hive-forge-notify grows a second binary, hive-github-notify. The two
share the notification half of the job — tolerant parse, classification,
formatting, dedupe, todo delivery — and nothing else: each binary owns
its host's protocol outright.
Two binaries rather than one multi-source daemon, and rather than a
cargo feature. A feature would unify across the workspace and cost every
crate its build cache. Two binaries keep the decision in nix: forge.nix
installs the forge unit, github.nix installs the github one under
hyperhive.github.enable, so a hive built without that module has no
github poller in its closure at all — GitHub access is separable (a
tier, a policy boundary), not merely switched off. Both binaries ship
from the existing derivation, so packages.nix is untouched.
The split is real at the code level too, not just at the unit level.
source.rs is a trait; the impls live in the binaries that use them, so
neither binary links the other's protocol code and the library names no
host at all. The forge-only assigned-issue rollup moves into the forge
binary for the same reason: it asks the forge what is assigned to this
agent, which is not a notification-protocol concern.
At runtime the github unit needs a PAT at <state>/github-token, the same
dashboard-provisioned token the gh wrapper and the git credential helper
already use. No PAT: it logs why and exits 0, which is why the unit is
Restart=on-failure and not always.
Forgejo's notifications API is modelled on GitHub's, so one tolerant
parse serves both — the differences (string thread ids, PullRequest vs
Pull) are absorbed by lenient deserializers rather than a second parse
path. Thread ids normalise to String at the parse boundary; they are
only ever opaque keys. Todo keys gain a per-source prefix so the two
hosts cannot collide, and the forge's is deliberately empty to keep
existing forge todo keys stable across the deploy that lands this.
The github loop honours the server's X-Poll-Interval, re-arming only
when the server asks for a slower cadence than ours; the hint is read
before the status check, because it arrives on error and empty pages too
and that is exactly when it matters. Reading the notification stream
needs the notifications scope on the PAT, which a token minted for push
access typically lacks; the failure mode is silence, so docs/github.md
says so explicitly.
The poller was a `tokio::spawn` inside the `hive-agent` serve loop. It
never needed anything from that loop except a socket path, so being
in-process bought nothing and cost two things: a harness restart took
forge notifications down with it, and the whole forge/HTTP dependency
tree was linked into the serve-loop binary.
It is now `hive-forge-notify`, a per-agent daemon with its own systemd
unit, a sibling of `hive-bash-daemon` and `hive-matrix-daemon`. Same
contract as those two: it reaches the harness only by upserting todos on
the in-agent socket, and nowhere else.
The module moves verbatim (`notify.rs`) — the formatters, the activation
gates, the dedupe map and all 33 tests are unchanged. Only the socket
call sites are rewritten, onto a small local `todo_client` rather than
the harness's. That mirrors what both sibling daemons already do, and
the etiquette differs on purpose: the harness's client carries a 60s
backoff schedule sized to ride out a hive-c0re restart, which its
callers need because they have no retry of their own. This poller's two
call sites both sit inside the 30s poll loop and both treat a failure as
"leave the thread unread, try next tick", so the poll interval already
is the retry; a second backoff would only stack sleeps and delay the
rest of the batch.
The unit is `Restart=on-failure`, not `always`. An agent with no forge
account is a supported configuration and the poller reports it by
logging why and exiting 0 — under `always` that clean exit would be a
restart loop on every forge-less agent.
`forgejo-api`, `url` and `time` drop out of `hive-agent`'s dependencies
with the module.
Also corrects docs that outlived the code they described: the persisted
`forge_cursor` field is long gone (forge's own read-state is the durable
record of what has been delivered), but `docs/persistence.md` and the
`harness_state` module docs still documented it as live.
- .prettierrc: proseWrap=preserve (no prose reflow)
- .prettierignore: exclude hivectl-cli.md (auto-generated) + 11 docs
with multi-line list-item continuations prettier would strip to col 0
(CommonMark limitation in prettier's list handling)
- format 16 markdown files: cosmetic only (*→_, table alignment,
heading normalisation) — verified no broken continuations, idempotent
Per operator directive: a silent automatic force-push is a bug. The
config-repo merge path already never force-pushes (run_merge_config_pr
lands via ff_push_to_main, a non-force git push). So set the branch
protection's enable_force_push to false — main only ever advances by
fast-forward.
The legacy push_config mirror does force-push (it re-points the status
tags and rewinds main on a failed-build rollback); the protection now
rejects those non-ff updates, so the mirror runs best-effort until the
agent-opened PR-merge flow retires it. Docs + comments updated to match.
Address iris review nit on #1948: the bullet above now reframes the
agent-configs entry as a config repo, so the dashboard-anchor bullet
should match. Pure wording, no behavior change.
Wires the per-agent config repo as the editable PR surface the #1838
merge handler (run_merge_config_pr) consumes, without yet retiring the
push_config force-mirror (that waits for #1838 P2 — the agent-opens-PR
MCP surface — so config changes keep working through the transition).
ensure_config_repo now, after creating agent-configs/<name>:
- adds the agent as a WRITE collaborator (can push config-change branches
+ open PRs);
- branch-protects main core-only via apply_config_repo_branch_protection:
push + merge whitelists are core-only (only hive-c0re lands on main, via
its verify-and-ff-push handler), operator-team approval required, the
agent can't push main directly or self-merge.
The protection sets enable_force_push=true as a TRANSITIONAL allowance so
push_config's applied->main force-mirror keeps working until P2 retires it
(a protected branch otherwise rejects force-push). Forgejo's force-push
allowlist is deploy-keys-only — no per-user list — so this is a plain
enable_force_push toggle; only core is in the push-whitelist so only core
can force-push anyway. At P2, flip it to false and keep core in the
push-whitelist so ff_push_to_main still lands.
Agent code repos (AGENTS_ORG) are intentionally untouched — their merge
flow has no auto-merge handler yet, so switching them to core-only-merge
would strand those PRs.
All steps idempotent (runs on every spawn + startup sweep). Updated the
CONFIG_ORG doc comment + docs/forge.md to drop the "mirror-only" framing.
forge_notify.rs is the biggest remaining #716 hotspot: ~22 attribution
cookies (#110#230#231#253#256#455#518#539#544) plus a thick
module-level docstring + per-function rationale blocks for the
notification pipeline. The substantive prose lives in a new
`docs/forge.md` covering the wider forge integration story:
- Per-agent forge accounts + agent-configs mirror (was implicit
across `hive-c0re/src/forge.rs` rustdocs).
- Notification poller: activation gates, self-notification filtering,
body excerpt + truncation + ATX heading escape pipeline, wrapper
formats (comment / review / new-item / state-change) with shape
table, meta suffix shape, review-request override, reason
drop-list rationale (drop vs allow), auto-unsubscribe on broad
watches.
In-code rustdocs reduced to 1-line semantic summaries + doc
pointers; inline cookie comments scrubbed. Net diff is ~150 lines
removed from the .rs file. All 16 forge_notify unit tests pass.
CLAUDE.md gets a new `docs/forge.md` file-map entry + reading-path
question entry. The existing `src/forge_notify.rs` file-map blurb
keeps its `(#539 / #544)` cookie — consistent with the rest of
CLAUDE.md's lineage attributions.
