Skip to content
Back to search

agent-scripts

Scripts for agents, shared between my repositories.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 11 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/steipete/agent-scripts with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/steipete/agent-scripts'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/steipete-agent-scripts/badge.svg)](/listings/steipete-agent-scripts)

HTML

<a href="/listings/steipete-agent-scripts"><img src="/api/listings/steipete-agent-scripts/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 117 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.

  1. # Agent Scripts
  2. Shared agent instructions, skills, and small portable helpers for Peter's local workspaces.
  3. This repo is the canonical place for:
  4. - `AGENTS.MD`: shared hard rules for Codex/Claude-style agents
  5. - `skills/`: reusable workflow skills, including repo-owned skills exposed by symlink
  6. - `scripts/`: dependency-light helpers used across projects
  7. - `hooks/`: local guardrails such as skill validation
  8. ## Skills
  9. Skills are the main routing layer. Each `skills/<name>/SKILL.md` has YAML front matter:
  10. ```yaml
  11. ---
  12. name: skill-name
  13. description: "Short generic trigger phrase."
  14. ---
  15. ```
  16. Rules:
  17. - Keep descriptions short and generic; optimize for routing, not documentation.
  18. - Keep skill bodies terse and operational.
  19. - Prefer helper scripts under `skills/<name>/scripts/` when a workflow has repeatable commands.
  20. - Validate after edits: `scripts/validate-skills`.
  21. - Quote `description` in front matter.
  22. Global discovery is built by `scripts/sync-skills` (idempotent; run on every Mac after cloning or adding skills):
  23. - Codex scans nested dirs, so it gets whole-root links: `~/.codex/skills/agent-scripts -> ~/Projects/agent-scripts/skills`, `~/.codex/skills/manager -> ~/Projects/manager/skills`.
  24. - Claude Code loads only `~/.claude/skills/<name>/SKILL.md` (exactly one level deep; per-entry symlinks are followed, category subfolders are not scanned — verified on 2.1.197). It gets a flat per-skill link mirror covering both repos plus machine-local `~/.codex/skills/<name>` extras.
  25. - Name collisions resolve agent-scripts > manager > codex-local; the script prints skipped duplicates and prunes broken/stale managed links.
  26. - Real destination files and directories are preserved. A real Claude skill directory with a Codex backlink to that same directory satisfies local ownership; other real destination conflicts are reported and make sync fail.
  27. For the specific legacy topology `~/.claude/skills/NAME/NAME -> ~/.codex/skills/NAME -> ~/.claude/skills/NAME`, invoke the sync owner directly with an explicit allowlist:
  28. ```bash
  29. /absolute/path/to/agent-scripts/scripts/sync-skills --repair-nested-self-links --dry-run -- boxd-cli boxd-setup-deploy
  30. ```
  31. ```bash
  32. /absolute/path/to/agent-scripts/scripts/sync-skills --repair-nested-self-links -- boxd-cli boxd-setup-deploy
  33. ```
  34. This mode validates every candidate before unlinking only the extra nested leaves. It preserves the real skill directories, assets, and valid Codex backlinks, and exits before creating roots, building mirrors, pruning, or touching instruction pointers. Names must start with an ASCII letter or digit and contain only letters, digits, `.`, `_`, or `-`; duplicates, missing names, and unknown arguments are rejected. A missing nested leaf is a no-op only with the expected surrounding topology. Redirected/inaccessible roots, unexpected objects or literal targets, and changed directory/link identities cause refusal. Rechecks before each unlink are not atomic concurrency protection; a later error stops the batch and reports removals already completed, without rollback.
  35. The read-only `skills/fleet-maintenance/scripts/agent-skill-links-audit.sh` reports same-name nested ancestor loops as `reason=nested-self-link`. This is narrow detection, not an exhaustive graph validator. Its `--repair` remains a broad sync through `~/Projects/agent-scripts/scripts/sync-skills`; it is not the scoped repair above. Run `scripts/test-sync-skills` for isolated fixture coverage; `scripts/test-sync-skills --recurrence-only /absolute/path/to/old-sync-skills` runs the unchanged recurrence assertion against an original helper.
