balaji-dutt/dotfiles/opencode-permission-review

Claude Code and agent skill by balaji-dutt/dotfiles

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Review OpenCode permission-capture logs, extract safe bash permission patterns, decide host vs devcontainer vs project scope, and clear the capture file after user approval. Use when the user says opencode permission review, permission prompt cleanup, or asks to reduce OpenCode approval prompts.

Install

npx vskill@latest install balaji-dutt/dotfiles/opencode-permission-review

SKILL.md

OpenCode permission review

Use this skill to turn captured OpenCode permission prompts into safer, more focused opencode.jsonc permission rules.

Inputs

  • Captured records from the permission_capture_read tool.
  • Existing permission config from the host and/or project config.
  • User preference for global vs repo-specific rules when the scope is unclear.

If the permission_capture_* tools are unavailable, tell the user to restart OpenCode so the local opencode-permission-capture plugin can load.

Workflow

  1. Read capture status and recent records with permission_capture_read.
  2. Group records by normalized command shape:
  • Trim leading/trailing whitespace.
  • Collapse repeated spaces outside quoted strings when describing patterns.
  • Keep shell operators (&&, ||, ;, pipes, heredocs) visible.
  • Treat exact commands, command probes, shell chains, and heredocs as separate

categories.

  1. For each group, classify risk and scope:
  • Safe allow candidate: read-only, deterministic, no network secrets, no

destructive side effects, narrow command prefix.

  • Keep ask: commits, pushes, package installs, network writes, generated

shell chains, long heredocs, commands with user data/secrets, or anything whose safety depends on context.

  • Deny candidate: destructive or credential-exfiltration patterns.
  • Global host+container: low-risk, platform-generic probes that are useful

everywhere, such as command -v ..., exact --help checks, or read-only version checks.

  • Project-specific: repo scripts, repo-relative validation commands, or

paths that only make sense in this checkout.

  • Host-only/container-only: commands tied to host paths, /workspaces,

Windows paths, macOS tools, WSL helpers, or devcontainer-only tooling.

  1. Propose permission-rule edits only after explaining the classification.
  2. Ask the user which proposals to apply when there is any ambiguity.
  3. After approved edits and validation, clear the capture file with

permission_capture_clear using confirm value clear permission capture.

Scope rules for this dotfiles repo

  • Host global OpenCode config:

private_dot_config/opencode/opencode.jsonc.

  • Devcontainer global OpenCode config:

private_Documents/development/container-dotfiles/dotfiles/private_dot_config/opencode/opencode.jsonc.

  • Repo/project OpenCode config:

.opencode/opencode.jsonc.

  • Shared OpenCode assets under private_dot_config/opencode/{commands,plugins,skills,prompts}

are mirrored into the devcontainer copy by ./assets/sync-devcontainer-assets.sh. Edit the host source and sync; do not hand-edit mirrored copies unless the config file is intentionally distinct.

Permission pattern policy

  • OpenCode bash rules are order-sensitive: the last matching rule wins.
  • Keep broad "*": "ask" rules before specific allows and keep explicit denies

after allows.

  • Prefer exact or narrow prefix patterns.
  • Do not recommend broad allows for:
  • arbitrary shell chains like * && *, * || *, or *; *;
  • arbitrary heredocs like python3 - <<*, bash <<*, or node <<*;
  • commits (git commit*, oc-commit*, cc-commit*) without user review;
  • force pushes, hard resets, rm, sudo, permission changes, disk tools, or

credential/cloud state access;

  • commands containing tokens, passwords, auth headers, or private URLs.
  • If a shell chain is repeatedly safe, prefer decomposing it into individual

allowed commands or allowing a repo-owned wrapper script instead of allowing the full chain shape.

  • Long validation heredocs should usually become checked-in scripts or stay

ask; do not hide arbitrary code execution behind a broad permission rule.

Review output format

Return a concise table with these columns:

Command/groupRiskRecommended actionScopeRule candidate

Then list:

  • rules to add globally to both host and container;
  • host-only/container-only rules;
  • project-specific rules;
  • commands that should remain ask or become scripts;
  • questions for the user.

Do not edit configs until the user approves the proposed rule set.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

7-day trend score
0

Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/balaji-dutt/dotfiles/opencode-permission-review/badge)](https://verified-skill.com/skills/balaji-dutt/dotfiles/opencode-permission-review)

Something wrong with this skill or its page? Tell us.