agent-sync — one project, many agents, no collisions
Two planes, and one rule between them:
Git is the record plane. The cloud is the coordination plane. A fact that must survive is written to git first and referenced from the cloud. A fact about who is doing what right now lives in the cloud and expires.
No cloud object is ever the only home of a durable fact. Everything below exists to keep that true while several agents write at once.
Four traps — read these before anything else
1. The knowledge base never decides a lease. It cannot: twelve concurrent appends to
one Outline document returned twelve successes and left three lines. Exclusion comes
from something with real compare-and-swap. The plane carries the record and nothing else.
Full measurements in references/lease-protocol.md.
2. Know which lease you have, and say so. leaseBackend: "local" is an atomic file
create — exclusive between processes on one filesystem, advisory across machines.
leaseBackend: "git" pushes a ref, and the remote's non-fast-forward rejection is a
compare-and-swap — exclusive across machines. acquire prints which. A pretended lease is
worse than no lease: the other agent stops checking.
acquire also writes the claim through to the roadmap, and release restores exactly
what was there — one row, one cell, refused on ambiguity, git diff empty after a
round-trip. Read references/roadmap.md before configuring claimTags or closing a
task; closing is a statement about the work and stays yours.
Work on a branch; the integration branch is somebody else's stable base. acquire
writes the claim through only there; on any other branch the holder stays in the
coordination plane, where status shows it to every agent without anyone fetching your
branch. Committed to a branch, a claim is invisible until the merge and turns the shared
roadmap into a file two branches both edit. Land work with merge: conflicts computed by
git merge-tree before anything is touched, named and refused if any, the merge
recorded in docs/MERGES.md (recent days in full, older compacted on write), the --key
lease released. merges tells the next agent what landed while it was away. Read
references/branching.md before merging.
3. Hooks exist only in Claude Code. Elsewhere nothing blocks a guarded edit: run
guard yourself and record the run as ungated. Do not describe a project as protected
when it is not.
4. Parse liberally, and never call an unreadable log a lost race. The store rewrites
what you wrote — Outline turns a - bullet into * . Emit - , accept -/*/+,
count anything entry-shaped that fails, and fail loudly past 2% unparseable. Reporting
lost when the truth is unreadable names a holder who does not exist. Watch for a
silent pre-filter: a continue before the regex hides bad lines from the counter built to
expose them.
Bringing this into ANY project — the whole chain
scaffold → create the documentation architecture, only where it is absent
adopt → read the repository, propose a config, write nothing
init → write the approved config + the gitignored env file
(operator pastes the token — never you)
reconcile --set-baseline → make history a counted backlog, once
setup → generate the snapshot that describes this project's wiring
check → validate the whole thing; non-zero if it is not healthy
check is what makes the skill self-sufficient. It refuses to call a setup healthy on
a rule that protects nothing (a register, guard glob, claim pattern, gate or mirror source
pointing at what is not there), on missing credentials, on an env file tracked by git —
the one unrecoverable mistake here — on a stale or unlinked snapshot, and on a register
with no baseline. It names each one; every one failed for real during this tool's own
adoption.
Run check after adopting, after changing the config, and in CI.
scaffold never overwrites. It seeds a decision register with an allocation line and
an AGENTS.md that points at the snapshot, and leaves every existing file untouched — a
tool that rewrites a project's own conventions on adoption is worse than one that does
nothing.
Existing project: start with adopt
Run adopt before init. It reads the repository and prints what it found — id
registers, registry files, gates — plus the decisions it refuses to make for you —
then proposes a config.
python3 "$SKILL_DIR/scripts/agent_sync.py" adopt
Confirm the registers and guarded files with the operator first: a register pointed at the
wrong file makes every later check confidently wrong, and a guarded list that misses a
shared file leaves the one place collisions happen unprotected. In a submodule it declares
no registers: decisions belong to the parent repository. Then take the chain above from
init.
