fulcra-agent-presence
Skill10 starsChanged 2 months agoMITView on GitHub
Add presence + liveness to a fulcra-agent-teams space: agents heartbeat a presence shard, and deterministic folds answer who's live/idle/stale, what each agent is working on, and who a broadcast must reach.
In this repository20 other files it gives its agents.
- AGENTS.mdAGENTS.md
- coordinator-disciplineSkill
- fulcra-agent-atcSkill
- fulcra-agent-automationSkill
- fulcra-agent-cloud-coordinatorSkill
- fulcra-agent-continuitySkill
- fulcra-agent-directivesSkill
- fulcra-agent-durable-stateSkill
- fulcra-agent-forgeSkill
- fulcra-agent-healthSkill
- fulcra-agent-operatorSkill
- fulcra-agent-reconcileSkill
- fulcra-agent-reviewSkill
- fulcra-agent-rolesSkill
- fulcra-agent-tasksSkill
- fulcra-content-reviewSkill
- fulcra-fdeSkill
- fulcra-lab-resultsSkill
- map-our-teamSkill
- sealed-secretsSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
Before beating or reading the roster, probe how far this session already got.
- Distinctive: line 30
All probes pass → you are present and (if role-scoped) holding your lease.
- Very specific: line 69
Session/host details are metadata, not address: Put them in the presence -s summary or the lease body if useful.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Fulcra Agent Presence
Enhances fulcra-agent-teams. Teams knows who its
members are (prose in index.md) but not who is alive right now. This skill adds a presence
heartbeat + deterministic liveness folds — the roster that directives' broadcast semantics, the operator
digest, and role-vacancy escalation all build on. Optional: without it, everything else still works;
broadcasts just degrade to "acked-by-me hides it for me".
Where to start — the re-entrancy probes
Before beating or reading the roster, probe how far this session already got. Enter at the first probe
that fails (per the repo's skill-quality pattern, docs/skill-quality-pattern.md); every step here is
safely re-runnable (a beat is a single-file overwrite, a claim is a refresh):
| Probe (run in order) | Command | Passes when | If it fails, enter at |
| Engine usable? | coord-engine doctor <team> | exits 0 and the last line is exactly doctor: healthy | fix engine/auth first (see fulcra-agent-reconcile) — do NOT beat against a broken engine |
| Own shard live? | coord-engine presence show <team> | a row for your agent id shows liveness [live ] (the <1h bucket) | Beat — run presence beat <team> (see Usage) to write/refresh your shard |
| If identity=role, lease held? | coord-engine roles status <team> <role> | prints role <role> in team/<team>: HELD for the role you act as | Claim — run roles claim <team> <role> (skip this probe entirely if you are not acting under a role identity) |
All probes pass → you are present and (if role-scoped) holding your lease; just keep beating on your cadence. A brand-new session fails the first or second probe and enters at Beat.
How it works
- Beat (single-file write, safe as a command):
presence beatwrites/refreshes your shardteam/<team>/presence/<agent-key>.md(collision-safe key) (OKFtype: Presence: agent, workstreams, summary, timestamp). Beat whenever you start work, on heartbeat/cron ticks, and when your focus changes. - Folds (deterministic, engine-side — never eyeball timestamps):
presence show— roster withlive(<1h) /idle(<24h) /staleper agent.agents— cross-agent digest: each agent's liveness, summary, and open work by status (union of presence ∪ task owners/assignees from the reconcile aggregate).- Roles:
roles claim <team> <role>writes/refreshes your lease shard;roles releasedeletes it;roles statusfolds HELD/VACANT/CONTESTED (see fulcra-agent-roles).
Usage
coord-engine presence beat <team> [--agent X] [-w workstream]... [-s "one-liner"]
coord-engine presence show <team> [--json]
coord-engine agents <team> [--json]
coord-engine roles claim <team> <role> [--agent X] # refresh = re-run
coord-engine roles release <team> <role> [--agent X]--agent defaults to $FULCRA_COORD_AGENT (or a derived host id). Stale shards drop out of the
presence fold's [live] view by age; the shard FILES are not currently garbage-collected (reconcile's
GC covers ack and health shards only). A stale agent reappears by simply beating again.
Pick your identity by ROLE, not by folder
Set FULCRA_COORD_AGENT to the role you are acting as (maintainer, preferences-maintainer,
release-reviewer), not a host/cwd-derived string. Folder-derived ids collide the moment two sessions
share a directory (shared inbox, clobbered presence, ambiguous acks) and rot when a hostname or checkout
path changes; a role-based id survives both and is what teammates actually want to address. Two rules
make it safe:
- Claim the role's lease while you act as it (
roles claim <team> <role>; see fulcra-agent-roles). Anexclusiverole turns two sessions acting as the same role under DIFFERENT ids into a visible CONTESTED state. It cannot see two sessions sharing one id string (same lease shard, last write wins) — see the roles skill's "Role-as-identity" guard matrix for the procedural check that covers that case. - Session/host details are metadata, not address. Put them in the presence
-ssummary or the lease body if useful; never in the agent id.
The derived host id remains only as a fallback for throwaway/anonymous sessions that never take
assignments — and note it is per-HOST, not per-session (coord-reconcile:<hostname>), so two env-less
sessions on one machine still share an id and clobber each other's shards. Any session that acts on the
bus should set an explicit role id.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

