concord

Findings

Measured on 2026-11-18 on macOS with jj 0.47.1, GitButler 0.23.4, Mergiraf 0.20.0 and Python 3.12. Agents were Claude Haiku subagents. Each setup ran once on a scenario built to collide, so these results show failure modes, not frequencies.

The cart scenario

cart.py has total(items) and item_count(items). Three agents start from the same base, none told about the others:

AgentTaskCollides with
Arename total to subtotalB (same function), C (C calls it)
Badd discount_pct=0 to totalA (same function); its formula makes the result a float
Cadd receipt() calling total; its test expects "3 items, total 25"A (name), B (the float prints 25.0)

A correct combined result passes 4 tests. Two of the bugs are invisible to text merging: the rename merges cleanly while the caller still calls total, and the float causes no textual conflict at all.

Results

SetupWhere the conflict surfacedCombined tests
plain git worktreemerge stops at B; C cannot merge until a human resolves itsame code as the Mergiraf row
git + Mergirafstops at B; Mergiraf resolves the rename and body change2/4
jj workspacesstored as a 3-sided conflict in the merge commit, nothing blocksB+C alone fails on 25.0
jj + Mergirafjj merge tools take 2 sides, so merges go pairwiseA+B already 1/3 failing
GitButler, one shared directoryhunks of different agents fuse while editing; 0/3 agents committed2/4, B's code never landed
GitButler + but worktree newcommits marked conflicted; refuses to land a conflicted commit2/4, landed B with failing tests
jj + claims and notes (v1)B stacked on A, C stacked on A3/4: rename caught, float missed
jj + concord board + gateA, B, C serialized on total; C resolved its own merge4/4 on the first gate run

What each layer is worth

Known gaps

Dogfood: concord developing concord

Three Claude Sonnet subagents, one workspace each, briefed only with the concord skill, their task and their scope, built three features of concord itself. Two of them planned to modify the same function: the second waited, claimed after the first one's done, and got its change merged in before editing. The gate passed on the first run with clippy clean and 25 tests passing.

The run showed two gaps in the skill, both now stated in it: an agent first planned modifies=- for a function whose body it changed, and edited before claiming. Docs are not functions: all three would have edited the design doc in parallel, so the orchestrator kept docs to itself and edited them after the gate.

On an existing git repository

Before landing was designed in, a git user saw a detached HEAD, a branch that did not move, an untracked file committed into the agents' base, and an index out of step with HEAD. The fixes: agents start from the last commit, the main workspace tracks files like git, and the gate tests in a scratch child and lands like a pull or restores everything on FAIL. experiments/git-landing/run.sh checks both paths.

Public benchmarks: merge tool A/B

ConflictBench: 159 real Java merge scenarios. A pass means the merged files equal the developer's resolution, text for text.

Gate merge toolsPassConflictWrong
plain git merge-file (2 cases error)01570
Mergiraf2710626
Mergiraf, then Weave378537

Weave turned 21 conflicts into merges: 10 equal the developer's resolution and 11 differ. Of those 11, 3 were equivalent, 3 kept both sides harmlessly, and 5 were broken, including code that does not compile and a dropped parameter. The rule set before the run was to keep Weave only if passes rose without wrong merges rising. It failed, so the gate uses Mergiraf alone. "Wrong" means text inequality, so some wrong merges are fine code; the column compares tools, it does not count bugs.