[auto-research 9.2] Audit-trail artifact in worktree#2174
Merged
Trecek merged 9 commits intoMay 7, 2026
Conversation
- plan-visualization: emit visualization-plan-trace.md and new structured tokens (disambiguation_rule_applied, tier_c_lens, methodology_tradition, visualization_plan_trace_path) - review-design: emit classification_timestamp structured output - research.yaml: capture new outputs, thread to create_worktree and generate_report with --experiment-type, --methodology-traditions, --design-review-verdict, --disambiguation-rule-applied, --tier-c-lens, --classification-timestamp flags - create_worktree.sh: create audit/ dir and copy evaluation_dashboard and visualization_plan_trace to audit/ - generate-report: add YAML frontmatter with full audit-trail schema, add Design Review Summary section - docs: add silent-type-convention.md and audit-trail-format.md - contracts: regenerate research.yaml contract - tests: add 5 test files for schema, artifacts, recipe contracts, and documentation Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
- Add pytestmark with layer("recipe") and small markers to all 5 new test files
- Declare classification_timestamp in review-design outputs contract
- Declare disambiguation_rule_applied, tier_c_lens, methodology_tradition,
visualization_plan_trace_path in plan-visualization outputs contract
- Add silent-type-convention.md and audit-trail-format.md to docs/README.md
- Remove unused monkeypatch parameter from test fixture
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…lidation The undeclared-capture-key rule reads from recipe/skill_contracts.yaml, not recipes/contracts/research.yaml. Add classification_timestamp to review-design and disambiguation_rule_applied, tier_c_lens, methodology_tradition, visualization_plan_trace_path to plan-visualization. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…vior The implementation commits generated the contract card with skill_hashes populated, but the test expects empty hashes (matching migration engine behavior). Regenerated without skills_dir. Also added missing write_behavior: always to make-groups in skill_contracts.yaml. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Syncs test_contracts.py with skill_contracts.yaml after adding write_behavior: always to make-groups. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Tests now perform actual copy operations and verify file existence and content, instead of tautological mkdir+exists or always-true not-exists assertions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Trecek
added a commit
that referenced
this pull request
May 8, 2026
## Summary
Persist classification decisions and review verdicts as committed audit
artifacts in the research bundle. This involves five coordinated
changes:
1. **Extend `create_worktree.sh`** to create a `research/{slug}/audit/`
directory and copy the evaluation dashboard and visualization plan trace
into it.
2. **Extend `plan-visualization/SKILL.md`** to write a
`visualization-plan-trace.md` artifact capturing Tier-C routing
decisions (primary_tradition, applied_union_rules, precedence_trace) and
emit `disambiguation_rule_applied` and `tier_c_lens` as structured
outputs.
3. **Extend `generate-report/SKILL.md`** to add YAML frontmatter to the
report template (the full audit-trail schema: 7 fields + nested
`audit_trail_path` map) and add a "Design Review Summary" section
referencing the committed dashboard.
4. **Update `research.yaml`** to capture new outputs from
`plan_visualization` and `review_design`, and thread them as flags to
`generate_report`.
5. **Author two new docs**: `docs/research/silent-type-convention.md`
(shared convention for #835 and #846) and
`docs/research/audit-trail-format.md` (structure of
`research/{slug}/audit/`).
## Requirements
(Extracted from issue #856 ## Scope section)
Persist the classification decisions and review verdicts as committed
audit artifacts in the research bundle: (a) copy
`evaluation_dashboard.md` from `{{AUTOSKILLIT_TEMP}}/review-design/` to
`research/{slug}/audit/design-review-dashboard.md`; (b) create
`research/{slug}/audit/visualization-plan-trace.md` documenting Tier-C
routing decisions; (c) add a YAML frontmatter metadata block at the top
of `report.md`. Author a shared convention doc
`docs/research/silent-type-convention.md` consumed by Work Items 2.3 and
4.7.
