7b031179a2
Merge pull request 'Give each role its own view of the transcript (M1.2)' (#88) from transcript-views into main
dongar2026-10-02 03:18:09 +00:00
51e144e26b
Merge pull request 'Work items in the state store, with a transcript bound to each (M1.1)' (#87) from work-items into main
dongar2026-10-01 23:36:36 +00:00
83804fe02a
Put work items in the state store, and bind the log to them (M1.1)
dongarandClaude Opus 5.52026-10-01 23:20:08 +00:00
e6aaa4783d
Merge pull request 'Add a bring-up script and a transcript dump; let nisse-dev install on its own VM (M0.2)' (#86) from bringup-script into main
dev-agent2026-10-01 16:18:15 +00:00
a18321b4b0
Add a bring-up script and a transcript dump; let nisse-dev install on its own VM (M0.2)
dongarandClaude Opus 5.52026-10-01 16:16:32 +00:00
678466c176
Merge pull request 'Retire the old build spec, and point every reference at the architecture (M0.1)' (#85) from retire-old-spec into main
dongar2026-10-01 00:52:53 +00:00
1ad0b2147d
Retire the old build spec, and point every reference at the architecture (M0.1)
dongarandClaude Opus 5.52026-10-01 00:17:48 +00:00
890708594b
Merge pull request 'Propose the Development Plan' (#84) from development-plan into main
dongar2026-09-30 21:07:39 +00:00
ae6513fd26
Development Plan: say which open issues are superseded, and which stay open
dongarandClaude Opus 5.52026-09-30 21:06:13 +00:00
f53b2777bc
Propose the Development Plan, and mark the architecture agreed
dongarandClaude Opus 5.52026-09-30 19:47:19 +00:00
73a035dfcc
Merge pull request 'Draft: rewrite the architecture (docs/Architecture.md)' (#83) from architecture-rewrite into main
dongar2026-09-30 19:40:21 +00:00
d302d337d6
Merge pull request 'Redeliver an interrupted review comment, and close the issue on completion' (#76) from redeliver-review-and-close-issue into main
dongar2026-09-29 20:37:04 +00:00
848bd9f122
Redeliver an interrupted review comment, and close the issue on completion
dongarandClaude Opus 5.52026-09-29 20:30:14 +00:00
83a4bd72b5
Merge pull request 'Adopt an already-published PR instead of publishing again' (#75) from adopt-published-pr into main
dongar2026-09-29 20:14:47 +00:00
6fdec6922b
Merge pull request 'Bind approvals and feedback to the revision they were about' (#74) from bind-approvals-to-revisions into main
dongar2026-09-29 19:45:07 +00:00
6c83839f4d
Merge pull request 'Model the orchestrator in Quint, and reproduce its counterexamples' (#73) from quint-orchestrator-spec into main
dongar2026-09-29 18:56:29 +00:00
d79e399933
Model the orchestrator in Quint, and reproduce its counterexamples
dongarandClaude Opus 5.52026-09-29 18:51:59 +00:00
26dabc3d54
Merge pull request 'Match approvals and commands regardless of case and trailing punctuation' (#72) from case-insensitive-commands into main
dongar2026-09-29 18:00:27 +00:00
6afd0cc914
Match approvals and commands regardless of case and trailing punctuation
dongarandClaude Opus 5.52026-09-29 17:57:34 +00:00
b893fba342
Merge pull request 'Add the abandon command' (#71) from abandon-command into main
dongar2026-09-29 17:52:30 +00:00
4e8ebea26a
Merge pull request 'Keep Gitea threads watched across restarts and interruptions' (#68) from pr-open-handling into main
dongar2026-09-29 17:37:25 +00:00
ab932f2ef4
Merge pull request 'Don't assert the exact word in the LiteLLM smoke test' (#67) from smoke-test-no-exact-word into main
dongar2026-09-29 17:24:43 +00:00
40b59e77bf
Merge pull request 'Never leave the agent stranded in a working state' (#66) from no-stranded-states into main
dongar2026-09-29 17:21:42 +00:00
b077906631
Let the live milestone test expect a waiting state after execution
dongarandClaude Opus 5.52026-09-29 17:17:19 +00:00
1ccea20a10
Merge pull request 'Publish work the executor committed itself, and gate when there is none' (#63) from publish-executor-commits into main
dongar2026-09-29 16:23:19 +00:00
f97e459bc7
Publish work the executor committed itself, and gate when there is none
dongarandClaude Opus 5.52026-09-29 16:17:41 +00:00
8283e9eb1c
Merge pull request 'Handle leading mentions in Matrix messages' (#60) from matrix-mentions into main
dev-agent2026-09-29 16:01:08 +00:00
9620404a2a
Merge pull request 'Keep a pending question across restarts, and make the restart question answerable' (#57) from fix-await-input-restart into main
dongar2026-09-29 15:54:01 +00:00
e5450e3e77
Keep a pending question across restarts, and make the restart question answerable
dongarandClaude Opus 5.52026-09-29 15:03:37 +00:00