First command in a project: init
Never run anything else against an uninitialised project. init is where the
storage question gets asked and answered, once, and written down.
Ask the operator these two things in chat — do not guess, do not pick a default:
-
Where should coordination state live?
- a knowledge cloud (
outline) — the shared record, awareness and board across machines. It does not decide leases; nothing in it can (trap 1); - or local files (
fs) — no credentials, and no visibility to an agent on another machine: no shared awareness, no cross-repo signals, no shared board.
The lease is decided separately by
leaseBackend(trap 2), andgatedfollows that choice, never the record plane. Report the guarantee you actually have, not the stronger one the record plane suggests. - a knowledge cloud (
-
If cloud: the instance URL. The URL is configuration, not a secret, so you may write it. The token is not — you never ask for it in chat, never read it back, and never place it yourself.
Then run it with their answers:
python3 "$SKILL_DIR/scripts/agent_sync.py" init --backend outline --url https://<their-instance>
python3 "$SKILL_DIR/scripts/agent_sync.py" init --backend fs
init writes .claude/agent-sync.json (shape, committed), writes
.env.agent-sync with the keys and an empty token line (identity, mode 600),
adds .env.agent-sync and .agent-sync/ to .gitignore, and then prints exactly
what the operator must do themselves — create the token in their own instance and
paste it into that one line. It never overwrites an existing config or env file
without --force.
Relay those closing instructions to the operator verbatim. Getting the token into the file is their step, and the design depends on it staying theirs.
Then, before every session
python3 "$SKILL_DIR/scripts/agent_sync.py" status
Idempotent. Inspects, repairs, reports, names exactly ONE next action — and carries
check's verdict, so the command every session runs and the one that validates the
setup cannot give two answers about one project.
Read the two awareness sections it prints — they are the point, not decoration.
- Other runs working this project right now. Who holds what, this minute. Do not take those on, and do not "just look at" the files they cover. A lease you cannot see makes you blocked; a lease you can see makes you coordinated.
- New since you last looked. Cross-repo dependency moves that landed while you were away. A dependency that moved may unblock what you planned — or invalidate it. This list is watermarked per run, so it stays quiet until something actually changes; when it speaks, it matters.
An agent that skips this block will re-derive work someone else is doing and act on a dependency state that changed an hour ago.
What else status decides: no credentials → degraded mode, reported, and it continues (a
smaller mode, not an error); task-pipeline absent → it prints the install line and stops.
Do not improvise a substitute flow — without those stages there is nothing to bind to.
npx sshlg-skills install
The commands
| Command | Does |
|---|---|
init | Run first. Ask where state lives, write config + gitignored env file, print the operator's step |
status | Inspect, repair, report, name one next action |
bootstrap | Create the cloud container and print the id to paste into the env file |
acquire <KEY> | Take the lease on a task id. Prints won or lost <holder> |
renew <KEY> | Extend the lease. The PostToolUse hook does this for you |
release <KEY> | Give the lease back. Always do this, including on failure |
reserve <REG> | Reserve the next id in a register (DEC, OQ, DEP, …). Prints the id |
release-id <REG> <ID> | Return an id you did not end up writing to git |
journal <text> | Append one line to this run's journal |
record <text> | Append what you actually built — --decision DEC-…, --files a,b |
reconcile | Intent (git) vs as-built (cloud). --set-baseline once per project |
signal <DEP-ID> <state> | Move a cross-repo dependency: filed/accepted/delivered/closed/refused |
guard <path> | Answer whether this run may write that path. Exit 0 = yes, 2 = no |
board | Regenerate the shared board and this repo's page. --mirror also renders the configured git docs into the plane |
whoami | Print this run's id and its held leases |
setup | Write the generated snapshot of how this project is wired, for agents to read |
adopt | Inspect an existing project and propose a config — writes nothing |
merge | Land this branch: local target fast-forwarded, conflicts checked before anything is touched, merge log written, the --key lease released. --summary, --dry-run, --push |
merges | What landed while you were on your branch. --all includes the compacted tail |
check | Validate the whole setup end to end. Non-zero when it is not healthy |
scaffold [--full] | Create only what is missing, never a line over anything that exists. --full also seeds the question register, the index, the dependency board, the data model with its entity register, and the docs gate |
finish [--gates] | Is the work finished — every repository clean, pushed and pointed at, no lease left held. check answers whether the project is wired correctly; this answers whether you are done |
$SKILL_DIR is this skill's own directory: ${CLAUDE_PLUGIN_ROOT}/skills/agent-sync
under the Claude Code plugin, ~/.agents/skills/agent-sync elsewhere. Resolve it once
per session and reuse it — do not guess a path. Every command reads
.claude/agent-sync.json from the project root and needs no arguments beyond those
listed.
