---

name: adapt-project description: >- Analyze a new or existing project repository, select matching agent-spec capability modules, skills, prompts, and shared context files, and adapt agent-spec governance to the workspace. Execute this skill when the user asks to adapt agent-spec, configure agent-spec for an existing codebase, recommend skills or prompts for a project, or set up workspace context governance. Do NOT execute for defining application foundations from scratch (use define-foundation), authoring new skills (use write-a-skill), or auditing prompt files (use prompt-auditor). version: 1.0.0 verified-on: [cline]

Adapt Project

0. Identity

• Role: Project adaptation engine. Audits workspace context and provisions agent-spec configurations, skills, prompts, and context templates for target projects.
• Authority: Owns the workspace audit, skill matching, prompt selection, and context provisioning workflow. Cannot alter core standards in core/ or modify target application code without user confirmation.
• Must not define: Core instruction hierarchy (core/), foundation PRD/architecture definitions (define-foundation), or skill authoring rules (write-a-skill).
• Normative base: shared/writing/writing-rules.md; docs/anti-patterns.md; skills/_template/SKILL.md; docs/skill-standard.md; skills/prompt-engineering/prompts-by-category/learning/prompt-engineer-mentor.md; references/module-mapping-guide.md.
• Anti-pattern gate: No step may trigger AP-1 (vague task), AP-4 (over-permissive agent), AP-16 (context dump), AP-26 (no scope boundary), AP-44 (unlocked filesystem), or AP-45 (no human review trigger). Never overwrite project context without user confirmation.

1. Intent (9 Dimensions)

| # | Dimension | Value | | --- | ---------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------- | | 1 | Task | Audit a target codebase and configure matching agent-spec skills, prompts, shared rules, and context templates. | | 2 | Target Tool | Any agent runtime reading markdown skills: Claude Code, Cursor, Copilot, Windsurf, Kiro, Cline, raw API. | | 3 | Output Format | Adaptation blueprint in chat summary, plus selected context/ templates with placeholders and .agents/ configurations upon approval. | | 4 | Constraints | Max 3 clarifying questions. Imperative voice. No em dashes. No banned words. Retain template placeholder markers. Require approval before writing files. | | 5 | Input | Target project path, codebase indicators, target AI agent runtime, and user goals. | | 6 | Context | Eliminates configuration friction for teams adopting agent-spec governance. | | 7 | Audience | Project developers and downstream AI coding agents. | | 8 | Success Criteria | Codebase stack identified, relevant skills and prompts selected, context templates scaffolded cleanly, and human review approval recorded. | | 9 | Examples | See §10. |

2. Trigger Matrix

| Trigger | Fire? | Notes | | ------------------------------------------------------- | ----- | --------------------------------------- | | "Adapt agent-spec to my project" | YES | Core trigger for new or existing repos. | | "Adapt agent-spec to my existing codebase / legacy app" | YES | Core trigger for brownfield projects. | | "Recommend skills and prompts for this codebase" | YES | Core trigger. | | "Set up agent-spec governance in this repository" | YES | Core trigger. | | "Define my app foundation from scratch" | NO | Route to define-foundation. | | "Write a new skill" | NO | Route to write-a-skill. | | "Audit an existing prompt file" | NO | Route to prompt-auditor. |

3. Execution Workflow

Step 1: Workspace Audit and Stack Discovery

• Action: Inspect the target project workspace (new or existing codebase). Identify package manifests, configuration files, directory structures, existing source code, and the active AI agent runtime. Preserve existing application source files.
• Input: Target workspace files and user prompt.
• Stop Condition: Primary programming language, framework, project domain, existing architecture, and active agent runtime recorded.
• Validation: Workspace stack and codebase structure categorized against references/module-mapping-guide.md.

Step 2: Capability Suite and Skill Selection

• Action: Match the discovered project domain with relevant capability suites in skills/ (such as dev-workflow, design-engineering, mobile-react-native, autonomous-dev, enterprise-business, prompt-engineering, research-and-productivity, content-and-growth).
• Input: Step 1 stack profile and references/module-mapping-guide.md.
• Stop Condition: List of specific skills and capability suites compiled with short justifications.
• Validation: Selected skills exist in skills/ and match project stack requirements.

Step 3: Prompt Engineering and Role Framing

• Action: Construct tool-specific prompt contracts, role framing declarations, and safety boundaries tailored to the project stack using prompt engineering principles.
• Input: Selected skills and target AI agent runtime.
• Stop Condition: System prompt templates and execution instructions compiled.
• Validation: Prompts strictly follow writing rules (no em dashes, no banned words, proper reasoning model constraints).

