fulcra-agent-tasks
Skill10 starsChanged 2 months agoMITView on GitHub
Give a fulcra-agent-teams space a typed task lifecycle: create tasks with structured status/priority/assignee, and move them through a validated state machine (proposed→active→done) instead of freeform markdown.
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-presenceSkill
- fulcra-agent-reconcileSkill
- fulcra-agent-reviewSkill
- fulcra-agent-rolesSkill
- 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
2 rules. About 0 are common, 1 is distinctive, and 1 is 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.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 20
Before creating or transitioning a task, probe engine health and what work is already waiting.
- Distinctive: line 31
All probes pass with nothing needing you → the engine is healthy, views are fresh, and your queue is clear.
≈ “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 Tasks
Enhances the fulcra-agent-teams skill. Bare teams
tracks long-running work as freeform task/<name>.md. This skill gives those docs a typed lifecycle —
a real status/priority/assignee and a validated state machine — so a task's state is queryable
(via fulcra-agent-reconcile) and can't take an illegal jump (e.g. done → active). Pairs with
fulcra-agent-reconcile: this skill writes task state; reconcile reads/heals the views.
Where to start — the re-entrancy probes
Before creating or transitioning a task, probe engine health and what work is already waiting. Enter at
the first probe that fails (per the repo's skill-quality pattern, docs/skill-quality-pattern.md);
task writes are parse→modify→write and a same-status update is idempotent, so re-entry never corrupts
state:
| Probe (run in order) | Command | Passes when | If it fails, enter at |
| Engine + auth usable? | coord-engine doctor <team> | exits 0 and the last line is exactly doctor: healthy | fix engine/auth first (see Usage / fulcra-agent-reconcile) — do NOT write tasks against a broken engine |
| Aggregate fresh? | coord-engine status <team> | output does NOT contain (no aggregate for team/ (the CLI's missing-aggregate hint) | run coord-engine reconcile <team> to build the aggregate, then re-probe |
| Work for me? | coord-engine needs-me <team> --agent <id> | prints 0 item(s) need <id>: (a non-zero count means work is waiting — enter there) — read the rows (including any [REVIEW] pending verdict review-pending rows) before starting new work | pick up the listed items (transition them via task update/task done) before creating more |
All probes pass with nothing needing you → the engine is healthy, views are fresh, and your queue is clear; proceed to create or advance tasks below.
Why the writes go through the engine
Writing OKF frontmatter correctly and enforcing which transitions are legal are deterministic
requirements — a malformed doc or an illegal waiting → done is a correctness bug, not a style choice.
So the lifecycle commands are the shared coord-engine tool (parse→modify→write, transition-checked),
not prose the agent hand-edits. Composing the human note in the task body is fine as prose; the
structured state is not.
The state machine
proposed → active | waiting | abandoned | done
active → waiting | blocked | done | abandoned
waiting → active | blocked | abandoned
blocked → active | waiting | abandoned
done, abandoned → (terminal)done requires evidence. A same-status update is always allowed (idempotent edit).
Usage
Needs fulcra-api authenticated and coord-engine installed — standalone:
uv tool install "git+https://github.com/ashfulcra/fulcra-tools.git@<latest-coord-engine-tag>#subdirectory=packages/coord-engine" (any
coord skill brings the same engine; installing once serves all).
# create a task doc at team/<team>/task/<slug>.md
coord-engine task start <team> "Fix the widget" \
--workstream web --priority P1 --status proposed --assignee user --summary "one-liner"
# move it through the machine (illegal transitions are rejected with a clear error)
coord-engine task update <team> fix-the-widget --status active --next "write the test"
coord-engine task block <team> fix-the-widget --blocked-on reviewer --unlock "review lands on PR #42"
# finish it — evidence is required
coord-engine task done <team> fix-the-widget --evidence "PR #42 merged"Each write stamps timestamp and appends a dated note to the task body; the Fulcra File Store versions
every write, so the full history is preserved. A task start with an --assignee also opens the obligation on
bus-v4 (when the team has a _coord/bus-v4/records.json), so the task shows in that assignee's fold-backed
needs-me/obligations at once. If that open fails, the doc is still written and the verb exits 3 naming the
obligations --export-open --force repair; it never fails quietly. After changing tasks, run
coord-engine reconcile <team> to refresh the index and views (or let a scheduled reconcile
do it).
See references/tasks-cli.md for the full flag list and the OKF Task shape.
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.