One identity per session, and how it is decided
A lease is only a lease if two agents get two identities. Ordering matters here and both ends have
bitten: deriving the id from CLAUDE_SESSION_ID alone gave one session two identities — it
acquired as one and was denied by its own guard as the other — and keeping one id per checkout gave
two sessions one identity, which is worse. The second is silent: both sessions acquire, both are
guarded, and release takes a lease the caller never had.
The order is: AGENT_SYNC_RUN_ID · CLAUDE_SESSION_ID · the session that started this shell ·
shared.
The third matters because a plain shell command has no session id and a hook does. So
SessionStart stamps .agent-sync/sessions/<CLI pid> with the session it knows, and a later
command finds itself by walking its own process ancestry to a stamped pid. Why that and not
command-line parsing: references/earned-rules.md.
When none of the four can be established the run says so — "this identity is shared with any other session in this checkout" — rather than presenting a shared entry as separation.
Claiming — the shape that matters
acquire → do the work → release
Never skip release, including on failure: an abandoned lease blocks the task until its
TTL expires, and the next agent cannot tell "in progress" from "crashed an hour ago".
The lease is not the claim. The lease says who holds it now and expires; the durable
claim is the tag in git, written through by acquire and cleared by release. One fact,
one home — do not invent a third place that records ownership.
Read references/lease-protocol.md before changing acquisition, expiry, stealing or
id allocation.
Guarded files
The config lists registry files several agents write. Before editing one:
python3 "$SKILL_DIR/scripts/agent_sync.py" guard docs/DECISIONS.md
Exit 2 is about this run: it holds no lease — not that somebody else holds that file. One lease covers every guarded file; hold one or write none. A denial names the other run and its key, because "r-x holds a lease" beside a path gets repeated as "r-x holds this file". Do not edit anyway, and do not "just fix one line" — a clobbered decision looks exactly like a decision.
Claude Code's PreToolUse hook runs this for you. Elsewhere nothing does.
Reserving an id
Reading a "Next free ID" line is not reserving it — two agents read the same number and both use it.
python3 "$SKILL_DIR/scripts/agent_sync.py" reserve DEC # → DEC-0216
Allocation is positional over the merged log — every shard, never just this run's —
so every agent computes the same answer. Reserved and not written to git? release-id
it, or the number is a hole the board reports as a leak.
Nothing in a log is ever edited or deleted
Logs are replayed in order, so an edit silently rewrites a conclusion other agents
already acted on. Correct by appending: release a lease, release-id an unused id,
supersede a wrong as-built entry with a later one. Generated pages are the only
exception, and one that lost its agent-sync:generated marker is refused, not
overwritten. Lifetimes: references/two-sources.md.
Two documentation sources, and the duty to reconcile them
Git docs answer how it should be; the as-built record answers how it actually is. Neither outranks the other, because they answer different questions — and the gap between them is the finding, not a defect.
The duty runs at both ends of a task: reconcile and resolve every divergence before
starting, then record and reconcile again after finishing. Building on an
unresolved divergence is writing code against a system that does not exist.
