Skip to content

Implementation Plan: T5-P6-A3-WP1 — Validate N=3 Extensibility#4148

Merged
Trecek merged 3 commits into
developfrom
t5-p6-a3-wp1-validate-n-3-extensibility-by-classifying-every/4052
Jun 28, 2026
Merged

Implementation Plan: T5-P6-A3-WP1 — Validate N=3 Extensibility#4148
Trecek merged 3 commits into
developfrom
t5-p6-a3-wp1-validate-n-3-extensibility-by-classifying-every/4052

Conversation

@Trecek

@Trecek Trecek commented Jun 28, 2026

Copy link
Copy Markdown
Collaborator

Summary

Create docs/design/paper-backend-n3-exercise.md — a six-section classification document that walks every CodingAgentBackend Protocol method (30 total), every BackendCapabilities field (41), and both BackendConventions fields against opencode-via-ACP as the N=3 paper backend candidate. Every surface is classified as TRIVIAL / SHIM-REQUIRED / GAP, and every GAP row produces a named Protocol change request in Section 6. Update docs/design/README.md with a new index row. Add the new filename to the segment-count allowlist in tests/docs/test_filename_naming.py (the filename has 4 kebab segments, exceeding the max-3 rule enforced by test_rule_4_max_three_segments).

Implementation Plan

Plan file: /home/talon/projects/autoskillit-runs/impl-20260628-145445-208417/.autoskillit/temp/make-plan/validate_n3_extensibility_plan_2026-06-28_150600.md

Closes #4052

🤖 Generated with Claude Code via AutoSkillit

Token Usage Summary

Step Model count uncached output cache_read peak_ctx turns cache_write time
plan* opus[1m] 1 880 18.8k 1.5M 104.6k 41 85.8k 13m 44s
verify* sonnet 1 60 9.6k 286.1k 55.8k 21 37.3k 3m 45s
implement* MiniMax-M3 1 91.7k 16.4k 2.1M 0 67 0 3m 44s
audit_impl* sonnet 1 18 5.0k 127.3k 65.6k 26 7.4k 8m 8s
prepare_pr* MiniMax-M3 1 43.2k 2.0k 113.7k 0 13 0 43s
compose_pr* MiniMax-M3 1 37.3k 2.1k 141.0k 0 12 0 24s
review_pr* sonnet 3 434 91.9k 2.9M 96.8k 129 223.4k 27m 5s
resolve_review* sonnet 2 378 32.0k 2.8M 82.7k 106 117.3k 12m 15s
Total 174.0k 177.8k 10.0M 104.6k 471.2k 1h 9m

* 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 425 5030.2 0.0 38.6
audit_impl 0
prepare_pr 0
compose_pr 0
review_pr 0
resolve_review 22 128740.5 5331.8 1453.3
Total 447 22477.3 1054.1 397.7

Model Usage Breakdown

Model steps uncached output cache_read cache_write time
opus[1m] 1 880 18.8k 1.5M 85.8k 13m 44s
sonnet 4 890 138.6k 6.1M 385.4k 51m 15s
MiniMax-M3 3 172.2k 20.4k 2.4M 0 4m 50s

Trecek and others added 3 commits June 28, 2026 15:16
…xercise

Classify every CodingAgentBackend Protocol method (30 total), every
BackendCapabilities field (41), and both BackendConventions fields against
opencode as the N=3 paper backend candidate. Sections 2-6 classify as
TRIVIAL / SHIM-REQUIRED / GAP and convert every GAP into a named Protocol
Change Request (PCR-001..PCR-013).

- docs/design/paper-backend-n3-exercise.md: new 6-section classification
  document with gap catalogue and PCR rollup (8 Protocol-change PCRs,
  3 implementation-only PCRs, 1 test-infrastructure PCR).
- docs/design/README.md: new index row pointing to the exercise doc,
  placed after acp-session-contract.md for normative-reference adjacency.
- tests/docs/test_filename_naming.py: add paper-backend-n3-exercise.md to
  ALLOWLIST["segment_count"] — 4 kebab segments exceeds the max-3 rule.

Refs #4052
- Add translate_model (row 20, SHIM-REQUIRED) to Identity/introspection
  cluster; update cluster SHIM 0→1, Total row SHIM 13→14, GAP 5→4,
  and prose '5 GAP rows' → '4 GAP rows'
- Fix PCR rollup 'Protocol change required' count 8→9 (9 IDs enumerated:
  PCR-001,002,003,004,005,007,008,010,013)
- Fix Section 6.2 cross-reference '§5 B4v pattern' → '§5 B3b pattern'
  to match the probe-category label defined in Section 5.3

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Extend legend scope from "Sections 2–4" to "Sections 2–5" (L129):
  Section 5 Probe-category rollup uses the same TRIVIAL/SHIM-REQUIRED/GAP
  taxonomy and was omitted from the legend's stated coverage.
- Correct Section 3.4 ACP-Mappable rollup (TRIVIAL 7→6, SHIM-REQUIRED 2→3):
  rows 1,2,5,13,14,17=6 TRIVIAL; rows 7,9,10=3 SHIM-REQUIRED; GAP=8 unchanged.
- Correct Section 3.4 autoskillit-Local Extension rollup (TRIVIAL 9→11, GAP 4→2):
  rows 19–26,29–31,33=11 TRIVIAL; rows 22,27,32,34,35=5 SHIM-REQUIRED; rows 18,28=2 GAP.
- Correct Section 3.4 Total row (TRIVIAL 21→22, SHIM-REQUIRED 8→9, GAP 12→10).
- Update prose "12 GAP rows" → "10 GAP rows" and "8 SHIM-REQUIRED rows" → "9 SHIM-REQUIRED rows".

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

@Trecek Trecek left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

AutoSkillit review passed. No blocking issues found.

@Trecek Trecek added this pull request to the merge queue Jun 28, 2026
Merged via the queue into develop with commit 490aacc Jun 28, 2026
3 checks passed
@Trecek Trecek deleted the t5-p6-a3-wp1-validate-n-3-extensibility-by-classifying-every/4052 branch June 28, 2026 23:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant