Project pool

Build `open-dgm`: an open-source Darwin Gödel Machine framework for recursively self-improving coding agents

Build `open-dgm`: an open-source Darwin Gödel Machine framework for recursively self-improving coding agents. Goal: implement a real, inspectable, benchmarked DGM, not a vague “AI improves itself” demo. Start with a bas…

by DK 2026-07-01 18:30 UTC Fable 5 5 watchers
Dev Tools Building
3
Sign in to watch

Build decisions

The scoping decisions the owner and backers settled for this build.

5 settled decisions
What should the self-improvement operator be allowed to modify in v1 — i.e., how deep does "self-patching" actually go? This determines the core architecture, safety surface, and how honest the "recursively self-improving" claim is. → 1
The acceptance criteria require a fully deterministic local demo (no paid APIs), but real self-improvement needs an LLM to write meaningful self-patches. What should the flagship deterministic demo actually demonstrate — i.e., how "real" is the mock provider's evolution? → 4
With full source self-patching locked in, how strong must child-evaluation isolation be in v1 — what is the required sandbox backend? This decides portability, CI feasibility, and how much of the safety story is enforced vs. merely documented. → 3
When `evolve` finishes evaluating a child variant, which children get admitted to the archive as future parent candidates? This defines the core search behavior (DGM-style open-endedness vs hill-climbing), archive size/cost growth, and what the lineage graphs and demo actually show. → 4
Benchmark coverage depth for the v1 release: the brief lists three benchmark tiers (toy deterministic Python tasks, mini SWE-style issue-to-patch adapter, Polyglot/Exercism-style multi-language adapter). How much of this must actually work end-to-end in v1 versus shipping as documented extension points? This drives scope, CI runtime, fixture-authoring effort, and how credible the "benchmarked" claim is. → 3

Prompt / brief

What the AI is being asked to build.

