SKILL PLAN · DETERMINISTIC
A small plan for a real job
Start with the minimum useful stack. Add nothing unless it has a clear responsibility in the run.
- Task
- Browser automation
- Agent
- Codex
- Priority
- Balanced
Change any option, then update this plan. Rules + reviewed metadata only. No LLM. Same inputs produce the same plan.
01Core execution
Native · Codex
agent-browser
Performs the browser navigation and interaction sequence. Open agent-browser profile and install guide ↗Review permissions, limitations, and evidence
- Agent fit
- Native · Codex
- Main limitation
- The checked-in SKILL.md is a discovery stub; full instructions come from the installed CLI version.
- First risk
- Browser-side actions · High
- Source
- Pinned source · d01253d9
02Guardrail
Native · Codex
Minimal Run and Audit
Limits the first run to the smallest useful path. Open Minimal Run and Audit profile and install guide ↗Review permissions, limitations, and evidence
- Agent fit
- Native · Codex
- Main limitation
- Detailed package files are not fully catalogued.
- First risk
- Review before install · Medium
- Source
- Pinned source · fb3ccdf5
03Evidence
Native · Codex
playwright-workflows
Checks the final state and preserves repeatable evidence. Open playwright-workflows profile and install guide ↗Review permissions, limitations, and evidence
- Agent fit
- Native · Codex
- Main limitation
- Detailed package files are not fully catalogued.
- First risk
- Review before install · Medium
- Source
- Editorial source
Core executionbrowser-use
- Choose this when
- you want fewer declared permissions for the same role
- Tradeoff to check
- Detailed package files are not fully catalogued.
Inspect browser-use ↗Guardrailcontext-map
- Choose this when
- you want fewer declared permissions for the same role
- Tradeoff to check
- The workflow is intentionally brief and may miss dynamic dependencies or runtime ownership.
Inspect context-map ↗Evidencewebapp-testing
- Choose this when
- you want stronger published documentation to start from
- Tradeoff to check
- It assumes Python Playwright and a runnable local application are available.
Inspect webapp-testing ↗ Combined permission budget
This is the union of declared permissions across the plan. Review each package before installation.
Browser controlNetwork accessRun commandsRead research codeRun bounded commandsWrite audit artifactsWrite screenshots
Finish with evidence
Convert a repeatable browser workflow into a controlled run with visible success criteria.
Use test accounts where possible; do not submit irreversible actions without confirmation.
- 3-step acceptance check
- Run the workflow once with harmless test data.
- Verify the final page state and any created record.
- Repeat the run from a clean browser state.
- Stop conditions
- The flow reaches payment, publishing, deletion, or account changes.
- A selector depends on unstable visual position only.
- The site requests an undeclared credential or permission.