Two privsep follow-ups, both in the hive-c0re service:
1. The migration 'chown -R hive-core /var/lib/hyperhive' (which only
started running once ExecStartPre stopped failing 203/EXEC) stomped every
agent's bind-mounted creds — agents/<name>/{claude,state,harness,config}
are owned by the per-agent/manager users — logging all agents out with no
way back in. Scope it to everything *except* agents/, plus the agents/
dir node itself so c0re can still create new per-agent subdirs. Each
container's hive-agent-user-migrate activation chowns the contents back.
2. nix (prebuild 'nix build', flake-check, meta eval in c0re; nixos-container
update->nix in priv) writes its cache under $HOME/.cache. Both services
run as users with no home -> HOME=/var/empty (unwritable) -> Lix cache
init fails, rebuilds error out. Set HOME to each service's StateDirectory
(adding one for hive-priv).
hive-priv had no PATH, so the nixos-container it runs for every container
op (incl. lifecycle::list) failed ENOENT. build_all swallows that into an
empty list — the 'no managed containers' symptom. Give the helper the
minimal set nixos-container shells out to (nixos-container, nix, util-linux,
e2fsprogs) on top of the systemd/coreutils/findutils already in the unit
PATH, rather than the whole system profile.
Also route forge/matrix is_present() through priv_client::list_containers
instead of spawning nixos-container directly from unprivileged hive-c0re.
The state-migration chown invoked ${pkgs.coreutils}/bin/sh, which does not
exist (coreutils ships chown, not sh), so ExecStartPre exited 203/EXEC on
every boot. hive-c0re hit its start limit and never came up, so the gateway
returned 502 Bad Gateway. Call chown directly with the +- prefix (run as
root, tolerate failure) instead of going through a shell for '|| true'.
The privsep drop to the hive-core user left four journalctl -M <container>
call sites shelling out directly. -M enters the container namespace via the
machine bus, which needs root, so all container-journal reads failed with
Permission denied. Add a ReadContainerJournal verb to hive-priv and route
dashboard get_journal, manager get_logs, the rebuild-failure journal tail,
and the agent host-journal -M path through it. Host-journal reads (no -M)
stay direct via systemd-journal group membership.
- Wrap table in .schedules-table-wrap (overflow-x: auto) so it
never exceeds the page width on narrow viewports
- Add .schedules-table-next-col (width: 8em) and
.schedules-table-every-col (width: 7em) with white-space: nowrap
to stop those columns from stealing space from the body column
- Align visual style with the perm page tables: unified border on
all cells (was: border-top row separator only), bg-elev thead,
font-size 0.82em (was 0.9em), min-width: 100% on the table itself
Drop the 'h-sock :8421' meta span from the agent card head.
ContainerView.container is now #[serde(skip)] since no frontend
code reads it; the field stays in Rust for internal use.
c.port remains serialized — used for fallback agent UI URLs.
hive_sh4re::Request uses #[serde(tag = "cmd")] so Wake serialises as
{"cmd":"wake",...}. The wake.rs in hive-matrix-mcp was sending
{"kind":"wake",...} — the harness could not deserialise the message
and silently discarded every incoming matrix event.
Also: drain the server's response line instead of just shutting the
write half. Without the drain the server got ECONNRESET writing back,
which logged a spurious error even though the wake itself was now
processed.
Root cause of the matrix notification blackhole reported in #1087.
add QueueKind::PermChange — dashboard tool-group and capability
handlers no longer write the shared JSON files inline. instead they
enqueue a PermChange entry; the FIFO worker applies the file write
then calls rebuild_agent so the updated env var takes effect.
concurrent batch-apply actions for different agents previously raced
on tool-groups.json / capabilities.json (last write wins, earlier
change silently dropped). serialising through the queue prevents this.
dedup check extended with perm-type discriminant so tool-groups and
capabilities changes for the same agent are kept as distinct entries
and never collapse into one slot.
- hive-sh4re: ToolGroup::description() and Capability::description() return
short human-readable strings for each variant
- hive-c0re: ToolGroupsSnapshot and CapabilitiesSnapshot now include a
`descriptions` map (name → description); get_capabilities now iterates
Capability::ALL instead of hardcoding the list
- tabs.js: renderToolGroups + renderCapabilities use descriptions[name] as
the column header title attribute (native browser tooltip on hover)
Argus flagged that since PR #1054 the nginx reload path is
`gateway_systemctl(["reload", "nginx"])` = `systemctl -M hive-gateway reload nginx`.
Replace the stale `systemd-run --machine=hive-gateway` reference.
Four modules were absent: capabilities.rs, build_logs.rs,
gateway_nginx.rs, priv_client.rs. Added at their natural
positions alongside related modules.