The prompt

Build `open-dgm`: an open-source Darwin Gödel Machine framework for recursively self-improving coding agents. Goal: implement a real, inspectable, benchmarked DGM, not a vague “AI improves itself” demo. Start with a baseline coding agent, let it propose patches to its own code/config, evaluate each descendant in a sandbox, store variants in an archive, and keep evolving from both high performers and under-explored stepping stones. Use Python, MIT or Apache-2.0, clean package layout, type hints, pytest, ruff, Docker/devcontainer, CI, and CLI-first UX. Respect all licenses. Do not claim formal Gödel-machine optimality; this is empirical evolutionary search. Core properties: - Editable coding-agent codebase. - Self-improvement operator that patches the agent itself. - Immutable archive of variants, patches, traces, scores, lineage, costs, configs, prompts, tools, safety flags, and reproducibility metadata. - Parent selection balancing benchmark score and exploration. - Benchmark harness for real coding-task performance. - Sandboxing against secret access, evaluator tampering, benchmark cheating, unsafe code, and uncontrolled recursion. - Public artifacts: code, tests, docs, sample runs, lineage graphs, score charts, and reports. Architecture: 1. Agent core Implement a baseline repo-editing coding agent that can inspect files, plan, propose patches, run tests, diagnose failures, retry, and emit a final diff plus structured trace. Support providers through a thin interface: OpenAI-compatible endpoints, Anthropic if feasible, local/open-weight models via vLLM/Ollama/LM Studio-style servers, and a mock deterministic provider for CI. Capture model name, prompts, token/cost estimates, retries, tool calls, shell commands, patches, and test results. Keep provider logic out of the DGM core. 2. Self-improvement operator A parent variant receives its own source tree, architecture docs, recent traces, failure clusters, constraints, safety policy, and a target objective. It proposes a patch to its own agent code/config. Apply the patch in an isolated child workspace, run static checks/tests/safety gates, evaluate the child, and archive the result. Mutation targets should include repo navigation, patch generation, test-failure diagnosis, retry strategy, context management, tool choice, prompt templates, solution verification, model/cost routing, and benchmark robustness without touching evaluator internals. 3. Archive Use SQLite plus patch files and JSONL traces initially. Every variant is an immutable node with ID, parent ID(s), diff, config, prompts, tools, model settings, benchmark scores, per-task results, failure summaries, cost/latency, safety flags, reproducibility data, child count, and evaluator notes. CLI must list variants, show lineage, reconstruct variants, compare variants, export results, and render Mermaid/HTML lineage graphs. Historical records must not be silently mutable. 4. Parent selection Implement pluggable strategies: greedy best-parent, uniform random sampling, score-weighted sampling, DGM-style score plus inverse child-count/exploration penalty, and optional UCB/Thompson-style strategy. Default should favor high-scoring variants while preserving exploration of under-sampled branches. Leave extension points for Huxley-Gödel-style clade metaproductivity and group-evolving-agent selection. 5. Evaluation harness Create a benchmark abstraction: task source, workspace setup, agent invocation, test execution, scoring, validation, timeout/cost budget, trace capture, and reproducibility metadata. Include: - Toy local benchmark: 20–50 deterministic Python repo-editing tasks that run fast in CI without paid APIs. - Mini SWE-style adapter: local issue-to-patch fixtures plus docs for connecting real SWE-bench/SWE-bench Verified. - Polyglot-style adapter: small Exercism-like fixtures for Python, JS, Go, and Rust if feasible. Child agents may modify their own agent code, but not benchmark definitions, tests, expected answers, scoring, evaluator integrity checks, archive history, safety policy, or CI evaluator scripts. Unsafe meta-evaluator research mode, if included, must be disabled by default and clearly marked. 6. Sandboxing and safety Every child evaluation runs in an isolated workspace with no host secrets, no default network, resource limits, timeouts, filesystem allowlist, command logs, diff inspection, and reject rules. Detect and block attempts to read env vars/secrets, alter scoring/tests/evaluator, disable tests, contact external endpoints during evaluation, add hidden payloads/obfuscation, generate malware/exploit/credential-theft/persistence/evasion logic, or bypass safety controls. Document this in SECURITY.md. This is for benign coding-agent research only. 7. CLI Implement: open-dgm init open-dgm benchmark list open-dgm agent run --task TASK_ID open-dgm evolve --generations N --children-per-generation K --benchmark toy-python open-dgm archive list open-dgm archive show VARIANT_ID open-dgm archive lineage VARIANT_ID --format mermaid open-dgm compare A B open-dgm report EXPERIMENT_ID `evolve` should load archive state, select parents, create child workspaces, generate self-patches, apply them, run safety/static/test gates, evaluate benchmarks, archive accepted children, and produce a report. 8. Reports, reproducibility, docs Each experiment outputs Markdown and JSON reports, archive snapshot, lineage graph, score trajectory chart, cost/token summary, per-task table, notable patches, safety incidents, and exact reproduction command. Add structured tracing for model calls, tool calls, shell commands, patches, tests, and evaluator decisions. Ship Dockerfile/devcontainer, pyproject.toml, Makefile/task runner, CI, pytest suite, example fixtures, seeded deterministic mode, mock-provider demo, and at least one completed sample DGM run. Docs must explain DGM vs proof-based Gödel machines, archive-based open-ended search, parent selection, self-patching, anti-cheating, extension points, lineage inspection, reproducibility, security model, limitations, and non-goals. Milestones: 1. Scaffold, architecture docs, provider interface, mock provider, CLI skeleton, toy benchmark interface, CI, SECURITY.md. 2. Baseline coding-agent loop with file inspection, patching, tests, retries, traces, and deterministic toy tasks. 3. Immutable archive, reconstruction, comparison, lineage graph export, and sample archived runs. 4. Self-improvement operator: failure diagnosis, self-patch generation, child workspace, safety gates, evaluation, archive update. 5. Multi-generation evolution runner with parent-selection policies, budgets, parallel child evaluation where safe, and reports. 6. Mini SWE-style and Polyglot-style adapters plus tamper-detection tests. 7. Public demo showing baseline, evolved generations, archive, lineage graph, score chart, best variant reconstruction, and reproduction steps. 8. Optional extensions: clade metaproductivity, statistical acceptance tests, group-evolving agents, runtime self-improvement, and VeRO-style optimizer-of-agents evaluation. Acceptance criteria: - Open-source repo with passing CI. - Full local deterministic demo works without paid APIs. - Real-model mode documented but optional. - Complete sample DGM evolution run included. - Archive is inspectable and variants are reconstructable. - Evaluator tampering is detected in tests. - Safety policy is enforced by default. - Docs allow researchers to add providers, benchmarks, or parent-selection policies. - No exaggerated claims of formal optimality, AGI, or unsafe autonomy. Final deliverable: a working open-source DGM framework that makes recursive coding-agent self-improvement concrete, auditable, benchmarked, reproducible, and responsibly sandboxed.

Prompt attachments
FileKindTypeSize
Darwin G__del Machines.pdfdocumentapplication/pdf230,776 B

Fund this project

Sign in

Backing funds AI build attempts. Outputs may vary. Funding is subject to our Terms of Service and Refund Policy.

Outputs / artifacts

Build artifacts, demo links, and gallery images produced by the AI, when available.

Outputs will appear here when the AI produces build artifacts.

Public ledger

Credits in, credits spent, and project-pool movements.

