fix: overhaul Claude Code plugin for correctness and spec compliance (#225 )

- Fix critical bugs: coder-agent name (CoderAgent→coder-agent), manifest
  path in session-start.sh, /oac:setup ghost command replaced with /install-context
- Fix hooks.json: add timeout/async, add description, restore matcher
- Fix session-start.sh: dual-format output (Claude Code + Cursor/OpenCode),
  remove invalid JSON escapes for $ and backtick
- Fix agent spec compliance: disallowedTools on read-only agents, WebFetch
  on external-scout, haiku model on context-scout, example blocks on 3 agents
- Fix skill descriptions: triggering conditions only (no workflow summaries)
  per superpowers empirical testing — prevents Claude shortcutting skill content
- Add Red Flags + Rationalization tables to 4 discipline skills
- Add install-context skill + command replacing oac-setup/oac-plan/oac-add-context
- Add cleanup-tmp.sh (was missing, broke /oac:cleanup)
- Add TypeScript installer (install-context.ts/js) with profile-based download
- Update plugin.json: keywords, homepage for marketplace discoverability
- Update README: correct skill count, fix structure diagram, remove stale refs
- Remove obsolete files: FIRST-TIME-SETUP.md, INSTALL.md, QUICK-START.md,
  download-context.sh, verify-context.sh, load-config.sh, context-manager skill,
  oac-setup/oac-plan/oac-add-context commands

2026-02-19 20:16:17 +00:00

2.7 KiB

Raw Blame History

name	description	context	agent
context-discovery	Use when coding standards, security patterns, or project conventions need to be discovered before implementation begins.	fork	context-scout

Context Discovery

Overview

Discover project standards and patterns before implementing features. Context-scout finds and ranks relevant files from .opencode/context/ based on your request.

Announce at start: "I'm using the context-discovery skill to find relevant standards for [feature/task]."

The Process

Step 1: Invoke Context-Scout

Run the skill with your implementation topic:

/context-discovery [what you're implementing]

Examples:

/context-discovery JWT authentication
/context-discovery React form validation
/context-discovery database migration workflow

Step 2: Load Critical Priority Files

Read EVERY file marked Critical Priority:

Read: .opencode/context/core/standards/code-quality.md
Read: .opencode/context/core/standards/security-patterns.md

These are mandatory—proceed only after loading.

Step 3: Load High Priority Files

Read files marked High Priority:

Read: .opencode/context/core/workflows/approval-gates.md

These are strongly recommended for your implementation.

Step 4: Load Medium Priority (If Needed)

Read Medium Priority files for additional context:

Read: .opencode/context/project-intelligence/architecture.md

These are optional but helpful.

Step 5: Apply to Implementation

Follow standards from loaded files
Apply patterns to your code
Use naming conventions discovered
Check workflows before executing

Delegation Pattern

When invoking subagents, pass discovered context files:

Invoke coder-agent:

Task: Implement JWT service

Context to load:
- .opencode/context/core/standards/code-quality.md
- .opencode/context/core/standards/security-patterns.md

Instructions: Follow functional patterns and security best practices.

Error Handling

"No context files found"

Run /install-context to download context first

"Too many files returned"

Be more specific (e.g., "TypeScript coding standards" not "coding")

"Which files do I load?"

Always: Critical → High → Medium (if needed)

Remember

Context FIRST, code SECOND—never skip discovery
Critical priority files are MANDATORY, not optional
Training data is outdated—context is current
Pass context forward when delegating to subagents
Only use file paths returned by context-scout

task-breakdown
code-execution
external-scout

Task: Discover context files for $ARGUMENTS

Follow navigation-driven discovery and return ranked recommendations.

2.7 KiB Raw Blame History