The cargo-fmt failure on this PR slipped to CI because the pre-commit
prek hooks were never installed in the working clone. The existing
cargo-fmt hook reformats in place but only when it runs. Add a
pre-push cargo-fmt-check hook (`cargo fmt --all --check`) that mirrors
CI's Dagger `check` step exactly, so an unformatted commit is blocked
locally before it can reach the runner — even if the pre-commit hook was
skipped or not installed. Filtered to .rs pushes so Rust-free pushes pay
nothing.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The project.resolve socket test had a node.create call over rustfmt's
line limit; CI's `cargo fmt --all --check` flagged it. Wrap it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
RPC clients (the hephaestus.nvim plugin, for its `:Heph version`
command) had no way to learn which hephd they are talking to — `health`
returns counts, not a version. Add a tiny `version` method returning
`{ version: heph_core::VERSION }`, the same `X.Y.Z (sha)` string the
binaries print for --version. No store access needed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Editing a task's canonical-context doc body previously meant looking up
its `canonical_context_id` (e.g. via `heph list --json`) and then
`heph node update <doc-id> --body`. Add a `heph context <task-id>`
command that resolves the canonical-context doc from the task's outgoing
links and:
* prints the body with no flag,
* `--body <text>` replaces it (`-` reads stdin, matching `node update`),
* `--append <text>` adds a blank-line-separated paragraph.
Errors clearly when the id has no canonical-context doc (e.g. a plain
doc node rather than a task). Purely a client-side CLI convenience — no
new RPC.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The `--project <name>` argument matched titles case-sensitively and
exactly, so `--project hephaestus` or `--project heph` failed against a
`Hephaestus` project. Make project-name resolution forgiving but
deterministic, via a tiered match in `resolve_project_id`:
1. exact (case-sensitive) — the historical behavior; always wins
2. case-insensitive exact — only when unambiguous
3. case-insensitive prefix — only when unambiguous
Ambiguous fuzzy matches resolve to None (callers report "no project
named X") rather than silently picking one. This single resolver already
backed `heph list --project` (via project_scope); route the CLI's
task/edit/promote/parent path through it too with a new `project.resolve`
RPC + `Store::resolve_project`, so every `--project` surface behaves the
same.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
heph-tui was the one daily-driver binary that did not answer --version.
Add the same clap `version = heph_core::VERSION` attribute that heph and
hephd already carry, so all three report `X.Y.Z (sha)` consistently.
Addresses the heph-tui half of the cross-binary --version task.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The build-time dirty check used `git status --porcelain`, which counts
untracked files — including cargo's own `.cargo-ok` marker in a
`cargo install --git` checkout — so a clean tagged build reported e.g.
`1.0.1 (bcab3c16b-dirty)`. Use `git diff --quiet HEAD` (tracked changes only).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the pre-release 'install from feature/v1-prototype' instructions with
`--tag v1.0.0` as the default, and document `--branch main` as the track-
unreleased-work alternative.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
heph-core gains a build.rs that captures the short git SHA and a
`heph_core::VERSION` const ("<crate-version> (<sha>)"); heph and hephd use it
for clap's --version. The crate version stays sourced from Cargo.toml.
release.yaml now bumps the workspace version into Cargo.toml + Cargo.lock on a
commit that only the tag points at, tags it manually, and pushes just the tag —
so cargo install --git --tag vX.Y.Z reports the real version while main stays at
0.0.0. The changelog commit moved ahead of the tag so the release includes its
own changelog.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Documents the desired end state before the code change (C1 docs-first):
- release.yaml bumps the workspace version into a commit only the tag points
at, keeping main at 0.0.0
- heph/hephd --version will report the release version plus the build SHA
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Dagger build_docs pipeline cloned Quartz from the default branch
unpinned. Quartz v5.0.0 restructured its config layout (.quartz/plugins,
../quartz imports), breaking the docs build against our existing
quartz.config.ts / quartz.layout.ts. Pin the clone to the last v4
release (v4.5.2) to restore known-good behavior.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The rust:1-bookworm CI image has no libdbus-1-dev, so libdbus-sys's
pkg-config build failed. Enable the dbus store's `vendored` feature to build
libdbus from bundled source (self-contained, the proven path the earlier
keyring-4 build used). `crypto-rust` keeps it OpenSSL-free; openssl-sys is only
an inert lock entry (the conditional `openssl?/vendored` reference), compiled
nowhere. Linux footprint unchanged at 235 crates; vendored libdbus is a
build-time C compile, not new crates.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
keyring 4's `keyring` meta-crate has no feature gating and compiles every
platform credential backend for the target. On Linux that dragged in the zbus
async stack, a redundant libdbus secret-service, the keyutils store, a
sqlite/zstd db-keystore, and OpenSSL (~290 crates in its subtree) — a real cost
on the RAM/CPU-constrained CI runner building with CARGO_BUILD_JOBS=1.
Depend on keyring-core (the API) + exactly one store crate per OS instead:
- macOS -> apple-native-keyring-store (keychain feature)
- Linux -> dbus-secret-service-keyring-store (crypto-rust; libdbus, no openssl)
oauth.rs registers the per-target store as the keyring-core default itself
(replacing keyring::use_native_store). Runtime behavior is unchanged (tokens
still go to the macOS Keychain / Linux Secret Service).
hephd's Linux dependency graph: 401 -> 235 crates (-166), dropping the zbus
ecosystem and two C builds (zstd-sys, plus the redundant secret-service path).
macOS builds + the full suite are green here (228 tests, clippy -D warnings,
fmt, prek); the Linux store path is CI-verified (API confirmed from source).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The cheap "seam" that keeps the single-owner hub from calcifying, ahead of
the gilbert -> indri bring-up:
- Replace the single-tenant gate `Store::authorize_owner_sub(sub) -> bool`
with `resolve_owner(sub) -> Option<owner_id>`. The hub auth middleware now
resolves the token's identity to the owner it may act as (Some -> allow,
None -> 403). Behavior is identical for the single-owner hub (claim-on-first;
strangers still 403), but the contract no longer assumes one global owner, so
serving N owners later is additive, not a rewrite. The per-request owner is
marked at the exact line where downstream scoping wires through.
- New how-to docs/how-to/set-up-sync-hub.md: stand up the hub and connect an
existing device as an offline-capable spoke, the data-safe way (Path A: the
hub adopts the device's identity rather than rewriting the device).
The decision (cheap seam now, defer full multi-tenancy + adoption rewrite) is
recorded in the Adoption + multi-tenant task's context doc. Two enabler gaps
the how-to surfaced (heph daemon hub/spoke service flags; Path-A seeding tool)
are filed as Hephaestus tasks.
Green: 228 tests, clippy -D warnings + fmt + prek clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The recent CI failures ("Cannot connect to the Docker daemon") are the
DinD build engine being OOM-killed mid-compile, not flakiness. Serialize
both the check and test_nvim cargo builds to jobs=1. Temporary mitigation
pending more host RAM.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The recent CI failures ("Cannot connect to the Docker daemon") are the
DinD build engine being OOM-killed mid-compile, not flakiness. Even 4
parallel rustc invocations spike memory too high on the runner; serialize
to jobs=1. Slower but survives. Temporary mitigation pending more host RAM.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Neovim plugin now lives at eblume/hephaestus.nvim (plugin at the repo
root). Remove heph.nvim/ from the monorepo and the build/test wiring that
referenced it:
- Dagger: drop the test_nvim function + the pinned-Neovim NVIM_VERSION
- build.yaml: drop the `dagger call test-nvim` step
- drop the mise run test-nvim task and .stylua.toml + the stylua prek hook
(no Lua remains in the monorepo)
- install-heph.md: install via a plain lazy.nvim spec pointing at the
plugin repo over SSH (no more local-dir checkout hack)
- README / AGENTS / heph-nvim.md: note the surface lives in its own repo
The CLI/TUI -> nvim integration is unchanged (they shell out to `nvim`
expecting the heph plugin installed). The v1-prototype tech-spec §14 build
record and prior changelog fragments are left as frozen history.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Bump all external crates to latest stable ahead of v1.0.0 (tech-spec §14.9):
keyring 3→4 (the keyring_core split + register-the-native-store model),
rusqlite 0.32→0.40, ratatui 0.29→0.30, rrule 0.13→0.14, yrs 0.26→0.27,
plus a cargo update for semver-compatible bumps.
keyring 4 moves Entry/Error into keyring_core and requires a credential
store to be registered before use; KeyringTokenStore now registers the
OS-native store once (lazily, via Once) and uses keyring_core types.
not_keyutils=true so Linux prefers Secret Service over the logout-wiped
kernel keyutils store.
Drop the fs4 dependency in favor of std::fs::File::try_lock (stable since
Rust 1.89); raise workspace MSRV 1.85→1.89. Remove the orphaned
.forgejo/scripts/build hook — CI invokes Dagger directly.
Green: 228 Rust tests + 25 heph.nvim headless e2e, clippy -D warnings + fmt.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
CI on main failed on a clippy::single_match lint in the heph-tui sidebar
key handler. Rewrite as `if let`. Also add a `cargo clippy -D warnings`
prek hook mirroring cargo-fmt, so lints are caught locally before CI
(prek previously only ran cargo fmt).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Run `cargo fmt --all` in place over the workspace on any staged .rs change,
matching the repo's other in-place formatters (ruff-format, stylua, shfmt).
