CORAL
Open-source autoresearch powered by autonomous coding agents. Run Claude Code, OpenCode, and Codex with grading, shared knowledge, and multi-agent evolution. Accepted at COLM 2026.
Runssource: GitHubhomepagePythonApache-2.0commit 0123dfb939b3
Python, Apache-2.0 licensed. The project labels itself: agent framework, agent orchestration, agentic ai, ai agents, alpha evolve, autonomous agents, autoresearch and claude code.
CORAL runs. An Argusic agent installed it in 5 minutes and hit 2 errors and fixed every one without help on a clean machine with no GPU, and the whole session was recorded.
coral 0.7.22.dev7+g0123dfb93.d20261118 installed in editable venv, 740/741 tests pass, CLI commands (init, validate, --version, --help) all work.
What the agent ended up with on a clean machine, in its own words. How this is measured
Measured by Argusic on a fresh machine every time. Every number links to its evidence. Argusic Score 100 of 100 (the mean of this project's run scores).
At a glance
Subject data from GitHub, linked at the top of this page, refreshed . Test data by Argusic (CC BY 4.0); every number links to a run page with the full log, the recording, and their sha256 hashes.
What was measured
From the run that decided this verdict. Every line is a recorded value, not an estimate.

What went wrong, in the agent's own words
The agent's own words, worst first, never rewritten into advice.
- tmux binary not available in container — test tried to spawn ['tmux', 'display-message', '-p', '#S'] as a subprocess and got FileNotFoundError10 minutes
- test_distribution_version_is_based_on_latest_reachable_release_tag failed — git describe couldn't find any tags reachable from HEAD (shallow clone)5 minutes
- Install time
- 5 minutes
- Cold machine to finish
- 20 minutes
- Errors hit and fixed
- 2 hit, 2 fixed with no human help
- How the result was proved
- pytest tests/ -v exited 0: 740 passed, 1 skipped (test_swebench_parse_with_real_harbor_result), 0 failed in 107s. Also exercised 'coral init' scaffold + 'coral validate' on a fresh task directory, both succeeded.
- Model tokens used
- 97,451
- Exact commit tested
- 0123dfb939b3
- Container image
- sha256:cdd920bce783
What was not measured
The other half, so you can see exactly how far this evidence goes. A gap is printed as a gap and never filled in.
- Peak GPU memory
- this ran on a CPU machine, so there was no GPU to sample
- Behaviour over a long run
- each attempt uses the smallest documented settings and a fixed time budget, so this measures whether it runs, not how it holds up under load
Watch the session
The whole run, from a clean machine, at the speed it actually happened. Nothing is cut and nothing is sped up.
A recording of attempt 1, run . The verdict above comes from a different attempt, so read this as one session and not as the current result. The terminal recording it was made from is the evidence, stored with a sha256 fingerprint; the video is a rendering of it. Nothing loads from the video host until you press play.
The end of the recording
The last lines of the terminal session that decided this verdict, exactly as they were recorded.
path Path to the task directory options: -h, --help show this help message and exit --json Output one machine-readable validation result Examples: coral validate my-task coral validate my-task --json coral 0.7.22.dev7+g0123dfb93.d20261118 Here's the final summary. All work done. tokens used 97,451 Here's the final summary. All work done.
Replay the whole session, every command from a clean machine to this point.
How it was tested
One agent, one container, no human help. It installs the project the way its own documentation says to, runs it, and fixes what breaks. Everything below is recorded as it happened: the terminal session, the log and the exact commit. The full method.
Strengths and limits
Measured facts, not opinions. How this is written.
What went well
- Reached a running state on a clean machine, with the session recorded.
- Recovered from all 2 errors without a human stepping in, which says the failures are documented well enough to solve.
- Ran without a GPU, so it does not need one to start.
- Licensed Apache-2.0, as reported by its host.
- Measured 2 times, so the result is not a one-off.
What did not, or is not known
- Took 5 minutes to install, slower than the median of the 63 comparable projects Argusic has measured.
Of the 64 agent-framework projects Argusic has installed and timed, CORAL was the 38th fastest to reach a running state, and 51 of 64 reached one at all.
What it is a reasonable choice for
- Trying it on a laptop or a small server: it reached a running state without a GPU.
- A quick evaluation: it was running 5 minutes after the clone.
Also tested, in the same area
Every one of these was installed and run by Argusic on a clean machine. Nothing appears here that was not tested.
Run history
Topics (from GitHub)
agent-frameworkagent-orchestrationagentic-aiai-agentsalpha-evolveautonomous-agentsautoresearchclaude-codecode-generationcodexcoding-agentevolutionary-algorithmlarge-language-modelsllm-agentsmulti-agentopen-sourceopencodescientific-discoveryself-evolvingself-evolving-agents
Embed the badge
Markdown for the project README. It links back here; terms on the terms page.
[](https://argusic.com/subject/coral)Questions
- Does CORAL run?
- Yes. CORAL runs. Argusic installed and launched it on a clean machine in 5 minutes, hitting 2 errors on the way, and recorded the session.
- How did Argusic test CORAL?
- On a fresh, disposable machine, with every command recorded and the repository pinned at commit 0123dfb939b3. 2 attempts are recorded, and the full method is on the methodology page.
- What did it cost to test CORAL?
- The run that produced this verdict cost $0.09: real compute and model cost, not a list price per million tokens. It is recorded on that run's page.
- How long does CORAL take to install?
- 5 minutes in the run that decided this verdict, from a clean machine to a running state. That is one measured attempt, not an average.
- Does CORAL need a GPU?
- Not to start. Argusic reached a running state on a machine with no GPU.
- What goes wrong when installing CORAL?
- 2 things broke in the recorded run, and 2 were fixed without human help. Each one, and the time it cost, is listed on this page.
- How does CORAL compare with the alternatives?
- Of the 64 agent-framework projects Argusic has installed and timed, CORAL was the 38th fastest to reach a running state, and 51 of 64 reached one at all.
- Where is the evidence for CORAL?
- All 2 recorded runs are on this page, each linking to its full log and terminal recording, stored with a sha256 fingerprint so it cannot be quietly altered.