2026-05-14 11:32:25 +02:00
|
|
|
[project]
|
|
|
|
|
name = "infospace-bench"
|
|
|
|
|
version = "0.1.0"
|
|
|
|
|
description = "Application-layer workspace for concrete structured knowledge spaces."
|
IB-WP-0014: archive integration with artifact-store (T01+T02)
Reframe IB-WP-0014 from "in-repo S3/git backend adapters" to "durable archive
surface via artifact-store". The live infospace stays in a local working folder;
finalized snapshots are bundled into content-addressed artifact-store packages.
- New module infospace_bench.archive: archive_infospace(), list_archives(),
ArchiveRecord. Self-bootstraps a SQLite + local-FS registry under
output/archives/.store/ when no Registry is passed in.
- New output/archives/index.yaml records each archive event (package id,
manifest digest, retention class, included paths, file count, note).
- artifactstore added as a path dep; Python floor bumped to 3.12 to match.
- Makefile for venv-based dev setup; stack-and-commands.md updated.
- tests/test_archive.py covers index write, list, recursive-capture guard,
caller-supplied include, and empty-include error. Full suite 65 passed.
Remaining tasks (T03 list CLI, T04 restore, T05 docs) tracked in the workplan.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-17 11:30:49 +02:00
|
|
|
requires-python = ">=3.12"
|
2026-05-14 14:53:16 +02:00
|
|
|
dependencies = [
|
|
|
|
|
"PyYAML>=6",
|
|
|
|
|
"markitect-tool @ file:///home/worsch/markitect-tool",
|
IB-WP-0014: archive integration with artifact-store (T01+T02)
Reframe IB-WP-0014 from "in-repo S3/git backend adapters" to "durable archive
surface via artifact-store". The live infospace stays in a local working folder;
finalized snapshots are bundled into content-addressed artifact-store packages.
- New module infospace_bench.archive: archive_infospace(), list_archives(),
ArchiveRecord. Self-bootstraps a SQLite + local-FS registry under
output/archives/.store/ when no Registry is passed in.
- New output/archives/index.yaml records each archive event (package id,
manifest digest, retention class, included paths, file count, note).
- artifactstore added as a path dep; Python floor bumped to 3.12 to match.
- Makefile for venv-based dev setup; stack-and-commands.md updated.
- tests/test_archive.py covers index write, list, recursive-capture guard,
caller-supplied include, and empty-include error. Full suite 65 passed.
Remaining tasks (T03 list CLI, T04 restore, T05 docs) tracked in the workplan.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-17 11:30:49 +02:00
|
|
|
"artifactstore @ file:///home/worsch/artifact-store",
|
IB-WP-0018-T01+T02+T05: routing bridge to llm-connect
T01 — task-type taxonomy. docs/routing-task-types.md names the five
generation stages as the default identity-mapped task types
(summarize-source, extract-entities, extract-relations,
evaluate-entity, synthesize-report) and records the recommended quality
floors per stage. The taxonomy explicitly does not decide which adapter
ships per task type, where the ledger lives, or what a quality score
means — those stay with the caller per the LLM-WP-0004 scope guardrail.
T02 — RoutingAssistedGenerationAdapter bridge in
src/infospace_bench/routing.py. Wraps any llm-connect RoutingPolicy or
AdaptiveRoutingPolicy as an infospace-bench AssistedGenerationAdapter:
maps stage_id -> task_type (overridable), resolves an LLMAdapter,
delegates execute_prompt with a configurable RunConfig, and surfaces
the resolved adapter id, task type, model, usage, and finish_reason
back on AssistedGenerationResult.metadata. Provider tag stays
back-compatible with the strings already used in run records and the
budget rollup (openrouter / claude_code / openai / gemini / mock /
routing).
T05 — eight tests in tests/test_routing_adapter.py cover: static-policy
per-stage resolution, stage_to_task_type overrides, default-mapping
completeness, fall-through for unmapped stage ids, the adaptive path
selecting the cheaper qualifying adapter when a quality_floor is set,
adaptive policy falling back to static when no floor is set, response
metadata round-trip with provider tagging, and estimated_cost_per_1k
pass-through.
Adds llm-connect as a path dependency on pyproject.toml and to the
pytest pythonpath. Static OpenRouter and fixture paths are unchanged;
this commit only adds the option of routing.
139 tests pass, 1 skipped (the OpenRouter live smoke, gated as before).
T03 (shadow-mode integration) and T04 (CLI + per-stage chosen-adapter
in the generation report) follow next.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-18 11:33:58 +02:00
|
|
|
"llm-connect @ file:///home/worsch/llm-connect",
|
2026-05-14 14:53:16 +02:00
|
|
|
]
|
2026-05-14 11:32:25 +02:00
|
|
|
|
|
|
|
|
[project.scripts]
|
|
|
|
|
infospace-bench = "infospace_bench.cli:main"
|
|
|
|
|
|
2026-05-14 19:33:22 +02:00
|
|
|
[tool.setuptools.package-data]
|
2026-05-17 19:54:30 +02:00
|
|
|
infospace_bench = ["profiles/**/*", "model_rates.yaml"]
|
2026-05-14 19:33:22 +02:00
|
|
|
|
2026-05-14 11:32:25 +02:00
|
|
|
[tool.pytest.ini_options]
|
IB-WP-0018-T01+T02+T05: routing bridge to llm-connect
T01 — task-type taxonomy. docs/routing-task-types.md names the five
generation stages as the default identity-mapped task types
(summarize-source, extract-entities, extract-relations,
evaluate-entity, synthesize-report) and records the recommended quality
floors per stage. The taxonomy explicitly does not decide which adapter
ships per task type, where the ledger lives, or what a quality score
means — those stay with the caller per the LLM-WP-0004 scope guardrail.
T02 — RoutingAssistedGenerationAdapter bridge in
src/infospace_bench/routing.py. Wraps any llm-connect RoutingPolicy or
AdaptiveRoutingPolicy as an infospace-bench AssistedGenerationAdapter:
maps stage_id -> task_type (overridable), resolves an LLMAdapter,
delegates execute_prompt with a configurable RunConfig, and surfaces
the resolved adapter id, task type, model, usage, and finish_reason
back on AssistedGenerationResult.metadata. Provider tag stays
back-compatible with the strings already used in run records and the
budget rollup (openrouter / claude_code / openai / gemini / mock /
routing).
T05 — eight tests in tests/test_routing_adapter.py cover: static-policy
per-stage resolution, stage_to_task_type overrides, default-mapping
completeness, fall-through for unmapped stage ids, the adaptive path
selecting the cheaper qualifying adapter when a quality_floor is set,
adaptive policy falling back to static when no floor is set, response
metadata round-trip with provider tagging, and estimated_cost_per_1k
pass-through.
Adds llm-connect as a path dependency on pyproject.toml and to the
pytest pythonpath. Static OpenRouter and fixture paths are unchanged;
this commit only adds the option of routing.
139 tests pass, 1 skipped (the OpenRouter live smoke, gated as before).
T03 (shadow-mode integration) and T04 (CLI + per-stage chosen-adapter
in the generation report) follow next.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-18 11:33:58 +02:00
|
|
|
pythonpath = ["src", "../markitect-tool/src", "../llm-connect"]
|
2026-05-14 11:32:25 +02:00
|
|
|
testpaths = ["tests"]
|