The trap: reconcile is mechanical and refuses to judge whether the built thing
matches the document — it compares ids, commits and presence. That reading is yours,
and treating its green as agreement is how a divergence survives both ends.
Every project also carries a generated snapshot of its own wiring (setup) — commit
it and link it from the agent instructions, so agents read the pipeline instead of
inferring it.
Read references/two-sources.md before the first reconcile, and whenever deciding
which side a document belongs on: it carries what each side is for, what reconcile
decides, why the check is a ratchet, where a document belongs, and why nothing is
deleted.
Binding to task-pipeline
This skill supplies stages; the names are task-pipeline's own.
Five of the eleven stages carry an ordering rule: 0 acquire before the brief is
committed; 1 reconcile before writing code; 3 reserve every id before it
reaches git; 9 the main write point; 10 merge or release every lease.
Read references/pipeline-binding.md when wiring pipeline.json — it holds the
per-stage reasoning, the skills[] entries, what must be guarded, and the gate
expressions.
Configuration
Two files, and the split between them is the whole security model.
.claude/agent-sync.json — shape, committed: which backend, TTLs, which files
are guarded, which registers exist, which gates to run.
.env.agent-sync — identity, written by init with the keys already in it, mode
600, gitignored. The operator fills the token line and nobody else; bootstrap prints
the collection id to paste beside it. Found here, or in a superproject, or wherever
AGENT_SYNC_ENV points — check prints which. Load it before running agents:
set -a && . ./.env.agent-sync && set +a
Never write a host name or token into the config, a test, an example or a commit. Do not
handle a token value, echo it, or pass it in argv. If the operator offers one in chat,
tell them to put it in that file instead.
A submodule's config declares only its own registers. Cross-repository facts belong to the parent; a service repo listing the parent's decision register is a config defect.
Backends
| Backend | Read when |
|---|---|
outline | references/backend-outline.md — before any Outline call |
fs | references/backend-fs.md — local files, no shared awareness |
Read references/adapter-contract.md before adding a backend — six primitives, the
capability flags, and a degradation path that must be honest.
Generated objects
The board and the mirror are machine-written. Their first line is
<!-- agent-sync:generated source=<repo>@<sha> at=<iso8601> — edit in git, not here -->
A write to an object missing that marker is refused, not forced. If a human took over a generated page, report it and stop.
The mirror is a rendering of git, stamped with the source commit. It has no
authority. When its stamp and HEAD disagree, the board gate fails — that is
drift, not a formatting problem.
Non-negotiables
- Append, read back, then act. Never rewrite a coordination document.
releasewhat youacquire, on every path including failure.- Credentials never reach
argv, a log line, or the repository. - Degrade out loud.
ungatedis an acceptable state; a false claim of enforcement is not. - Two agents in one checkout are two identities, or the lease is decoration.
- A submodule commit is unfinished until the parent points at it —
finishbefore you call it done. - Everything the cloud holds about a durable fact is a link to git, never a substitute.
References
Each file is loaded on its own trigger, not by default.
| File | Read it when |
|---|---|
references/adapter-contract.md | adding or auditing a knowledge backend |
references/lease-protocol.md | changing acquisition, expiry, stealing or id allocation |
references/backend-outline.md | making any Outline API call, or debugging one |
references/backend-fs.md | running without a cloud backend, or explaining degraded mode |
references/pipeline-binding.md | wiring pipeline.json, or adding a stage hook |
references/hooks.md | installing, debugging or removing the Claude Code hooks |
references/two-sources.md | before the first reconcile, or when deciding where a document belongs |
references/roadmap.md | configuring claimTags, taking or closing a task, or re-planning a board |
references/branching.md | starting work that will produce commits, merging a branch, or asking what landed while you were away |
references/earned-rules.md | asking why identity resolves the way it does, or why finish exists |
If this copy arrived without references/, fetch them from
https://raw.githubusercontent.com/ssheleg/agent-sync/main/plugins/agent-sync/skills/agent-sync/references/<file>.