Acceptance criteria:
- `research/{slug}/audit/design-review-dashboard.md` materializes in the
worktree post-review (copied from TEMP, committed)
- `research/{slug}/audit/visualization-plan-trace.md` materializes
post-plan-visualization
- `report.md` has valid YAML frontmatter matching the schema
- `docs/research/silent-type-convention.md` authored and consumed by WI
2.3 and 4.7
- `docs/research/audit-trail-format.md` documents the audit-artifact
structure
- Test: generate a report; parse YAML frontmatter with a YAML loader;
verify all fields present and well-formed
- Test: audit files exist in the worktree and are committed (in git log)
- `task test-check` passes
## Implementation Plan
Plan file:
`/home/talon/projects/autoskillit-runs/impl-20260507-075657-573707/.autoskillit/temp/make-plan/audit_trail_artifact_in_worktree_plan_2026-05-07_080000.md`
🤖 Generated with [Claude Code](https://claude.com/claude-code) via
AutoSkillit
<!-- autoskillit:pipeline-signature
steps=prepare_pr,run_arch_lenses,compose_pr,annotate_pr_diff,review_pr
-->
## Token Usage Summary
| Step | Model | count | uncached | output | cache_read | peak_ctx |
turns | cache_write | time |
|------|-------|-------|----------|--------|------------|----------|-------|-------------|------|
| plan | claude-sonnet-4-6 | 1 | 79 | 20.0k | 610.4k | 68.1k | 96 |
51.6k | 8m 37s |
| verify | claude-opus-4-6 | 1 | 55 | 15.8k | 2.2M | 83.1k | 173 |
131.3k | 9m 53s |
| implement* | MiniMax-M2.7-highspeed | 1 | 4.4M | 26.2k | 2.4M | 114.2k
| 189 | 92.1k | 9m 27s |
| fix | claude-opus-4-6 | 5 | 270 | 38.5k | 8.4M | 95.8k | 291 | 341.9k
| 47m 56s |
| prepare_pr* | MiniMax-M2.7-highspeed | 1 | 136.8k | 9.4k | 172.2k |
28.7k | 22 | 40.9k | 2m 16s |
| compose_pr* | MiniMax-M2.7-highspeed | 1 | 39.9k | 1.8k | 169.3k |
28.7k | 14 | 15.0k | 44s |
| **Total** | | | 4.5M | 111.8k | 13.9M | 114.2k | | 672.8k | 1h 18m |
\* *Step used a non-Anthropic provider; caching behavior may differ.*
## Token Efficiency
| Step | LoC Changed | cache_read/LoC | cache_write/LoC | output/LoC |
|------|-------------|----------------|-----------------|------------|
| plan | 0 | — | — | — |
| verify | 0 | — | — | — |
| implement | 0 | — | — | — |
| fix | 19147 | 436.4 | 17.9 | 2.0 |
| prepare_pr | 0 | — | — | — |
| compose_pr | 0 | — | — | — |
| **Total** | **19147** | 726.8 | 35.1 | 5.8 |
## Model Usage Breakdown
| Model | steps | uncached | output | cache_read | cache_write | time |
|-------|-------|----------|--------|------------|-------------|------|
| claude-sonnet-4-6 | 1 | 79 | 20.0k | 610.4k | 51.6k | 8m 37s |
| claude-opus-4-6 | 2 | 325 | 54.3k | 10.6M | 473.2k | 57m 50s |
| MiniMax-M2.7-highspeed | 3 | 4.5M | 37.5k | 2.7M | 148.0k | 12m 28s |
---------
Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Persist classification decisions and review verdicts as committed audit artifacts in the research bundle. This involves five coordinated changes:
create_worktree.shto create aresearch/{slug}/audit/directory and copy the evaluation dashboard and visualization plan trace into it.plan-visualization/SKILL.mdto write avisualization-plan-trace.mdartifact capturing Tier-C routing decisions (primary_tradition, applied_union_rules, precedence_trace) and emitdisambiguation_rule_appliedandtier_c_lensas structured outputs.generate-report/SKILL.mdto add YAML frontmatter to the report template (the full audit-trail schema: 7 fields + nestedaudit_trail_pathmap) and add a "Design Review Summary" section referencing the committed dashboard.research.yamlto capture new outputs fromplan_visualizationandreview_design, and thread them as flags togenerate_report.docs/research/silent-type-convention.md(shared convention for [auto-research 2.3] Review-design handling of all-silent types #835 and [auto-research 4.7] No-mandatory-figures path in vis-lens-methodology-norms #846) anddocs/research/audit-trail-format.md(structure ofresearch/{slug}/audit/).Requirements
(Extracted from issue #856 ## Scope section)
Persist the classification decisions and review verdicts as committed audit artifacts in the research bundle: (a) copy
evaluation_dashboard.mdfrom{{AUTOSKILLIT_TEMP}}/review-design/toresearch/{slug}/audit/design-review-dashboard.md; (b) createresearch/{slug}/audit/visualization-plan-trace.mddocumenting Tier-C routing decisions; (c) add a YAML frontmatter metadata block at the top ofreport.md. Author a shared convention docdocs/research/silent-type-convention.mdconsumed by Work Items 2.3 and 4.7.Acceptance criteria:
research/{slug}/audit/design-review-dashboard.mdmaterializes in the worktree post-review (copied from TEMP, committed)research/{slug}/audit/visualization-plan-trace.mdmaterializes post-plan-visualizationreport.mdhas valid YAML frontmatter matching the schemadocs/research/silent-type-convention.mdauthored and consumed by WI 2.3 and 4.7docs/research/audit-trail-format.mddocuments the audit-artifact structuretask test-checkpassesImplementation Plan
Plan file:
/home/talon/projects/autoskillit-runs/impl-20260507-075657-573707/.autoskillit/temp/make-plan/audit_trail_artifact_in_worktree_plan_2026-05-07_080000.md🤖 Generated with Claude Code via AutoSkillit
Token Usage Summary
* Step used a non-Anthropic provider; caching behavior may differ.
Token Efficiency
Model Usage Breakdown