Pin the claude-native-practice skill as a first-party manifest row and raise the listing ceiling to 11,000

[seathatflowsinourveins]

Scope

• What this PR changes, in one or two sentences: pins this repository's own claude-native-practice skill in the central skills manifest as a first-party row (listed on for Claude Code, enabled in Codex) at Ground the Claude Code native practice in primary sources and one cross-client mapping #958's landed commit, and raises the manifest's own ceiling on the on-listed description sum from 10,500 to 11,000 characters, as the command center ruled (option (a) of four, 2026-10-10). The row costs about 90 tokens in every session whose listing includes it.
• Base commit: 9d4c00cd05d79b3fd5074df2e46bacab87ee67ff (main, which has Ground the Claude Code native practice in primary sources and one cross-client mapping #958 landed); the row's ref is Ground the Claude Code native practice in primary sources and one cross-client mapping #958's landed commit 7a6bfbcf0a4f7201c64e5410c2ccbed1b224b511 and its tree_sha the tree of .claude/skills/claude-native-practice there.
• Lane: lane:foundation
• Owned paths touched (12 files): adoption/skills/manifest.json (the row, the three description sums, the ceiling in trial.on_description_char_cap and budget.claude_on_cap, the Codex catalog count in client_budgets), adoption/templates/claude.settings.template.json (skillOverrides entry), adoption/templates/codex.config.template.toml (the comment's count), blueprints/runtime-workers/skills/manifest.json and README.md (the worker exclusion and the reuse count), catalogs/landscape/skills-lifecycle.json (the source and the agent-docs task), catalogs/foundation/upstream-surface-dispositions.json (one cited manifest line moved), docs/decisions/2026-10-10-skills-claude-native-practice-row.md (new), three tests, manifests/evidence.json.

Revisit note (the ruling's): the ceiling comes back down, to 10,500 if the on-listed sum then fits, when the 2026-10-10 wave's thirteen skills-slot verdicts land; I did not read that wave's verdict status for this PR. After this lands the designated host shell installs the skill by adoption/skills/lifecycle.md; a pin does not install itself.

SOTA sources

• The skill itself: .claude/skills/claude-native-practice of this repository at 7a6bfbcf0a4f7201c64e5410c2ccbed1b224b511 (tree 3a485d2a4bbe63dc1ae54e23101c472d21450b8c, SKILL.md sha256 a52a24bc7691fa18b3e4c532c6cbd15b79968d82a54acb4202c9fcced2ad8bcb, 2,224 bytes, description 349 characters), added by Claude Code native practice, 2026-10: one evidenced default per role slot #923 and extended by Ground the Claude Code native practice in primary sources and one cross-client mapping #958; no body is copied.
• Claude Code's skill listing budget and per-entry cut, as the manifest already cites them: skillListingBudgetFraction (default 0.01, the template sets 0.05 under the 2026-09-30 directive), the 8,000-character fallback and the 1,536-character entry cut (budget.client_budgets, docs/decisions/2026-09-30-skills-llm-native-listing.md); the manifest's claude_on_cap is this repository's own ceiling, not a client budget.
• Codex's catalog fit: [skills] max_context_tokens (6,000 in the template) and codex-rs/ext/skills/src/render.rs L19-22, L126-152 and L154-160 at rust-v0.159.2, as cited in budget.client_budgets; the repository counter scripts/skills_status.py estimates 3,171 tokens for the 28-skill catalog against 6,000.
• The install route: the pinned Vercel Skills CLI (cli block of the manifest, v1.7.0, https://github.com/vercel-labs/skills/tree/v1.7.0), skills add <tree URL> --skill claude-native-practice -g -y, as the other rows install.