Unformatted Rust now fails the commit locally (it reformats + reports
"files were modified"), so the fmt-dirty commits that slipped through this
session can't recur. CI still enforces `cargo fmt --check` via Dagger as the
backstop. Verified: passes clean, catches + fixes a deviation.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`heph list --project <name>` lists a project's outstanding tasks by name
(subtree-expanded, resolved server-side via a new project.scope path that
reuses the view machinery; errors on unknown names). `--json` prints raw
rows — node_id, canonical_context_id, attention/state/do_date/late_on/
recurrence/project_id — for scripting and agents. Store::project_scope on
the trait + LocalStore + RemoteStore; new project.scope RPC and a flattened
ListParams so `list` accepts an optional project name. Test covers
resolve-by-name + unknown-name error.
AGENTS.md thinned to tight command/pattern sections: dropped the historical
parity narrative and the verbose roadmap section; added a "Working state"
section documenting `heph list --project Hephaestus [--json]` as the way to
inspect heph's self-hosted roadmap.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Run rustfmt over files landed in earlier commits this session that weren't
fmt-checked (heph-quickadd, the heph-tui undo/move wave, the hephd quickadd
supervisor). Pure formatting (struct/if-else expansion, line wrapping); no
behavior change. Restores `cargo fmt --check` clean for CI.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
v1 reached Todoist feature-parity, so remaining/future work is now tracked
in heph itself — tasks in the Hephaestus project (heph view ondeck) — not in
a doc. Renamed docs/reference/tech-spec.md -> v1-prototype-tech-spec.md and
rewrote all 27 [[tech-spec]] wiki-links + README/changelog path refs (docs
checks green). Retitled + bannered the spec as a historical v1 build record
and froze its §14 tracker. AGENTS.md gains a "Planning future work" section
(capture via `heph task --project Hephaestus`, triage in heph-tui On Deck);
README status reflects parity + the three daily-driver surfaces. The design
doc remains the living rationale.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Project delete previously tombstoned only the project node, leaving its
tasks with a live in-project link to a dead project — orphaned (not in the
Inbox, unbrowsable, blank project) rather than unfiled as intended. New
atomic Store::delete_project tombstones every in-project link to the project
(tasks fall to the Inbox), then tombstones the project node; tasks are never
deleted. Exposed as the project.delete RPC (LocalStore + RemoteStore); the
heph-tui sidebar `D` now routes through it. Core test asserts the task
survives and becomes unfiled; the project node is tombstoned.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Triage gestures (x/d/S/A/b/e/m/D) now fire only when the task pane is
focused, so a stray key while in the sidebar can't drop or delete a task;
hints are focus-aware. `u` undoes the last triage action (drop/done/skip/
attention/move) and Ctrl-z redoes it, restoring from a pre-action snapshot
(multi-level, cap 200; tombstone-delete excluded — no restore path yet). `D`
in the sidebar deletes the highlighted project (y/N), unfiling its tasks to
the Inbox. The sidebar's Projects section now rebuilds after create/delete,
so a new project appears without a restart. Tests cover undo/redo, the
empty-undo no-op, and sidebar project delete.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The `m` move-to-project overlay is now a filterable picker: a prompt line
narrows the project list by fuzzy subsequence match as you type (↑/↓ or
Ctrl-n/p move, Enter selects, Esc cancels), so there's no scrolling a long
list. When the filter names no existing project, a "+ New project" row
creates it and files the task there in one step (Backend::create_project →
node.create). Tests cover fuzzy narrowing and the create-then-file flow.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A sixth built-in filter view (listed below On Deck) showing every
outstanding task filed under no project — the capture inbox to triage. New
ListFilter.unfiled predicate kept purely in matches(); the inbox ViewSpec is
un-gated (no attention/do-date filter) so nothing hides from triage. Surfaces
automatically in the heph-tui sidebar and `heph view`. Tests cover the
predicate and the view spec; navigation tests updated for the 6th view.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A macOS global quick-capture popover to retire Todoist. New `heph-quickadd`
crate: an always-warm eframe/egui agent that registers ⌘' (global-hotkey,
Carbon — no Accessibility permission) and toggles a hidden, pre-created
window visible + focused on press — never spawning on the keypress, so it's
a muscle reflex. A single field live-parses Todoist-style inline syntax via
the shared parser; chips show ⚑ attention · 📁 project · ⏰ do-date · ↻
recurrence as you type. Enter saves optimistically (hide now, task.create
on a bg thread; a failed RPC re-shows with the text restored). #project
autocomplete (Tab/↑↓/click; focus-locked so Tab completes instead of
traversing). Example hints rotate, fading in only after ~2s idle.