Step 4: Shared Rules and Context Scaffolding

• Action: Select required context/ templates (PRD.md, ARCHITECTURE.md, SCHEMA.md, DESIGN.md, RULES.md) and shared/ writing rules needed for project governance. If the project lacks existing context files, apply the Zero-Context Repository Onboarding Protocol from references/module-mapping-guide.md.
• Input: Project architecture, domain needs, and workspace context status.
• Stop Condition: Selected context templates and shared rules prepared for workspace provisioning.
• Validation: All placeholders retain [PLACEHOLDER: ...] markers while confirmed codebase facts are pre-filled into ARCHITECTURE.md.

Step 5: Adaptation Blueprint and Handoff

• Action: Present the complete adaptation blueprint summary to the user. Request explicit confirmation before writing any files to the project repository.
• Input: Compiled results from Steps 1 through 4.
• Stop Condition: User approves or requests modifications to the adaptation blueprint.
• Validation: Approval recorded. On pass, provision approved context/ templates and .agents/ configurations.

4. Output Specification

The summary output in chat follows this structure:

# Agent-Spec Adaptation Blueprint

- **Target Project:** [Project name / path]
- **Detected Stack:** [Languages, frameworks, tools]
- **Target Runtime:** [Active AI agent tool]

## Recommended Modules & Skills

- [module-name/skill-name]: [1-sentence justification]

## Recommended Context Templates

- context/[TEMPLATE].md: [1-sentence purpose]

## System Prompt Framing

[Role statement, file scope bounds, and execution constraints]

## Action Plan

- [ ] Confirm file provisioning

Upon user approval, write selected templates with retained [PLACEHOLDER: ...] markers to context/ or .agents/.

5. Validation Gate

Run before declaring completion:

• [ ] Workspace audit completed and stack accurately categorized.
• [ ] Recommended skills exist in skills/ and match project scope.
• [ ] Prompt recommendations follow tool-specific rules without adding Chain of Thought to reasoning models.
• [ ] Context templates preserve all [PLACEHOLDER: ...] markers without pre-filling mock data.
• [ ] No em dashes and no banned words present in generated content.
• [ ] Explicit human review approval recorded before writing files.

6. Anti-Triggers and Calibration

• Under-execution threshold: Recommending generic modules without checking codebase files, or writing files without user approval.
• Over-execution threshold: Overwriting existing application source code, or pre-filling project facts into context templates instead of leaving placeholders.
• Calibration default: Err toward precise codebase inspection and strict human approval gates.

7. Anti-Pattern Compliance

| Step | Prevents AP | Mechanism | | ----------------- | --------------------------------------- | ------------------------------------------------------ | | 1 (Audit) | AP-1, AP-41 (vague / no starting state) | Reads repository manifests to establish true baseline. | | 2 (Skill match) | AP-27 (no stack constraints) | Selects skills matched to detected stack indicators. | | 3 (Prompt design) | AP-35, AP-54 (CoT on reasoning models) | Tailors prompts strictly to target runtime rules. | | 4 (Context rules) | AP-26, AP-44 (unlocked filesystem) | Provisions scope bounds and context rules. | | 5 (Handoff) | AP-45 (no human review trigger) | Requires explicit user approval before writing files. |

8. Versioning & Changelog

• Version: 1.0.0
• Changelog:
  • 1.0.0 (2026-08-30) — Initial release of Tier-5 project adaptation skill per docs/skill-standard.md.

9. Portability Matrix

| Runtime | Status | Notes | | -------------------- | -------- | ------------------------------ | | Claude Code | untested | | | Cursor | untested | | | Copilot | untested | | | Windsurf | untested | | | Kiro | untested | | | Cline | verified | Executed in current workspace. | | Raw API (no tooling) | untested | |

10. Examples

Input: "I have a Next.js 15 app with Tailwind and Prisma. Adapt agent-spec to my repo and tell me what I need."

Output: Audit detects Next.js 15, React 19, TypeScript, Tailwind CSS, Prisma. Recommends skills/dev-workflow (plan-feature, api-endpoint-generator, database-migration) and skills/design-engineering (01-foundations-and-systems). Scaffolds context/PRD.md, context/ARCHITECTURE.md, context/SCHEMA.md, and context/DESIGN.md with placeholder markers intact. Presents adaptation summary for approval before writing.

Input (Existing/Brownfield Repo): "We have an established Django and React mobile project with 50k lines of legacy code. How do we integrate agent-spec without breaking our code?"

Output: Non-destructive audit detects Python, Django, React Native, Expo. Preserves existing codebase untouched. Recommends skills/mobile-react-native (react-native-best-practices, `