Remove web_tools from the C4P4B1L1T13S table in dashboard.md and add
it to the T00L GR0UPS section with a note that it gates Claude built-ins
rather than MCP tools. Fix turn-loop.md to say 'tool-group-gated' and
'web_tools tool group' throughout.
The hardcoded vec was a maintenance hazard — any new capability added
to Capability::ALL would silently be omitted from the permissions UI
column list until get_capabilities was manually updated.
Now both the GET and POST handlers derive their known-capability lists
from the same Capability::ALL source of truth.
Capabilities were added with the permissions tab but conventions.md only
documented tool groups. Adds a full Capabilities section covering: the
known capabilities table, config storage path, HIVE_CAPABILITIES env var
injection, runtime resolution, operator-only grant constraint, and the
pattern for adding a new capability.
/logs.html was undocumented — only mentioned in passing in API
endpoint lists. Adds a proper section covering BUILD / AGENT /
SYSTEM sub-tabs, their API endpoints, and behaviour.
The meta/ section only described the flake. Added entries for
topology.json, tool-groups.json, and capabilities.json — the three
system-level config files also committed there — with their writers,
readers, and injected env vars.
The deny list takes precedence over --allowedTools, so granting the
web_tools group had no effect — both tools were silently blocked.
WebFetch/WebSearch are now gated solely via --tools (builtin_tools_arg):
agents without the web_tools tool group don't get them in --tools so
claude never sees them; agents with the group get them in both --tools
and --allowedTools.
Replaces the earlier capability-based approach (closed#1069) with a
ToolGroup — capabilities are for privileged system access, web egress is
a tool permission.
Add ToolGroup::WebTools to hive-sh4re:
- tools() returns &[] (no MCP tools gated)
- builtin_tools() returns &["WebFetch", "WebSearch"] — new method on ToolGroup
- Present in ALL and as_str() → "web_tools"
In hive-ag3nt/mcp.rs:
- allowed_tools_arg() now iterates group.builtin_tools() to prepend any
group-gated built-ins alongside the base ALLOWED_BUILTIN_TOOLS set
- builtin_tools_arg_for_flavor(flavor) replaces builtin_tools_arg() so
the flavor-correct effective groups are used when building --tools
- builtin_tools_arg() kept as a flavor=Agent convenience alias
- turn.rs updated to call builtin_tools_arg_for_flavor(files.flavor) so
manager sessions also see web tools when web_tools is in their groups
The dashboard T00L GR0UPS table gains a web_tools column automatically
(ToolGroup::ALL drives the columns).
Tuwunel 1.6.1 appends " 💕" to every new user's display name via the
new_user_displayname_suffix config option. Set it to empty string so
agent and operator matrix accounts get clean names.
Existing users (atlas, mara, root) need a one-time display name reset.
Atlas's was already corrected via the client API.
stats.html had a <style> block with all its page-specific CSS inline.
Move to agent.css under a '--- /stats page ---' section, renaming
the generic .grid/.card/.empty-note classes to .stats-grid/.stats-card/
.stats-empty-note to avoid future collisions with other agent pages.
- rename cap-cap-col → cap-col; add CSS rules for .cap-col and .cap-save-col
- drop zero-width-space replace in capability header cells (nowrap makes it a no-op)
- add Capability::ALL to hive-sh4re; validate incoming cap strings in post_capabilities
Add a new P3RM1SS10NS tab to the dashboard that consolidates all
per-agent permission configuration:
Backend:
- GET /api/capabilities returns { caps: [...], assignments: {...} }
driven by Capability::ALL variants (manage_root_agent,
read_host_journal, query_agent_state)
- POST /api/capabilities/{agent} writes capabilities.json and queues
a rebuild so HIVE_CAPABILITIES takes effect
Frontend:
- New 'permissions' entry in TABS, placed after 'system'
- P3RM1SS10NS tab pane with two sections:
C4P4B1L1T13S — agents × capabilities checkbox matrix (.cap-*)
T00L GR0UPS — agents × tool-groups checkbox matrix (.tg-*) moved
from SYST3M tab
- activateTab('permissions') fetches both tables; neither has an SSE
channel so they re-fetch on each activation to stay fresh
- CSS for .cap-* mirrors the .tg-* layout (scrollable, Catppuccin)
hive-ci-register.service now runs unconditionally on every boot (not
just when .runner is absent). Before fetching a registration token it
validates existing .runner credentials via the forge admin API:
- 200: runner still registered, write dummy token and exit
- 404: runner deleted from forge, purge .runner and re-register
- 000: forge unreachable, keep credentials (runner surfaces the error)
- other non-200 or malformed .runner: purge and re-register
Removes ConditionPathExists so stale credentials from a wiped forge
no longer block the runner indefinitely. Updates docs/ci.md to match.