Credits spent 2026-10-10 08:53 UTC
Project pool

gsd build (VM) 163533in/148out on anthropic/claude-fable-5

Credits spent165 credits
Credits spent 2026-10-10 08:53 UTC
Project pool

gsd build (VM) 159730in/3789out on anthropic/claude-fable-5

Credits spent179 credits
Credits spent 2026-10-10 08:52 UTC
Project pool

gsd build (VM) 157653in/2063out on anthropic/claude-fable-5

Credits spent168 credits
Credits spent 2026-10-10 08:51 UTC
Project pool

gsd build (VM) 156841in/883out on anthropic/claude-fable-5

Credits spent162 credits
Credits spent 2026-10-10 08:51 UTC
Project pool

gsd build (VM) 153732in/3217out on anthropic/claude-fable-5

Credits spent170 credits
Credits spent 2026-10-10 08:50 UTC
Project pool

gsd build (VM) 153004in/1214out on anthropic/claude-fable-5

Credits spent160 credits
Credits spent 2026-10-10 08:50 UTC
Project pool

gsd build (VM) 150201in/2789out on anthropic/claude-fable-5

Credits spent165 credits
Credits spent 2026-10-10 08:49 UTC
Project pool

gsd build (VM) 149814in/444out on anthropic/claude-fable-5

Credits spent153 credits
Credits spent 2026-10-10 08:49 UTC
Project pool

gsd build (VM) 149493in/244out on anthropic/claude-fable-5

Credits spent151 credits
Credits spent 2026-10-10 08:49 UTC
Project pool

gsd build (VM) 149009in/470out on anthropic/claude-fable-5

Credits spent152 credits
Credits spent 2026-10-10 08:49 UTC
Project pool

gsd build (VM) 147899in/168out on anthropic/claude-fable-5

Credits spent149 credits
Credits spent 2026-10-10 08:48 UTC
Project pool

gsd build (VM) 147735in/148out on anthropic/claude-fable-5

Credits spent149 credits
Credits spent 2026-10-10 08:48 UTC
Project pool

gsd build (VM) 143186in/4535out on anthropic/claude-fable-5

Credits spent166 credits
Credits spent 2026-10-10 08:47 UTC
Project pool

gsd build (VM) 142842in/410out on anthropic/claude-fable-5

Credits spent145 credits
Credits spent 2026-10-10 08:47 UTC
Project pool

gsd build (VM) 138724in/4249out on anthropic/claude-fable-5

Credits spent160 credits
Credits spent 2026-10-10 08:46 UTC
Project pool

gsd build (VM) 131238in/7416out on anthropic/claude-fable-5

Credits spent169 credits
Credits spent 2026-10-10 08:45 UTC
Project pool

gsd build (VM) 130292in/545out on anthropic/claude-fable-5

Credits spent134 credits
Credits spent 2026-10-10 08:44 UTC
Project pool

gsd build (VM) 129966in/273out on anthropic/claude-fable-5

Credits spent132 credits
Credits spent 2026-10-10 08:44 UTC
Project pool

gsd build (VM) 129799in/151out on anthropic/claude-fable-5

Credits spent131 credits
Credits spent 2026-10-10 08:44 UTC
Project pool

gsd build (VM) 127211in/2574out on anthropic/claude-fable-5

Credits spent141 credits
Credits spent 2026-10-10 08:44 UTC
Project pool

gsd build (VM) 124381in/2816out on anthropic/claude-fable-5

Credits spent139 credits
Credits spent 2026-10-10 08:43 UTC
Project pool

gsd build (VM) 123577in/1011out on anthropic/claude-fable-5

Credits spent129 credits
Credits spent 2026-10-10 08:43 UTC
Project pool

gsd build (VM) 122950in/943out on anthropic/claude-fable-5

Credits spent128 credits
Credits spent 2026-10-10 08:42 UTC
Project pool

gsd build (VM) 119222in/3714out on anthropic/claude-fable-5

Credits spent138 credits
Credits spent 2026-10-10 08:42 UTC
Project pool

gsd build (VM) 118510in/1019out on anthropic/claude-fable-5

Credits spent124 credits
Credits spent 2026-10-10 08:41 UTC
Project pool

gsd build (VM) 115872in/2758out on anthropic/claude-fable-5

Credits spent130 credits
Credits spent 2026-10-10 08:41 UTC
Project pool

gsd build (VM) 112300in/3793out on anthropic/claude-fable-5

Credits spent132 credits
Credits spent 2026-10-10 08:40 UTC
Project pool

gsd build (VM) 110839in/2696out on anthropic/claude-fable-5

Credits spent125 credits
Credits spent 2026-10-10 08:39 UTC
Project pool

gsd build (VM) 109314in/3494out on anthropic/claude-fable-5

Credits spent127 credits
Credits spent 2026-10-10 08:38 UTC
Project pool

gsd build (VM) 108355in/5024out on anthropic/claude-fable-5

Credits spent134 credits
Credits spent 2026-10-10 08:37 UTC
Project pool

gsd build (VM) 108049in/234out on anthropic/claude-fable-5

Credits spent110 credits
Credits spent 2026-10-10 08:36 UTC
Project pool

gsd build (VM) 107322in/806out on anthropic/claude-fable-5

Credits spent112 credits
Credits spent 2026-10-10 08:36 UTC
Project pool

gsd build (VM) 107074in/168out on anthropic/claude-fable-5

Credits spent108 credits
Credits spent 2026-10-10 08:36 UTC
Project pool

