killed partway, the next tick sees the same "changed" range again rather than silently skipping work — correctness over tidiness.
Agent half (TICK.md), five steps, each its own commit:
- Triage
inbox.orgonly (other files changing is someone else's edit): events →esdeveniments.org, tasks → wherever they actually belong,:diwan:items get moved but not acted on yet, new topics only get a new file when nothing existing fits. - Tag consistency: any
:session:/:chat:heading without a:diwan:tag on itself or an ancestor gets:diwan:added directly to it. Then resolve:chat:ownership: chats under a:session:ancestor are left for that heading's own session; orphan chats get answered inline if trivial, or escalated (see step 3's three-way decision) if not. - Handle every
:diwan:-tagged heading without an open:session:: trivial ones get answered inline, no tag beyond:diwan:ever assigned. Non-trivial ones split two ways — self-contained within the heading's own subtree gets a:session:tag anddiwan do <file>:<line> --diwan-id <slug>run by the tick agent itself (the tick agent is the only thing that ever originates a freshDIWAN_ID); inherently cross-cutting work (spans multiple files/headings, can't be parallelized — archiving, bulk cleanup, moving things between files) is done by the tick agent directly, right there, no session at all, possibly spread across several tick runs if it needs Marc's input first. The same handling applies to any open item underdiwan.org'sTasksheading — diwan's own self-maintenance backlog, no tag needed there since the location already signals it. Already-open sessions (:session:tag +DIWAN_IDpresent,Close diwan sessionstillTODO) get their subtree's content left alone as always, but are checked for a nudge: if any commit in this tick's change range not authored by the diwan git identity touched a line inside that subtree, the tick agent re-runsdiwan do <file>:<line> --diwan-id <id>on the existing ID to wake it up; no nudge at all if nothing hand-authored touched the subtree. A session whoseClose diwan sessionis markedDONEgets retired: the tick agent closes itsDIWAN_ID's herdr tab (herdr tab close) and removes the:session:tag from the heading, leaving theDIWAN_IDproperty in place as a historical record. - Delete any already-
DONEReports subheadings first, then write one short, titledTODOsubheading per reportable thing this run did intodiwan.org's Reports log — the bar is deliberately broad (items moved, dispatches, closures, inline answers, findings, notes left for Marc to read), excluding only a run where nothing happened at all and a bare nudge (no information yet on whether it did anything) — each with the title itself linking to the one heading it's about, and a 1-2 line signed body. - Only if 1–4 all succeeded: update
state.json.last_tick_committo currentHEADand commit under the diwan identity.
Verified live, repeatedly, against the packaged flake build and diwan-dev-notes: the no-op gate (an immediate second tick after one that found nothing new correctly does nothing), triage, the tag-consistency/escalation decision in step 3 (confirmed doing the right thing for a cross-cutting archive task and for a stale leftover agent it correctly declined to nudge), and Reports writing. Not yet verified live: the current (broadened) Reports bar and TODO/DONE format, and the session-closure teardown path (nobody has checked off a Close diwan session and had a tick actually process it yet) — both designed and committed, but no real run has exercised either.
What diwan do actually does
Same split: cmd_do (CLI) decides identity/dispatch mechanics; DO.md governs what the agent does once running.
CLI half (cmd_do):
- Parse
target: if it matches<existing file>:<line number>(resolved against the content repo), it's heading mode; anything else is treated as an ad-hoc free-text prompt with no org heading involved (useful for manual testing, or tasks that genuinely don't belong in the notes). - Heading mode: read the headline at that line, check its
DIWAN_IDproperty.- Already has one → continuation. (If
--diwan-idwas also passed and disagrees with the stored one, that's an error, not a silent overwrite — a mismatch means something's confused and a human should look.) The existing ID is validated against herdr's naming rule before anything else happens, so a corrupted/hand-edited property fails with a clear message instead of a confusing herdr-side error later. - No
DIWAN_IDyet → fresh. Compute one:--diwan-idif given, else a slug of the heading text (slugify, capped and shaped to satisfy herdr's agent-name rule — see below), de-duplicated against every otherDIWAN_IDalready in the content repo's tree (unique_id). Write the property and insert** TODO Close diwan sessionas the first child (assign_diwan_id), in one commit under the diwan identity. This is the only place aDIWAN_IDis minted or a Close-session item is inserted — mechanical, not left to agent judgment, so it's always consistent.
- Already has one → continuation. (If
- Ad-hoc mode:
diwan_idis--diwan-idor a generatedadhoc-<8 hex chars>; always goes through the continuation dispatch path (see below) since there's no heading state to distinguish fresh from resumed. - Dispatch: fresh headings use
herdr_dispatch_fresh; everything else (continuation headings, ad-hoc) usesherdr_dispatch_continuation. Prints thediwan_ideither way — this is the command's only stdout contract, which is what letsTICK.mdstep 3 just rundiwan do ...as an ordinary shell command and know what session it created.
Continuation dispatch, concretely (herdr_dispatch_continuation): check herdr agent get <id> — alive → just agent prompt <id> <text> (fire-and-forget, no --wait). Not alive (the normal case between ticks; CLI agent processes exit, they don't idle forever) → look up the tab by label in the diwan workspace, start a fresh agent in its existing root pane, prompt it. Tab itself gone too (e.g. closed by hand) → falls through to the same fresh-tab path herdr_dispatch_fresh uses. Continuity across resumptions is carried by the heading's own content, not by agent conversation memory — a resumed do agent has no idea what it said last time except what's written in the org file. This is a real, current limitation, not an oversight to "fix" without thinking about it — it's a direct consequence of principle #1 (org is ground truth) and keeps the system's state in exactly one place.
