Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add crenshawdev/cadence/plugin install cadenceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/crenshawdev/cadence/cad-land)<a href="https://agentmods.dev/skills/crenshawdev/cadence/cad-land"><img src="https://agentmods.dev/badge/skills/crenshawdev/cadence/cad-land.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00032 | $0.04620 |
| Opus 5 | $0.00016 | $0.02310 |
| Sonnet 5 | $0.00006 | $0.00924 |
| Haiku 4.5 | $0.00003 | $0.00462 |
Grade A, and why
cad-land scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 298 lines — stays where its author put it; the contents beside it link to each section on GitHub.
<execution_context> @${CLAUDE_PLUGIN_ROOT}/cadence-core/references/git-guard.md </execution_context>
-
Report git state. Current branch; the base =
$ARGUMENTS, elsegit.base_branch, else the firstgit.protected_branchesentry that exists here (references/git-guard.md's fallback); commits ahead of base; unpushed commits; uncommitted/untracked changes; and the remote host detected from the origin URL (gitlab -> MR, github -> PR, any other host whereteaholds a login -> Forgejo/Gitea PR viatea, else local only). Show this plainly before doing anything.Then the tracker, in the same report. Run it here, before any publish ask, on both step-3 arms:
node "${CLAUDE_PLUGIN_ROOT}/cadence-core/bin/issue-check.mjs" check --dir <root> --base <base>
Branch on action alone. On report: say in ONE sentence which issues this
branch's commits reference and which of them are still open ("your branch
references #42 and #47; #42 is still open"), naming a not-found number as
not found and an unresolved one unresolved, never closed or not found.
Print the open list ONLY when
referenced is empty - it is the fallback, never the headline, because a
bare list is what a reader skims past. On skip: print reason verbatim as
ONE line and carry on - never block, never retry, never ask, and never list
an issue the seam did not read. On off - git.issue_check: false, decided
before any forge CLI runs - say NOTHING about the tracker: not the reason,
not that it was skipped. The off switch is off, so this paragraph produces
no output at all.
This report never writes: landing closes no issue, and closing one stays an explicit ask you make at publish time.
-
Uncommitted changes. If the tree is dirty, do NOT auto-commit. Ask (ask-user seam): commit them first (then continue), leave them out of this land, or stop. If HEAD is a protected branch, the protected-branch guard (references/git-guard.md) applies to any commit here.
-
Publish - branch on
git.auto_close.First the deferred queue, on BOTH arms and ahead of the branch below. A gate resolved
deferredran its reviewer and let the run continue, so the finding stops the LAND. Ask what is still queued:
node "${CLAUDE_PLUGIN_ROOT}/cadence-core/bin/planning.mjs" deferred list
On ANY entry in members, STOP the land here: no publish ask, no seam call,
no merge, no tag, neither arm. Print each member's trigger,
discriminator and finding count with the path its bodies are in, then
point the reader at
${CLAUDE_PLUGIN_ROOT}/cadence-core/references/triage-gate.md, whose
deferred arm says how to clear one - triage the member's findings and
record the rulings with planning.mjs adjudication for the same trigger,
discriminator and round, which supersedes it. A non-empty unreadable
arrives as ok:false and refuses exactly as a member does: the queue could
not be PROVEN empty, and this gate never reports "nothing deferred" about
input it could not read, the disposition land-cleanup.mjs gate already
takes for a findings payload it could not parse.
It is NOT land-cleanup.mjs gate and must not be folded into it: that gate
halts only when git.auto_close is true and reads only risk_surface
survivors, so a default-configured project would publish straight over a
deferred plan, diff or phase_diff finding.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 8d ago First seen · 298 lines · 32 tokens per session scan A e94ef0e9e2ea
cad-land is a skill published in the GitHub repository crenshawdev/cadence (5 stars, last pushed today), licensed MIT. It adds 32 tokens to every session and 4,620 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
branch-and-worktree-workflow
Isolates feature work in its own branch or worktree and integrates it cleanly when done. Use this when starting work that should not disturb the current workspace, when several efforts must proceed in parallel on one repository, or when implementation is finished and the change needs merging, rebasing, or splitting…
loop-on-ci
Monitor PR checks and fix failures until green. Uses gh pr checks as the source of truth for PR-attached checks.
git-operations
Git command patterns, branching strategy, and safety protocols. TRIGGER when: managing branches, resolving merge conflicts, or running commit/merge/push operations. SKIP: worktree lifecycle and isolation recovery (use worktree-management); CI automation (use github-actions-template).
dockerized-service-release-deployment-workflow
Create a Dockerized-service release contract with clean GitHub Actions builds, main-anchored tags, immutable digest manifests, published-release deployments, production approval, health checks, and exact-digest rollback.
coordinate-worktrees-and-threads
Assign worktree, branch, write, validation, integration, and cleanup ownership before parallel repository work. Use when a worker will inspect or modify repository state outside the coordinator's worktree.
agile-ledger-workspace
Optional multi-repo orchestrator for Agile-Ledger. Install once at a workspace root to manage many repositories at once: discover new repositories on a GitHub org (including ones nobody told you about), clone and bootstrap them, run a single cross-repo "what changed while I was away" sync, and reconstruct undocumented…