  36. Shared personal skills live as real folders in `skills/`. Public OpenClaw shared skills live in `../agent-skills` and are exposed here with tracked relative symlinks. Repo-owned skills stay canonical in their repo and are exposed here the same way, for example:
  37. ```text
  38. skills/autoreview -> ../../agent-skills/skills/autoreview
  39. skills/discrawl -> ../../discrawl/.agents/skills/discrawl
  40. skills/peekaboo -> ../../peekaboo/skills/peekaboo
  41. ```
  42. Current symlinked repo-owned skills include `birdclaw`, `discrawl`, `gog`, `imsg`, `peekaboo`, `slacrawl`, `wacli`, and `wacrawl`.
  43. Keep `~/Projects/peekaboo` cloned and current before syncing the Peekaboo skill. Its `skills/peekaboo/SKILL.md` owns command, permission, capture, and interaction guidance; edit that source instead of adding another copy here. After updating both repositories, run `scripts/sync-skills` and verify the Peekaboo skill resolves through the Codex and Claude mirrors.
  44. ## Agent Instructions
  45. Shared hard rules live in `AGENTS.MD`.
  46. Global setup (also maintained by `scripts/sync-skills`; Claude Code reads `CLAUDE.md` only, so it links to the shared `AGENTS.MD`):
  47. - `~/.codex/AGENTS.md -> ~/Projects/agent-scripts/AGENTS.MD`
  48. - `~/.claude/CLAUDE.md -> ~/Projects/agent-scripts/AGENTS.MD`
  49. - `~/.claude/AGENTS.md -> ~/Projects/agent-scripts/AGENTS.MD`
  50. Downstream repos should use a pointer-style `AGENTS.MD`:
  51. ```text
  52. READ ~/Projects/agent-scripts/AGENTS.MD BEFORE ANYTHING (skip if missing).
  53. ```
  54. Repo-specific rules go below that pointer. Do not copy the shared blocks into downstream repos.
  55. ## Helpers
  56. `scripts/sync-skills`
  57. - Builds the per-machine skill mirror: Codex whole-root links, Claude flat per-skill links, shared `AGENTS.MD` pointers.
  58. - Idempotent; prints changes only, prunes broken/stale managed links, never clobbers real files.
  59. - No arguments runs ordinary sync; `--help` prints usage. Only the scoped `--repair-nested-self-links` mode accepts `--dry-run`.
  60. `scripts/validate-skills`
  61. - Checks every `skills/*/SKILL.md`.
  62. - Verifies YAML front matter plus required `name` and `description`.
  63. - Enable as a local hook with `git config core.hooksPath hooks`.
  64. `scripts/docs-list.ts`
  65. - Walks `docs/`.
  66. - Enforces `summary` and `read_when` front matter.
  67. - Prints onboarding summaries for repos that wire it in.
  68. `scripts/browser-tools.ts`
  69. - Standalone Chrome DevTools helper.
  70. - Profile copies are staged before replacement so failed copies preserve existing data. Failed startup stops only the process it launched. The picker supports SVG classes, and serialized console reads release object handles.
  71. - `content` and `search` fail on navigation errors; a loaded article remains readable when only network idle times out. Individual `search --content` failures are reported per result.
  72. - Common commands: `start --profile`, `nav <url>`, `eval '<js>'`, `screenshot`, `console`, `network`, `search --content "<query>"`, `content <url>`, `inspect`, `kill --all --force`.
  73. - Build optional binary with `bun build scripts/browser-tools.ts --compile --target bun --outfile bin/browser-tools`.
  74. ## Syncing
  75. Treat this repo as canonical for shared agent rules and portable helper scripts.
  76. When syncing downstream repos:
  77. - Pull latest here first.
  78. - Ensure each target repo starts with the pointer-style `AGENTS.MD`.
  79. - Preserve repo-local rules below the pointer.
  80. - Copy helper changes both directions only when the helper is meant to stay byte-identical.
  81. - Keep scripts dependency-free and portable; no repo-specific imports or path aliases.
  82. For submodules, repeat the pointer check inside each subrepo, push those changes, then bump submodule SHAs in the parent repo.
  83. ## Releases
  84. See [RELEASING.md](RELEASING.md) for the signed-tag GitHub release workflow and its no-publish rehearsal.

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.