Current feature checkpoint: 2026-08-18. Historical XTtape publication details remain below as provenance, not as the current feature inventory.
Use CAM_Codx to state the desired outcome. CAM_Codx is the normal control plane for every managed CAM_CAM capability.
- Direct CAM_CAM use remains supported for troubleshooting, recovery, runtime development, regression isolation, and expert scripts.
- One canonical
cam-codxskill removes the normal need to choose among specialized CAM_Codx skills. - Six everyday SWE intents are approved:
assess,plan,build,fix,verify, andrecord. - One source-to-outcome SWE Run will use existing CAM-SEQ storage rather than a second reuse database.
- Tasks 1-13 are implemented and release-audited in the recovered CAM_Codx and
CAM_CAM worktrees. The canonical skill, 142-path registry, outcome router,
managed-run seam, setup migration, graph route, release gates, and Task 13
fixture chain are current; see
reports/2026-08-18-task-10-13-release-audit.md. - A real MatrAIx/SESA slice is not claimed. Its target, source revisions,
licensing, privacy boundary, provider policy, and landing scope are the
separate discovery/approval work in
../GOAL_TASK_14_MATRAIX_SESA.md.
See CAM_CAPABILITY_AUDIT_2026-08-12.md and the 2026-08-12 control-plane
design and implementation plan under docs/plans/.
- CAM_Codx is documented as the Codex-native workflow hub.
- CAM_CAM remains the runtime/base engine.
- Repo Necromancer packet workflow is documented.
- Codex goal templates exist.
- Generated CAM Agent Packs exist for Claude Code, Gemini, and Grok Build.
- The packet/approval manager supports fixed operation packets, single-use approvals, digest-only receipts, and separately gated model/self-enhancement promotion or rollback.
- One canonical
cam-codxskill is installable for normal CAM_Codx work. - The Development Brief gives early new-project and continue/rescue guidance with direct-precedent, transferable-analogy, and new-hypothesis labels.
- Development Brief recall uses a supplied primary corpus only and is read-only: no mining, provider calls, target edits, retrieval telemetry, or sibling corpus search occurs by default.
- The Pull Mine Directory skill is an explicit bounded phase:
--source-rootselects another user's repository directory,--dry-runperforms no Git/CAM writes, and a live pass can only reach one supervised--skip-swapcandidate after its evidence gate. It cannot swap, promote, roll back, or edit live configuration. - A shared capability contract exists at
agent-packs/contract/cam_agent_capabilities.json. docs/AGENT_PACKS.mdexplains the generated-pack architecture.- XTtape showpiece artifacts exist under
docs/showpieces/xttape-cam-comparison/. - The XTtape case study exists at
docs/examples/XTTAPE_CAM_SHOWPIECE_CASE_STUDY.md. - Local overlay directories exist at
/Volumes/WS4TB/CAM_ALL. - Archive staging exists at
/Volumes/WS4TB/CAM_ARCHIVE.
- GitHub remotes for CAM_Codx, CAM_CAM, and moriahcareframe were queried.
CAM_CAM/data/claw.dbexists locally and was inspected using metadata-only SQLite commands.- Clean clones were created under
/Volumes/WS4TB/CAM_ALL/repos. - CAM_CAM MCP runtime ownership was inspected from
src/claw/mcp_server.py,src/claw/tools/schemas.py, anddocs/MCP_INTEGRATION_GUIDE.md. - XTtape showpiece evidence was checked for stale CAM paths, excluded source material, and obvious secret patterns before commit.
- The XTtape comparison published a final merged build brain and app implementation plan, but no XTtape runtime app code.
The control-plane feature and main refs were published from the recovered
worktrees on 2026-08-18. The exact execution heads and test receipts are
recorded in reports/2026-08-18-task-10-13-release-audit.md; verify the live
remote refs with:
git rev-parse HEAD
git ls-remote origin refs/heads/main refs/heads/feat/cam-codx-control-planeThe older XTtape publication details that follow are historical and are not a claim about the current control-plane head.
- Working-checkout cross-repo verification passed in the previous batch and is
recorded in
docs/reports/CAM_REPO_REORG_VERIFICATION_2026-06-21.md. docs/repo_inventory/PUBLIC_REPO_CLEANUP_MANIFEST.jsonnow governs tracked public-file cleanup.- Fresh-clone proof passed and is recorded at
docs/reports/CAM_PUBLIC_REPO_DONE_FOREVER_2026-06-21.md. - Agent pack verification is:
python tools/generate_agent_packs.py --check,python -m pytest -q tests/test_agent_packs.py, andgit diff --check. - XTtape showpiece verification included
git diff --check, secret-pattern scan over the committed showpiece docs, and remote head verification.
- CAM_Codx local
mainfast-forwarded tof8f6ef3; this was intentionally a local merge and was not pushed at this checkpoint. - CAM_CAM local
mainfast-forwarded to6b64ca4under the same local-only integration decision. - CAM_Codx verification passed:
python -m pytest -q -p no:cacheprovider tests(44 passed),python tools/generate_agent_packs.py --check, andgit diff --check. - CAM_CAM focused verification passed:
PYTHONPATH=src python -m pytest -q tests/test_read_only_brief_query.py tests/test_tool_schemas.py tests/test_integration_wiring.py(78 passed),PYTHONPATH=src python -m claw.cli brief-query --help, andgit diff --check. - The no-mutation proof is fixture-backed against a synthetic SQLite database; it is not a live-corpus read test.
- Deleting, moving, renaming, or archiving old folders.
- Publishing
claw.db. - Publishing real API keys or local-only config.
- Merging CAM_CAM, CAM_Codx, and generated products into one monorepo.
- Creating separate
CAM_Claude,CAM_Gemini, orCAM_Grokproduct forks. - Claiming that XTtape app runtime code has been built. The published XTtape artifact is a planning/evidence showpiece and implementation contract.