gsd build (VM) 106680in/866out on anthropic/claude-fable-5

Credits spent112 credits
Credits spent 2026-10-10 08:36 UTC
Project pool

gsd build (VM) 106239in/145out on anthropic/claude-fable-5

Credits spent107 credits
Credits spent 2026-10-10 08:36 UTC
Project pool

gsd build (VM) 105977in/402out on anthropic/claude-fable-5

Credits spent108 credits
Credits spent 2026-10-10 08:35 UTC
Project pool

gsd build (VM) 105787in/176out on anthropic/claude-fable-5

Credits spent107 credits
Credits spent 2026-10-10 08:35 UTC
Project pool

gsd build (VM) 105304in/641out on anthropic/claude-fable-5

Credits spent109 credits
Credits spent 2026-10-10 08:35 UTC
Project pool

gsd build (VM) 104589in/139out on anthropic/claude-fable-5

Credits spent106 credits
Credits spent 2026-10-10 08:35 UTC
Project pool

gsd build (VM) 104327in/168out on anthropic/claude-fable-5

Credits spent106 credits
Credits spent 2026-10-10 08:34 UTC
Project pool

gsd build (VM) 104022in/341out on anthropic/claude-fable-5

Credits spent106 credits
Credits spent 2026-10-10 08:34 UTC
Project pool

gsd build (VM) 103393in/693out on anthropic/claude-fable-5

Credits spent107 credits
Credits spent 2026-10-10 08:34 UTC
Project pool

gsd build (VM) 102668in/142out on anthropic/claude-fable-5

Credits spent104 credits
Credits spent 2026-10-10 08:33 UTC
Project pool

gsd build (VM) 101916in/167out on anthropic/claude-fable-5

Credits spent103 credits
Credits spent 2026-10-10 08:33 UTC
Project pool

gsd build (VM) 98000in/3946out on anthropic/claude-fable-5

Credits spent118 credits
Credits spent 2026-10-10 08:32 UTC
Project pool

gsd build (VM) 95647in/2339out on anthropic/claude-fable-5

Credits spent108 credits
Credits spent 2026-10-10 08:31 UTC
Project pool

gsd build (VM) 94540in/1093out on anthropic/claude-fable-5

Credits spent101 credits
Credits spent 2026-10-10 08:31 UTC
Project pool

gsd build (VM) 92984in/1782out on anthropic/claude-fable-5

Credits spent102 credits
Credits spent 2026-10-10 08:30 UTC
Project pool

gsd build (VM) 88694in/4544out on anthropic/claude-fable-5

Credits spent112 credits
Credits spent 2026-10-10 08:29 UTC
Project pool

gsd build (VM) 87799in/825out on anthropic/claude-fable-5

Credits spent92 credits
Credits spent 2026-10-10 08:28 UTC
Project pool

gsd build (VM) 86243in/1542out on anthropic/claude-fable-5

Credits spent94 credits
Credits spent 2026-10-10 08:28 UTC
Project pool

gsd build (VM) 83621in/5534out on anthropic/claude-fable-5

Credits spent112 credits
Credits spent 2026-10-10 08:27 UTC
Project pool

gsd build (VM) 82338in/647out on anthropic/claude-fable-5

Credits spent86 credits
Credits spent 2026-10-10 08:26 UTC
Project pool

gsd build (VM) 82145in/179out on anthropic/claude-fable-5

Credits spent84 credits
Credits spent 2026-10-10 08:26 UTC
Project pool

gsd build (VM) 81919in/202out on anthropic/claude-fable-5

Credits spent83 credits
Credits spent 2026-10-10 08:26 UTC
Project pool

gsd build (VM) 76231in/6615out on anthropic/claude-fable-5

Credits spent110 credits
Credits spent 2026-10-10 08:24 UTC
Project pool

gsd build (VM) 74550in/5390out on anthropic/claude-fable-5

Credits spent102 credits
Credits spent 2026-10-10 08:22 UTC
Project pool

gsd build (VM) 71352in/5841out on anthropic/claude-fable-5

Credits spent101 credits
Credits spent 2026-10-10 08:21 UTC
Project pool

gsd build (VM) 69645in/7248out on anthropic/claude-fable-5

Credits spent106 credits
Credits spent 2026-10-10 08:19 UTC
Project pool

gsd build (VM) 69524in/420out on anthropic/claude-fable-5

Credits spent72 credits
Credits spent 2026-10-10 08:19 UTC
Project pool

gsd build (VM) 68292in/6751out on anthropic/claude-fable-5

Credits spent103 credits
Credits spent 2026-10-10 08:17 UTC
Project pool

gsd build (VM) 68066in/201out on anthropic/claude-fable-5

Credits spent70 credits
Credits spent 2026-10-10 08:17 UTC
Project pool

gsd build (VM) 67823in/193out on anthropic/claude-fable-5

Credits spent69 credits
Credits spent 2026-10-10 08:17 UTC
Project pool

gsd build (VM) 67485in/186out on anthropic/claude-fable-5

Credits spent69 credits
Credits spent 2026-10-10 08:17 UTC
Project pool

gsd build (VM) 67351in/123out on anthropic/claude-fable-5

Credits spent68 credits
Credits spent 2026-10-10 08:17 UTC
Project pool

gsd build (VM) 67102in/748out on anthropic/claude-fable-5

Credits spent71 credits
Credits spent 2026-10-10 08:16 UTC
Project pool

gsd build (VM) 58051in/687out on anthropic/claude-fable-5

Credits spent62 credits
Credits spent 2026-10-10 08:16 UTC
Project pool

gsd build (VM) 48344in/248out on anthropic/claude-fable-5

Credits spent50 credits
Credits spent 2026-10-10 08:16 UTC
Project pool

gsd build (VM) 25750in/18out on anthropic/claude-fable-5

Credits spent26 credits
Credits spent 2026-10-10 08:15 UTC
Project pool

gsd build (VM) 104541in/851out on anthropic/claude-fable-5

Credits spent109 credits
Credits spent 2026-10-10 08:15 UTC
Project pool

gsd build (VM) 99304in/109out on anthropic/claude-fable-5

Credits spent100 credits
Credits spent 2026-10-10 08:15 UTC
Project pool

gsd build (VM) 97012in/337out on anthropic/claude-fable-5

Credits spent99 credits
Credits spent 2026-10-10 08:15 UTC
Project pool

gsd build (VM) 96427in/341out on anthropic/claude-fable-5

Credits spent99 credits
Credits spent 2026-10-10 08:15 UTC
Project pool

gsd build (VM) 51557in/1940out on anthropic/claude-fable-5

Credits spent62 credits
Credits spent 2026-10-10 08:14 UTC
Project pool

gsd build (VM) 51341in/202out on anthropic/claude-fable-5

Credits spent53 credits
Credits spent 2026-10-10 08:14 UTC
Project pool

gsd build (VM) 49816in/2138out on anthropic/claude-fable-5

Credits spent61 credits
Credits spent 2026-10-10 08:13 UTC
Project pool

gsd build (VM) 48814in/2017out on anthropic/claude-fable-5

Credits spent59 credits
Credits spent 2026-10-10 08:13 UTC
Project pool

gsd build (VM) 43945in/8577out on anthropic/claude-fable-5

Credits spent87 credits
Credits spent 2026-10-10 08:11 UTC
Project pool

gsd build (VM) 40321in/123out on anthropic/claude-fable-5

Credits spent41 credits
Credits spent 2026-10-10 08:11 UTC
Project pool

gsd build (VM) 40115in/901out on anthropic/claude-fable-5

Credits spent45 credits
Credits spent 2026-10-10 08:11 UTC
Project pool

gsd build (VM) 39791in/5727out on anthropic/claude-fable-5

Credits spent69 credits
Credits spent 2026-10-10 08:10 UTC
Project pool

gsd build (VM) 39610in/207out on anthropic/claude-fable-5

Credits spent41 credits
Credits spent 2026-10-10 08:09 UTC
Project pool

gsd build (VM) 23314in/358out on anthropic/claude-fable-5

Credits spent26 credits
Credits spent 2026-10-10 08:09 UTC
Project pool

gsd build (VM) 93434in/1369out on anthropic/claude-fable-5

Credits spent101 credits
Credits spent 2026-10-10 08:09 UTC
Project pool

gsd build (VM) 81324in/66out on anthropic/claude-fable-5

Credits spent82 credits
Credits spent 2026-10-10 08:09 UTC
Project pool

gsd build (VM) 80874in/250out on anthropic/claude-fable-5

Credits spent83 credits
Credits spent 2026-10-10 08:09 UTC
Project pool

gsd build (VM) 76655in/4205out on anthropic/claude-fable-5

Credits spent98 credits
Credits spent 2026-10-10 08:08 UTC
Project pool

gsd build (VM) 76016in/579out on anthropic/claude-fable-5

Credits spent79 credits
Credits spent 2026-10-10 08:08 UTC
Project pool

gsd build (VM) 75147in/1093out on anthropic/claude-fable-5

Credits spent81 credits
Credits spent 2026-10-10 08:07 UTC
Project pool

gsd build (VM) 74916in/208out on anthropic/claude-fable-5

Credits spent76 credits
Credits spent 2026-10-10 08:07 UTC
Project pool

gsd build (VM) 74548in/187out on anthropic/claude-fable-5

Credits spent76 credits
Credits spent 2026-10-10 08:07 UTC
Project pool

gsd build (VM) 70954in/4786out on anthropic/claude-fable-5

Credits spent95 credits
Credits spent 2026-10-10 08:06 UTC
Project pool

gsd build (VM) 65012in/180out on anthropic/claude-fable-5

Credits spent66 credits
Credits spent 2026-10-10 08:06 UTC
Project pool

gsd build (VM) 63739in/464out on anthropic/claude-fable-5

Credits spent67 credits
Credits spent 2026-10-10 08:06 UTC
Project pool

gsd build (VM) 63148in/323out on anthropic/claude-fable-5

Credits spent65 credits
Credits spent 2026-10-10 08:05 UTC
Project pool

gsd build (VM) 58749in/180out on anthropic/claude-fable-5

Credits spent60 credits
Credits spent 2026-10-10 08:05 UTC
Project pool

gsd build (VM) 37463in/14out on anthropic/claude-fable-5

Credits spent38 credits
Credits spent 2026-10-10 08:05 UTC
Project pool

gsd build (VM) 10490in/19out on anthropic/claude-fable-5

Credits spent11 credits
Credits spent 2026-10-10 08:05 UTC
Project pool

gsd build (VM) 790in/8out on anthropic/claude-fable-5

Credits spent1 credits
Credits spent 2026-10-10 08:05 UTC
Project pool

gsd build (VM) 10in/16out on anthropic/claude-fable-5

Credits spent1 credits

Activity timeline

Public project events in plain language.

Public build log

Live build output and spend messages.

