docs(evals): one eval page named for the evals, naming aos-eval #957

Merged

coilysiren merged 2 commits from aos/claude/eval-docs-one-page into main

2026-08-18 16:14:07 +00:00

coilyco-ops commented

2026-08-18 15:58:23 +00:00

Member

The eval surface lived on a page called sirens-echo-board.md, so nothing in the docs index reads as the eval page, and it never named the tool.

Renames it to sirens-echo-eval.md, matching cmd/sirens-echo-eval, and rewrites it as the one page about evaluation here: the two layers, the triple, the pairing rule, the derived board, and the battery.

Adds the section that was missing. aos-eval is the shared grading half in agentic-os, and eval/boundaries.yaml is already in its declaration shape, so aos-eval boundaries derive reads it as-is. The board dataset is not: board-deep emits cases: keyed on clause/history/current where aos-eval reads dataset: keyed on role/test_type/prompt, so annotate, taxonomy, and export cannot read a board record until a profile reconciles them. Grading here is still local, and the page says so rather than implying adoption.

Two things stay local on purpose and the page records why: the source-drift check in scripts/boundaries.sh, which verifies an origin#fragment still resolves and aos-eval boundaries check does not do, and the battery, which gates a deployment rather than grading it.

Repoints AGENTS.md, both packs, five Go files, FEATURES, and the identity doc, and drops a doubled reference in the evaluation pack header.

pre-commit run --all-files and just test both pass.

🤖 Generated with Claude Code

The eval surface lived on a page called `sirens-echo-board.md`, so nothing in the docs index reads as **the** eval page, and it never named the tool. Renames it to `sirens-echo-eval.md`, matching `cmd/sirens-echo-eval`, and rewrites it as the one page about evaluation here: the two layers, the triple, the pairing rule, the derived board, and the battery. **Adds the section that was missing.** `aos-eval` is the shared grading half in agentic-os, and `eval/boundaries.yaml` is already in its declaration shape, so `aos-eval boundaries derive` reads it as-is. The board dataset is not: `board-deep` emits `cases:` keyed on `clause`/`history`/`current` where `aos-eval` reads `dataset:` keyed on `role`/`test_type`/`prompt`, so `annotate`, `taxonomy`, and `export` cannot read a board record until a profile reconciles them. Grading here is still local, and the page says so rather than implying adoption. Two things stay local on purpose and the page records why: the source-drift check in `scripts/boundaries.sh`, which verifies an `origin#fragment` still resolves and `aos-eval boundaries check` does not do, and the battery, which gates a deployment rather than grading it. Repoints `AGENTS.md`, both packs, five Go files, FEATURES, and the identity doc, and drops a doubled reference in the evaluation pack header. `pre-commit run --all-files` and `just test` both pass. 🤖 Generated with [Claude Code](https://claude.com/claude-code)

coilyco-ops added 1 commit

2026-08-18 15:58:23 +00:00

docs(evals): one eval page named for the evals, naming aos-eval

ci / image-build (pull_request) Successful in 40s

Details

ci / test (pull_request) Successful in 1m13s

Details

ci / publish-echo-image (pull_request) Has been skipped

Details

ci / publish-observed (pull_request) Has been skipped

Details

2093d07926

The eval surface was on a page called sirens-echo-board.md, so nothing in the
docs index reads as the eval page, and it never named the tool. Rename it to
sirens-echo-eval.md, matching cmd/sirens-echo-eval, and rewrite it as the one
page about evaluation here: the two layers, the triple, the pairing rule, the
derived board, and the battery.

Adds the section that was missing. aos-eval is the shared grading half in
agentic-os, and eval/boundaries.yaml is already in its declaration shape, so
`boundaries derive` reads it as-is. The board dataset is not: it keys on
clause/history/current where aos-eval reads role/test_type/prompt, so annotate,
taxonomy, and export cannot read a board record yet and grading here is still
local. The source-drift check in scripts/boundaries.sh is not a duplicate of
`boundaries check`, which compares slots to a dataset instead.

Repoints AGENTS.md, both packs, the Go comments, FEATURES, and the identity doc,
and drops a doubled reference in the evaluation pack header.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Kai Siren <coilysiren@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>

coilyco-ops requested review from coilysiren

2026-08-18 15:58:23 +00:00

coilyco-ops referenced this pull request

2026-08-18 15:58:58 +00:00

Grade the board through aos-eval instead of locally #960

coilyco-ops added 1 commit

2026-08-18 16:10:27 +00:00

Merge branch 'main' into aos/claude/eval-docs-one-page

ci / image-build (pull_request) Successful in 51s

Details

ci / test (pull_request) Successful in 1m19s

Details

ci / publish-echo-image (pull_request) Has been skipped

Details

ci / publish-observed (pull_request) Has been skipped

Details

116c903226

main unwrapped the FEATURES index lines while this branch renamed the eval
page. Takes main's single-line shape with the renamed target.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Co-authored-by: Kai Siren <coilysiren@gmail.com>
Co-authored-by: Claude <noreply@anthropic.com>