Evidence-class table

Claim	Evidence class	Command / receipt
The row's pin, tree, SKILL.md hash, byte count and description length are computed from the landed commit's own bytes; the three sums equal the sums over the rows and stay within the ceiling	local_integration	skills_row.py (lane tool, in the research record) and tests.test_skills_manifest (BudgetTests, SkillEntryTests)
Claude on-listed descriptions 10,374 to 10,723 (+349) under a ceiling of 11,000 (277 characters left); Codex enabled and catalog descriptions 10,055 to 10,404; Codex catalog skills 27 to 28	local_integration	scripts/skills_status.py --json: catalog matches the manifest, within budget (the host has not installed the skill yet, so its per-skill install states read missing, by design)
A first-party row has no skills.sh page: the audit test holds it to this repository as its source and to the not-applicable audit form, every other row keeps the skills.sh requirement	local_integration	tests.test_skills_manifest.SkillEntryTests.test_audit_url_is_on_skills_sh
The skill is excluded from the runtime-worker selection by the existing adoption_ref contract, the landscape catalog lists it once under agent-docs, and the Claude template's overrides equal the manifest	local_integration	tests.test_runtime_worker_skills, tests.test_landscape_sweep_skills, tests.test_skills_manifest.ListingBudgetTemplateTests

Declared test contract change

Three main test methods change by design: tests.test_new_wsl_client_config.AgentGapTests.test_the_gaps_are_the_servers_and_skills_the_agents_name_beyond_the_wired_ones_and_the_plans_skills (the plan's selected skill count 27 to 28), tests.test_skills_status.SkillsStatusTests.test_real_manifest_codex_catalog_fits_the_configured_token_budget (the Codex catalog count 27 to 28; still must match the manifest total and fit the unchanged 6,000-token setting) and tests.test_skills_manifest.SkillEntryTests.test_audit_url_is_on_skills_sh (a first-party row is held to the repository source and the not-applicable audit form). Main's versions of these three fail against this head; docs/decisions/2026-10-10-skills-claude-native-practice-row.md declares them.

Local commands run

Head 9d78b7d00d079680f51f97b9c373b79593d20dd6, rebased once onto main 9d4c00cd0 (#968: workflow tests and one manifest entry; clean). The unit tests, the generators and the
convergence check ran on the same change set on #958's landed commit before the rebase (gates head 7a6bfbcf0); validate.py and the manifest
check were rerun at this head. CPython 3.13.16 locally, TMPDIR outside /tmp and outside home:
$ python3 -E -s -B scripts/validate.py   # at this head
{"components": 70, "hashed_files": 11422, "profiles": 4, "receipts": 239, "status": "passed"}   (rc 0)
$ python3 -E -s -B scripts/evidence_manifest.py --check   # at this head
{"files": 11422, "status": "passed"}   (rc 0)
$ python3 -E -s -B scripts/validate_convergence.py --all-recorded   # before the rebase
rc 0
$ python3 -E -s -B scripts/component_matrix.py --write && python3 -E -s -B scripts/new_host_grand_list.py --write   # before the rebase
rc 0; no change to a generated report
$ python3 -E -s -B -m unittest <the 24 manifest, template, catalog and consumer modules>   # before the rebase
Ran 1461 tests in 589.421s; OK (skipped=26)
$ git diff origin/main --unified=0 | grep '^+'   # added lines, scanned for the host's private names
hits 0

12 files changed, 159 insertions(+), 35 deletions(-). The local full-suite shards in a CI-like environment are noisy on this host (four of eight red at a CI-green head on tests no change touches), so the list above is the targeted run and CI's eight shards are the full-suite measurement.

Decision record

docs/decisions/2026-10-10-skills-claude-native-practice-row.md: the ceiling raise and its cost, the listing check, what follows, the declared test changes, the inverse and the revisit.

Host evidence

Not applicable: no file under evidence/hosts/ changes.

Checklist

• New/changed GitHub Actions are pinned to a full commit SHA with a version comment (no floating tags). (none changed)
• New/changed workflows declare top-level permissions: {} and grant each job only what it needs. (none changed)
• No secrets are printed, logged or committed; no new required secret was added without a documented owner.
• No new paid hosting, subscription or billing surface was introduced.
• Peer-owned untracked files and worktrees were preserved (not deleted, moved or overwritten).

🤖 Generated with Claude Code

[socket-security]

Dependency limit exceeded — report not shown.

This pull request scan exceeded the 10,000-dependency limit applied to this scan, so the results are incomplete and may be inaccurate. To avoid reporting false positives, Socket has not posted a report.

Upgrade your plan to raise the dependency limit and get complete reports, or view the partial scan in the dashboard.

Socket is always free for open source. If this is a non-commercial open source project, contact us to request a free Team account.