Agent half (DO.md): stay inside your own heading's subtree; if something needs doing elsewhere, ask in :chat: rather than reaching out of scope; use the chat protocol for questions/FYIs; commit progress in small units under the diwan identity through the flock-wrapped recipe; never touch your own Close diwan session item — that's Marc's call, acted on by the next tick. One explicit exception to subtree-only: a do agent may append its own entry to diwan.org's Reports log once it finishes something worth telling Marc about, since that's its only channel to him outside of a tick run — it still never edits or deletes an existing entry there, only tick does that.
Verified live: real web research dispatched against a heading, writing a structured result into it and asking clarifying questions via a signed :chat: entry instead of guessing, committed under the diwan identity. Ad-hoc dispatch and continuation (reusing the same pane/tab/agent on a second call rather than creating a new one) are both confirmed working, including cold-starting in a content repo pi had never run in before (no trust-dialog block — unlike some other harnesses, pi doesn't have one). Not yet verified live: diwan do in heading mode (fresh DIWAN_ID minting) specifically with pi as the dispatched agent — the heading-mode mechanics have only been exercised end-to-end via a tick's own self-contained-work dispatch, not tested in isolation since the pi-only switch.
DIWAN_ID ⇄ herdr, precisely
A DIWAN_ID is a herdr tab label in the diwan workspace, and (while its agent process is alive) also that agent's name. This is why it must satisfy herdr's agent-name rule: ^[a-z][a-z0-9_-]{0,31}$. slugify() enforces this at mint time (lowercase, non-alnum → hyphen, truncated to 24 chars, forced to start with a letter) and cmd_do additionally validates every DIWAN_ID it's about to dispatch — including ones already stored in a heading — against that same regex before touching herdr, erroring out with a clear message rather than letting herdr reject it after the CLI has already created a tab (which used to leave an orphan tab behind; fixed by closing the just-created tab on any agent start failure).
Lifecycle: fresh (no property) → diwan do mints an ID, creates the tab+agent, heading now has DIWAN_ID + open (TODO) Close diwan session → active (any further tick/do on this heading is a continuation; tick explicitly leaves active sessions alone) → Marc marks Close diwan session DONE → next tick retires it: closes the herdr tab, removes the :session: tag, leaves the DIWAN_ID property in place as a historical record — retired (the heading can still be found later by that DIWAN_ID, but there's no live terminal and no tag pointing at one).
Known gaps / explicit non-goals (don't "fix" these without discussing)
- Tick scheduling beyond a plain systemd timer.
samfelag's module runsdiwan tickon a configurableOnCalendarcadence — that's it, no smarter scheduling, no overlap detection between runs. Fine for a personal single-host setup; revisit only if that stops being true. - Retired-session teardown is designed but not yet verified live.
TICK.mdstep 3 has the tick agent close the herdr tab (herdr tab close) and strip the:session:tag onceClose diwan sessionis markedDONE— but no real tick run has exercised this path yet. Worth a close look the first few times it fires: does the tab lookup byDIWAN_IDlabel reliably find the right tab, does tag-stripping leave:diwan:/other tags intact, does it behave sanely if the tab was already closed by hand. - The new (broadened) Reports bar and
TODO/DONEformat are unverified live. Designed and committed, but no real tick run has written a Reports entry under the current rules yet. - No real conversation continuity across resumption. Covered above — intentional for now, but worth revisiting if resumed
dosessions turn out to need more context than the heading text carries. diwan doin heading mode hasn't been exercised in isolation since the pi-only switch. FreshDIWAN_IDminting works as part of a tick's own dispatch decision, but hasn't been separately smoke-tested the way ad-hoc dispatch and continuation have.diwan remote <machine>is implemented but not verified against a real remote herdr server. Argument parsing, the no-herdr-on-PATHfallback, and a fast failure on an unknown--machinelabel are all confirmed; an actualherdr machine add+diwan remoteround trip against a live remote server is not.diwan.org'sConfigheading is a placeholder, not wired up. PROTOCOL.md documents an intended home for configuration values governing diwan's own behavior (tick cadence, tab retention, that kind of thing), but the CLI reads none of it — the env vars (DIWAN_DIR,DIWAN_NOTES_DIR,DIWAN_PI_APPROVE,DIWAN_MAX_TICK_TABS) remain the only real knobs. Don't assume anything in that heading does something until the CLI actually reads it.- Concurrency is scope discipline, not real locking, beyond the git-commit flock. Two
doagents editing truly disjoint subtrees should be fine; nothing stops two agents editing the same subtree if the scope rule is violated. There's no enforcement, only convention (PROTOCOL.md). - No automated tests. Everything described as "verified live" above was exercised by hand against a real herdr +
piinstall. There is no regression suite — a future session changingdiwanshould re-run at least the manual smoke tests (tick no-op gate, tick dispatch, do fresh dispatch, do continuation, a deliberately-invalidDIWAN_ID) before trusting a change.