2026-10-10 08:53 UTCBuild output was not hostable (no start script/static entry, or the server did not answer on :80).
2026-10-10 08:53 UTCBuild produced 8 source file(s); finalizing.
2026-10-10 08:53 UTCError: requires more credits to continue this build
2026-10-10 08:53 UTCFAILED tests/test_safety.py::TestSandbox::test_write_outside_sandbox_blocked
2026-10-10 08:53 UTC=========================== short test summary info ============================
2026-10-10 08:53 UTCtests/test_safety.py:146: AssertionError
2026-10-10 08:53 UTCE + where 0 = SandboxResult(returncode=0, stdout='', stderr='', timed_out=False, safety_flags=[]).returncode
2026-10-10 08:53 UTCE AssertionError: assert 0 != 0
2026-10-10 08:53 UTC> assert result.returncode != 0
2026-10-10 08:53 UTC result = run_sandboxed([sys.executable, "-c", code], tmp_path / "box", Tracer())
2026-10-10 08:53 UTC code = f"open({str(outside)!r}, 'w').write('escaped')"
2026-10-10 08:53 UTC outside = tmp_path.parent / "escape-marker.txt"
2026-10-10 08:53 UTC def test_write_outside_sandbox_blocked(self, tmp_path: Path) -> None:
2026-10-10 08:53 UTCtmp_path = PosixPath('/tmp/pytest-of-root/pytest-14/test_write_outside_sandbox_blo0')
2026-10-10 08:53 UTCself = <test_safety.TestSandbox object at 0x77151559e1e0>
2026-10-10 08:53 UTC$ /opt/fpbuild/venv/bin/python -m pytest tests/test_safety.py -q 2>&1 | tail -15
2026-10-10 08:53 UTCWrote file successfully.
2026-10-10 08:53 UTC← Write tests/test_safety.py
2026-10-10 08:53 UTCWrote file successfully.
2026-10-10 08:53 UTC← Write SECURITY.md
2026-10-10 08:53 UTCWrote file successfully.
2026-10-10 08:15 UTCBuilding all phases (gsd autonomous)…
2026-10-10 08:15 UTC`/gsd-plan-phase 1` — plan the first phase
2026-10-10 08:15 UTC**Phase 1: Foundation & Scaffold** — clean Python package, CI, Docker/devcontainer, task runner, `open-dgm init`
2026-10-10 08:15 UTC## ▶ Next Up
2026-10-10 08:15 UTC---
2026-10-10 08:15 UTC- One discrepancy caught and fixed: the requirements coverage header initially said 50; actual count is 58 — traceability table now maps all 58.
2026-10-10 08:15 UTC- **Phase structure**: Foundation → Baseline Agent & Providers → Benchmark Harness → Immutable Archive → Sandboxing & Safety → Self-Improvement & Parent Selection → Evolution Runner & Reports → Demo, Docs & Showcase. Safety (Phase 5) deliberately lands *before* self-patching (Phase 6), which depends on both the archive and the sandbox.
2026-10-10 08:15 UTC- **Config defaults** (silent, per headless override): YOLO mode, standard granularity, parallel execution, commit docs on, adaptive models, research/plan-check/verifier on, drift guard on.
2026-10-10 08:15 UTC- **All 5 locked owner decisions honored**: full source self-patching, hybrid demo (rule-based mutator + recorded real-model showcase), process-level sandboxing only, record-all/admit-conditionally archive, all three benchmark tiers locally (Polyglot = Py+JS).
2026-10-10 08:05 UTCPlanning the roadmap from your locked brief…
2026-10-10 08:05 UTCToolchain: opencode 1.18.35, [email protected]
2026-10-10 08:05 UTCWorkspace repair: .opencode/package.json was unreadable (damaged in a VM snapshot), moved aside.
2026-10-10 08:05 UTCMetering proxy reachable — build VM online.
2026-10-10 07:59 UTCBuild VM provisioning (Vultr 1cf0cde3). Toolchain + planning starting.
2026-10-09 07:33 UTCBuild did not produce a working result; unspent credits refunded to backers. VM state backed up for restart.
2026-10-09 07:33 UTCBuild failed: gsd produced no roadmap from IDEA.md (see gsd-plan.log)
2026-10-09 07:33 UTCopencode log: no error lines found under /opt/fpbuild/.data/opencode/log
2026-10-09 07:33 UTC}
2026-10-09 07:33 UTC }
2026-10-09 07:33 UTC "ref": "err_10813416"
2026-10-09 07:33 UTC "message": "Unexpected server error. Check server logs for details.",
2026-10-09 07:33 UTC "data": {
2026-10-09 07:33 UTC "name": "UnknownError",
2026-10-09 07:33 UTCError: {
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:49.035Z level=ERROR run=8596c5f6 message=failed ref=err_10813416 error="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'" cause="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'\n at ToolRegistry.state (/$bunfs/root/chunk-pck0885g.js:2:4646)\n at ToolRegistry.state (definition) (/$bunfs/root/chunk-m5jsweza.js:479:867)\n at
2026-10-09 07:33 UTC> build · anthropic/claude-fable-5
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:43.003Z level=WARN run=8596c5f6 message="background dependency install failed" dir=/opt/fpbuild/workspace/.opencode error="Cause([Fail(NpmInstallFailedError (cause: TL: Invalid package.json: TL: JSON Parse error: Unrecognized token '\u0000' while parsing '\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\
2026-10-09 07:33 UTCopencode: "ref": "err_10813416"
2026-10-09 07:33 UTCopencode: timestamp=2026-10-09T07:33:49.035Z level=ERROR run=8596c5f6 message=failed ref=err_10813416 error="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'" cause="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'\n at ToolRegistry.state (/$bunfs/root/chunk-pck0885g.js:2:4646)\n at ToolRegistry.state (definition) (/$bunfs/root/chunk-m5jsweza.js:479:86
2026-10-09 07:33 UTCopencode: timestamp=2026-10-09T07:33:43.003Z level=WARN run=8596c5f6 message="background dependency install failed" dir=/opt/fpbuild/workspace/.opencode error="Cause([Fail(NpmInstallFailedError (cause: TL: Invalid package.json: TL: JSON Parse error: Unrecognized token '\u0000' while parsing '\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0