Parser lifted from heph-tui into hephd's lib (hephd::quickadd) so the TUI
and the popover share one parser. hephd supervises the helper as a child in
local mode on macOS (opt-in HEPH_QUICKADD=1, set by the installed launchd
plist) — one service to manage, no second launch agent; the helper
self-exits when orphaned so killing hephd leaves nothing behind.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
A one-cell gap between the ⚑ attention flag and the project-colored ●
bullet; the blank-flag case already reserves the same width, so rows stay
aligned whether or not a flag is present.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- `<Enter>` now opens the selected task's context doc in nvim (App::enter:
from the sidebar it drills into the task list first); the `o` binding is
retired. Hint line updated.
- BUILTIN_VIEWS reordered to the owner's preference — Top of Mind, Tasks,
Work Tasks, Chores, On Deck — which drives the TUI sidebar and
`heph view`. Tests that walked to On Deck by a fixed offset now seek it
by title.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The picker listed every node, so each task showed up twice (itself + its
same-titled canonical-context doc) plus tag/log noise — and the new
preview made the duplicates look identical. New `Store::list_linkable_nodes`
/ `node.linkable` returns non-tombstoned nodes minus `tag`s and the docs
that are a task's canonical-context or log attachment (you link the task,
not its body). The Telescope picker now sources from it.
Tests: a socket test (5 nodes → 2 linkable: task + standalone doc).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add a buffer previewer that shows the highlighted node's body as you
filter — for a task (no body of its own), it previews the canonical-
context doc instead. RPC-fed (heph nodes aren't files, unlike
obsidian.nvim's notes), pcall-guarded so a fetch miss just blanks the
preview. Reuses the global fzy_native sorter.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Replace the "guess a query, then filter a frozen result set" flow with a
live fuzzy filter over every node when Telescope is present: type to
narrow `node.list`, <CR> inserts the highlighted node, <C-x> creates a
doc named the current prompt (a miss flows straight into making it). The
search-then-`vim.ui.select` two-step stays as the no-Telescope fallback.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`wikilink::to_ids` rewrites name-addressed links to the canonical id
(id-first resolve: an already-id target is left alone, a name → its id
with any label preserved). `Store::migrate_wikilinks_to_ids` runs it over
every body and re-saves through update_node (which collapses + materializes
by id); idempotent. Surfaced as the `migrate.wikilinks` RPC + RemoteStore
forward + the `heph migrate-links` CLI command (not auto-run — the owner
runs it once per store).
Name-resolution + the canonical-context hack stay for now so legacy links
keep resolving pre-migration; retiring them is a later tidy. Tests:
to_ids unit + a heph-core migrate integration (rewrite + materialize +
idempotency).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
`conceal.lua` hides the `[[id|` prefix and `]]` suffix with conceal
extmarks (refreshed on edit), leaving the label as a styled `HephLink`;
`conceallevel=2` + empty `concealcursor` reveal the raw `[[id|Name]]` on
the cursor's line so it stays editable. The `[[` picker now inserts the
labelled `[[id|Name]]` form (readable + conceal-ready; collapses to bare
on save). e2e asserts the conceal extmarks + conceallevel.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Keep canonical `[[NODEID]]` links readable without storing names. New pure
`heph-core::wikilink` (injected id→title): `expand` turns a bare `[[id]]`
into `[[id|Current Name]]`, `collapse` turns a name-matching `[[id|text]]`
back to bare (a custom label is preserved as an override).
- `node.get` expands on every read (nvim buffer + TUI preview both
readable), then prepends frontmatter when asked.
- `update_node` strips frontmatter, then collapses links, then CRDT-diffs
— so neither projection ever persists and an unchanged read→write is a
no-op to the bare id.
Tests: wikilink unit (expand/collapse/round-trip), a heph-core collapse
+ materialize-by-id integration test, and a socket expand→collapse
round-trip. `heph export` still emits raw ids (later polish).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Backend: `links::resolve_id` now checks for an exact live node id before
alias/title, so a canonical `[[NODEID]]` link resolves to its node and
can't be shadowed by a like-named node. Legacy `[[Name]]` links still
resolve by name (until the migration), so this is additive.
heph.nvim: `link.insert` (bound to insert-mode `[[` and `:Heph link`)
searches via the `search` RPC and inserts `[[NODEID]]`, with a "+ Create
new doc" entry; `<CR>` follow resolves the id directly. e2e covers
search→insert→materialize and the create path.
Remaining (§8.4): read-expansion/conceal display + the one-time
[[Title]]→[[NODEID]] migration (then retire name-resolution + the hack).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
