Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add rules/andr-ca/agentharness/branchinggit clone --depth 1 https://github.com/andr-ca/agentharnessWhat 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 | $0.00038 | $0.01181 |
| Opus 5 | $0.00019 | $0.00590 |
| Sonnet 5 | $0.00008 | $0.00236 |
| Haiku 4.5 | $0.00004 | $0.00118 |
Grade C, and why
branching scanned grade C with 1 finding 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 yesterday.
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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- **Remove with git, not `rm -rf`** — `git worktree remove <dir>` (or How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Branching
Full reference: .github/BRANCHING_STRATEGY.md (worktree deep-dive,
.gitignore policy, lifecycle walkthrough). This skill is the actionable
summary.
The core rule
Never commit directly to main/master/trunk/develop/production/
release/*. Always: branch → commit → push → PR → merge. This repo
enforces it locally via git config core.hooksPath .github/hooks
(already set here) — don't rely on the admin bypass.
Branch naming
{type}/{description}, lowercase, hyphens not underscores, short and
specific.
| Type | Purpose |
|---|---|
feature/ |
New feature or enhancement |
fix/ |
Bug fix |
refactor/ |
No behavior change |
test/ |
Testing improvements |
docs/ |
Documentation only |
chore/ |
Maintenance, deps, config |
perf/ |
Performance improvement |
ci/ |
CI/CD changes |
Good: feature/user-authentication, fix/email-validation-crash.
Bad: update, Feature/UserAuth, fix_everything.
Worktrees
A worktree is a second working directory backed by the same repo, so you can have several branches checked out at once without stashing or re-cloning.
Reach for one when:
- Running long tests/builds on one branch while you keep coding another.
- Reviewing a PR branch without disturbing your in-progress work.
- Running agents in parallel — give each agent/task its own worktree so concurrent runs never fight over one working tree or index. This is the highest-value case for an agent harness.
Skip it for a single quick edit on an already-clean, already-scoped branch — a plain branch switch is cheaper. That exception does not apply once either mandatory condition below is true.
Mandatory, not a preference, before starting new feature/fix implementation work (not Q&A, not read-only investigation) when either is true:
git status --porcelainis non-empty (a dirty tree), or- the current branch is unrelated to the task (someone else's topic
branch, a
chore/*tree, anything not already scoped to this work).
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.
- yesterday First seen · 117 lines · 38 tokens per session scan C 531e0a298d18
branching is a cursor rule published in the GitHub repository andr-ca/agentharness (1 stars, last pushed yesterday), licensed MIT. It adds 38 tokens to every session and 1,181 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other cursor rules, from other repositories
dreamd-recall
Recall lessons, decisions, and prior context from the .agent/ memory daemon. Use when starting work in a project that has a .agent/ folder, when the user references a past decision, or when you are about to make a choice that has a documented prior.
handoff
Export a Waybill Bundle by default, or import one when explicitly requested.
ui-motion-workflow
Orchestrate motion-aware UI work by sequencing direction, provider choice, implementation, and browser validation.
lemoncrow.execute
LemonCrow execute mode reference for Cursor.
prompt-routing
Route tasks to the correct Universal AI Engineering Prompt.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.