2026-10-09 07:33 UTC}
2026-10-09 07:33 UTC }
2026-10-09 07:33 UTC "ref": "err_10813416"
2026-10-09 07:33 UTC "message": "Unexpected server error. Check server logs for details.",
2026-10-09 07:33 UTC "data": {
2026-10-09 07:33 UTC "name": "UnknownError",
2026-10-09 07:33 UTCError: {
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:49.035Z level=ERROR run=8596c5f6 message=failed ref=err_10813416 error="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'" cause="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'\n at ToolRegistry.state (/$bunfs/root/chunk-pck0885g.js:2:4646)\n at ToolRegistry.state (definition) (/$bunfs/root/chunk-m5jsweza.js:479:867)\n at
2026-10-09 07:33 UTC> build · anthropic/claude-fable-5
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:43.003Z level=WARN run=8596c5f6 message="background dependency install failed" dir=/opt/fpbuild/workspace/.opencode error="Cause([Fail(NpmInstallFailedError (cause: TL: Invalid package.json: TL: JSON Parse error: Unrecognized token '\u0000' while parsing '\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\
2026-10-09 07:33 UTCPlanning hit an opencode server error; retrying as a plain prompt from fresh opencode state.
2026-10-09 07:33 UTCWorkspace: .git IDEA.md .opencode ; git: ?? .opencode/ ?? IDEA.md ; locks:
2026-10-09 07:33 UTCopencode log: no error lines found under /opt/fpbuild/.data/opencode/log
2026-10-09 07:33 UTCopencode: "ref": "err_caed9eac"
2026-10-09 07:33 UTCopencode: timestamp=2026-10-09T07:33:38.983Z level=ERROR run=beed5d31 message=failed ref=err_caed9eac error="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'" cause="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'\n at ToolRegistry.state (/$bunfs/root/chunk-pck0885g.js:2:4646)\n at ToolRegistry.state (definition) (/$bunfs/root/chunk-m5jsweza.js:479:86
2026-10-09 07:33 UTCopencode: timestamp=2026-10-09T07:33:33.084Z level=WARN run=beed5d31 message="background dependency install failed" dir=/opt/fpbuild/workspace/.opencode error="Cause([Fail(NpmInstallFailedError (cause: TL: Invalid package.json: TL: JSON Parse error: Unrecognized token '\u0000' while parsing '\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0
2026-10-09 07:33 UTC}
2026-10-09 07:33 UTC }
2026-10-09 07:33 UTC "ref": "err_caed9eac"
2026-10-09 07:33 UTC "message": "Unexpected server error. Check server logs for details.",
2026-10-09 07:33 UTC "data": {
2026-10-09 07:33 UTC "name": "UnknownError",
2026-10-09 07:33 UTCError: {
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:38.983Z level=ERROR run=beed5d31 message=failed ref=err_caed9eac error="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'" cause="ResolveMessage: Cannot find module '@opencode-ai/plugin' from '/opt/fpbuild/workspace/.opencode/tools/ask_decision.ts'\n at ToolRegistry.state (/$bunfs/root/chunk-pck0885g.js:2:4646)\n at ToolRegistry.state (definition) (/$bunfs/root/chunk-m5jsweza.js:479:867)\n at
2026-10-09 07:33 UTCtimestamp=2026-10-09T07:33:33.084Z level=WARN run=beed5d31 message="background dependency install failed" dir=/opt/fpbuild/workspace/.opencode error="Cause([Fail(NpmInstallFailedError (cause: TL: Invalid package.json: TL: JSON Parse error: Unrecognized token '\u0000' while parsing '\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\u0000\
2026-10-09 07:33 UTCPlanning the roadmap from your locked brief…
2026-10-09 07:33 UTCToolchain: opencode 1.18.35, [email protected]
2026-10-09 07:33 UTCMetering proxy reachable — build VM online.
2026-10-09 07:27 UTCBuild VM provisioning (Vultr 4aab119a). Toolchain + planning starting.
2026-10-09 07:05 UTCBuild did not produce a working result; unspent credits refunded to backers. VM state backed up for restart.
2026-10-09 07:05 UTCBuild failed: gsd produced no roadmap from IDEA.md (see gsd-plan.log)
2026-10-09 07:05 UTCopencode log: no error lines found under /opt/fpbuild/.data/opencode/log
2026-10-09 07:05 UTC}
2026-10-09 07:05 UTC }
2026-10-09 07:05 UTC "ref": "err_47cec72f"
2026-10-09 07:05 UTC "message": "Unexpected server error. Check server logs for details.",
2026-10-09 07:05 UTC "data": {
2026-10-09 07:05 UTC "name": "UnknownError",
2026-10-09 07:05 UTCError: {
2026-10-09 07:05 UTC}
2026-10-09 07:05 UTC }
2026-10-09 07:05 UTC "ref": "err_47cec72f"
2026-10-09 07:05 UTC "message": "Unexpected server error. Check server logs for details.",
2026-10-09 07:05 UTC "data": {
2026-10-09 07:05 UTC "name": "UnknownError",
2026-10-09 07:05 UTCError: {
2026-10-09 07:05 UTCPlanning hit an opencode server error; retrying once with fresh opencode state.
2026-10-09 07:05 UTCopencode log: timestamp=2026-10-09T07:04:08.214Z level=ERROR run=f6365c2e message="stream error" providerID=openrouter modelID=google/gemini-nano-banana-2.1 session.id=ses_ee0861cdaffeFoSC17WvE8T76o small=true agent=title mode=primary error.error="AI_RetryError: Failed after 2 attempts with non-retryable error: 'Not Found'"
2026-10-09 07:05 UTCopencode log: timestamp=2026-10-09T07:04:06.073Z level=ERROR run=f6365c2e message="stream error" providerID=openrouter modelID=anthropic/claude-fable-5 session.id=ses_ee0861cdaffeFoSC17WvE8T76o small=false agent=build mode=primary error.error="AI_APICallError: Bad Gateway"