e5ed32bc2b
Merge pull request 'Use qwen3.8-27b as the example executor' (#56) from executor-qwen3.8-27b into main
dongar2026-09-29 14:20:25 +00:00
dfa08d4900
Merge pull request 'Benchmark the finalists on all eight tasks, and archive older results' (#55) from bench-results-harder-tasks into main
dongar2026-09-28 12:41:48 +00:00
4b6c48cc96
Benchmark the finalists on all eight tasks, and archive older results
dongarandClaude Opus 5.52026-09-28 03:14:40 +00:00
a891932739
Merge pull request 'Spec: prefer one local model for all roles; record host sharing, park vision' (#54) from spec-shared-inference-host into main
dongar2026-09-28 01:47:08 +00:00
2ab86bf0da
Spec: OpenSCAD work needs vision, in nisse-dev or its own agent
dongarandClaude Opus 5.52026-09-28 01:32:20 +00:00
f2c48d8b16
Merge pull request 'Keep runs with different tuning apart in the report, and raise max_turns to 40' (#52) from report-keeps-tuning-apart into main
dongar2026-09-27 22:43:26 +00:00
f881a18ca4
Keep runs with different tuning apart in the report, and raise max_turns to 40
dongarandClaude Opus 5.52026-09-27 22:19:57 +00:00
a77c84c128
Merge pull request 'Render one assistant message per turn, and make the request timeout a setting' (#51) from fix-transcript-and-timeout into main
dev-agent2026-09-27 22:24:50 +00:00
d49435d4e5
Render one assistant message per turn, and make the request timeout a setting
dongarandClaude Opus 5.52026-09-27 22:15:30 +00:00
f41cbb00cf
Merge pull request 'Record rate-limit waits in benchmark runs, and leave them out of time' (#50) from record-rate-limit-retries into main
dev-agent2026-09-27 21:23:56 +00:00
4e45dfe06a
Record rate-limit waits in benchmark runs, and leave them out of time
dongarandClaude Opus 5.52026-09-27 20:57:15 +00:00
a716d5ec8d
Merge pull request 'Benchmark seven new RTX 4090 models, and local-fast on the 4090' (#49) from bench-4090 into main
dev-agent2026-09-27 20:31:46 +00:00
1cc8df8dab
Merge pull request 'Point the example config at Opus 5.5, with Haiku as consultant' (#48) from example-opus-5-5 into main
dev-agent2026-09-27 20:31:34 +00:00
82f171c2f5
Benchmark seven new RTX 4090 models, and local-fast on the 4090
dongarandClaude Opus 5.52026-09-27 18:43:55 +00:00
3cb70c0656
Point the example config at Opus 5.5, with Haiku as consultant
dongarandClaude Opus 5.52026-09-25 20:22:49 +00:00
8a8d8dc27f
Merge pull request 'Add an executor benchmark that keeps its results' (#47) from benchmark into main
dongar2026-09-25 02:23:57 +00:00
0cccfe1e94
Add the local baseline: qwen3-8b-awq and qwen3-14b, 5 runs per task
dongarandClaude Opus 5.52026-09-25 01:25:13 +00:00
b2bbf70b7b
Merge pull request 'Enforce the tool-call contract, asking servers only where it works' (#45) from enforce-tool-calls into main
dongar2026-09-25 00:05:46 +00:00
b934c46714
Enforce the tool-call contract, asking servers only where it works
dongarandClaude Opus 5.52026-09-24 22:54:49 +00:00
389b71a5d1
Merge pull request 'Fit every model request to its context window' (#44) from context-budget into main
dongar2026-09-24 13:41:15 +00:00
31851b7396
Merge pull request 'Move configuration to YAML, and make the tuning knobs configurable' (#43) from yaml-config into main
dongar2026-09-23 23:49:44 +00:00
889615e3e2
Merge pull request 'Record the MCP server deployment decisions in the spec' (#36) from mcp-deployment into main
dongar2026-09-23 02:03:59 +00:00
b0d7adb70d
Merge pull request 'Discover a model's vendor from LiteLLM instead of its name' (#35) from alias-discovery into main
dongar2026-09-23 01:49:36 +00:00
e38adf40ca
Discover a model's vendor from LiteLLM instead of its name
dongarandClaude Opus 52026-09-23 01:30:09 +00:00
231b406393
Merge pull request 'Choose the LLM route by model, and add a capability matrix' (#34) from model-routing into main
dongar2026-09-23 01:47:34 +00:00
8cfa116aa1
Preserve delegation work through any upstream outage, not just budget
dongarandClaude Opus 52026-09-22 23:57:05 +00:00
8cfb45aed3
Enforce the token cap, and treat a spent budget as an interruption
dongarandClaude Opus 52026-09-22 23:54:57 +00:00
249bcb10fc
Detect truncated responses, and raise the output ceiling
dongarandClaude Opus 52026-09-22 23:45:28 +00:00
ed2f5dfc7f
Choose the LLM route by model, and add a capability matrix
dongarandClaude Opus 52026-09-22 23:20:38 +00:00
1da94d2581
Merge pull request 'Add the composition root and a runnable entry point' (#31) from agent-entrypoint into main
dongar2026-09-08 22:33:45 +00:00