From dekspec
> **Vendored asset paths (INT-097):** Paths below like `dekspec/...` reference the consumer-vendored layout. Pip-only installs resolve via `dekspec resource ...`. See [`_lib/vendored_assets.md`](../_lib/vendored_assets.md).
How this skill is triggered — by the user, by Claude, or both
Slash command
/dekspec:land-intentThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> **Vendored asset paths (INT-097):** Paths below like `dekspec/...` reference the consumer-vendored layout. Pip-only installs resolve via `dekspec resource ...`. See [`_lib/vendored_assets.md`](../_lib/vendored_assets.md).
Vendored asset paths (INT-097): Paths below like
dekspec/...reference the consumer-vendored layout. Pip-only installs resolve viadekspec resource .... See_lib/vendored_assets.md.
Review-and-land phase-executor for an Intent. The review-side counterpart to /exec-coding-session <intent>: where that skill executes all of an Intent's beads, land-intent drives all of the Intent's IB-aggregate PRs through the review pipeline to a landed state, in dependency order, from one launch — reusing the existing trigger + review-pr + REVIEW_PR_FAIL machinery unchanged. It is a thin orchestrator: its only net-new behavior is the per-Intent PR enumeration/ordering and the operator-confirmed merge gate.
/dekspec:land-intent INT-129
Drive every open PR for INT-129 through review to a landed state. For each PR:
fire the review trigger, run the fix-loop to a terminal verdict, and on GO show
me the squash-merge to confirm. Stop and ask if any PR comes back NO-GO.
See _lib/mode_detection_template.md. Default mode: Land Mode.
--help flag. See Help Mode.INT-NNN / path positional). Proceed to Land Mode.See _lib/help_mode_template.md for the canonical Help rendering contract. Manifest for this skill:
skill_name: "/dekspec:land-intent"
one_line: "Drive all of an Intent's IB-aggregate PRs through review to operator-confirmed merge."
modes:
- { flag: "", args: "<INT-NNN>", description: "Land mode — enumerate the Intent's PRs, review-and-land each in dependency order." }
- { flag: "--help", args: "", description: "Show this help message." }
examples:
- "/dekspec:land-intent INT-129"
- "/dekspec:land-intent --help"
Identify the target Intent INT-NNN. Build the ordered set of its open IB-aggregate PRs using only existing surfaces (no new tooling):
Resolves IB-NNN PR-body convention (ADR-025) + bead external_ref linkage — gh pr list for the open set, git for branch state.br blocked-by topology — the same br-driven dependency walk /exec-coding-session Phase 1 performs.For each PR, in dependency order:
dekspec review trigger review-pr <PR-#> (INT-124). This enters the REVIEW_PR state and auto-invokes /dekspec:review-pr (INT-107).REVIEW_PR_FAIL action-handler (INT-115), which seeds /code-review --comment, fixes real findings, and re-fires review-pr until a terminal verdict. Do not hand-drive it.MERGE GATE — never merge a PR without explicit operator confirmation. This skill is RECOMMEND-only at landing per ADR-026; unattended merge-on-GO is the AUTO graduation tier, gated on the INT-117 calibration corpus, and is out of scope here.
gh pr merge <PR-#> --squash --delete-branch) and wait for an explicit confirmation keystroke. The skill does not merge until the operator confirms. After a confirmed merge, advance to the next PR.When every PR is landed (or the operator halts), report a per-PR outcome roll-up for the Intent (landed / stopped-NO-GO / pending), and name the next action.
land-intent only sequences dekspec review trigger + review-pr + the REVIEW_PR_FAIL handler, which already exist; adding a parallel loop drifts from the canonical pipeline.br blocked-by topology.br dependency order.dekspec review trigger review-pr → review-pr → terminal verdict (no hand-rolled review).After the run, dekspec audit relink against the repo root to restitch any backlinks touched by merges.
npx claudepluginhub dektora/dekspec-public --plugin dekspecProvides CDSS development patterns for drug interaction checking, dose validation, clinical scoring (NEWS2, qSOFA), and alert classification integrated into EMR workflows.