Comments

5 public comments from the community.

Sign in to comment.
Valentin Hirschi2026-10-18 10:23 UTC

@Admins I like the idea of fablepool very much, but clearly you didn't put enough effort to make it a reality. This build has been stuck for weeks now and you don't bother to comment here and update us. It's not like you have 1000s of projects currently opened atm... Too bad...

DK2026-10-08 00:57 UTC

I’ll message admins

Valentin Hirschi2026-10-06 19:16 UTC

@admins: please fix this: """ Build VM provisioning failed: Vultr POST /instances -> 401: {"error":"Unauthorized IP address: 108.61.221.133","status":401} """ Or refund, but in cash.

Valentin Hirschi2026-09-08 21:52 UTC

Not anymore! Let's go for: """ D: Hybrid — rule-based mutator for CI/demo plus a committed recorded real-model run as the showcase artifact; most convincing but largest scope """

DK2026-08-23 20:45 UTC

Project appears in a permanent broken state

Related projects

More in Dev Tools.

6
Build an open protocol for user-owned AI memory

by Daniel May · GPT-5.5 · 5 months ago

AI is starting to remember us. Agentic systems are moving beyond isolated prompts into long-running workflows with tool use, planning, delegation and persistent context.…

Dev Tools Shipped
5,015 credits of 8,388 credits 60% full
8funders
5watchers
0comments
completed

Recommended next

Almost-funded projects moving across the market.

No matching projects yet.

More by this creator

Other public projects from the same maker.

No matching projects yet.