PromptScript language expert for reading, writing, modifying, and troubleshooting .prs files. Use when working with PromptScript syntax, creating or editing .prs files, adding blocks like @identity, @standards, @restrictions, @shortcuts, @skills, or @agents, configuring promptscript.yaml, resolving compilation errors, understanding inheritance (@inherit), composition (@use, @extend, @override), contextual @header metadata, or migrating AI instructions to PromptScript. Also use when asked about the 49 built-in compilation targets, including GitHub Copilot, Claude Code, Cursor, Antigravity, Factory AI, and AGENTS.md-based platforms.
PromptScript language expert for reading, writing, modifying, and troubleshooting .prs files. Use when working with PromptScript syntax, creating or editing .prs files, adding blocks like @identity, @standards, @restrictions, @shortcuts, @skills, or @agents, configuring promptscript.yaml, resolving compilation errors, understanding inheritance (@inherit), composition (@use, @extend, @override), contextual @header metadata, or migrating AI instructions to PromptScript. Also use when asked about the 49 built-in compilation targets, including GitHub Copilot, Claude Code, Cursor, Antigravity, Factory AI, and AGENTS.md-based platforms.
PromptScript is a domain-specific language that compiles .prs files into native instruction formats for AI coding assistants (GitHub Copilot, Claude Code, Cursor, Antigravity, Factory AI, OpenCode, Gemini CLI). One source of truth, multiple outputs.
File Structure
A .prs file contains ordered declarations. Syntax 1.5.0 applies @inherit,
@use, local blocks, @extend, and @override in source order. Put @meta
first.
Contextual @header entries appear inside supported owner blocks, not at the
top level.
Content Types
PromptScript has four canonical content shapes inside blocks:
Text Content
Use triple quotes (three double-quote characters) to wrap multiline text.
Text is automatically dedented - leading whitespace from source indentation is stripped.
Use for prose, markdown, or freeform content.
Example: @identity with a text block describing an AI persona starting with "You are..."
Defines AI persona. Start with "You are..." for consistent output across all formatters.
Contains a triple-quoted text block with the persona description.
@context
Project context with structured properties (project, team, languages, runtime)
plus optional triple-quoted text for architecture details, diagrams, etc.
@standards
Category-based conventions. Any category name is valid:
Category names are arbitrary. @standards can also contain free-form text:
@standards {
"""
## Formatting
Preserve heading structure and use four-space indentation.
## Testing
Add regression coverage for every behavior change.
"""
typescript: ["Strict mode", "Named exports only"]
git: {
format: "Conventional Commits"
}
}
Free-form text is dedented and rendered with its Markdown heading structure. Factory
monolith output nests it under Conventions & Patterns; split Factory rules adjust
heading levels relative to the generated section. Custom structured categories remain
available to formatters that support them.
@restrictions
Hard rules as a list of dash-prefixed strings:
@restrictions {
- "Never expose API keys"
- "Never commit secrets to version control"
- "Always validate user input"
}
@shortcuts
Simple strings appear as documentation. Objects with prompt: true generate
executable prompt/command files for GitHub Copilot and Cursor:
@shortcuts {
"/review": "Review code for quality"
"/test": {
prompt: true
description: "Write unit tests"
content: (triple-quoted text with instructions)
}
}
@commands is a backwards-compatible alias for @shortcuts — prefer @shortcuts in new files.
Allowed file types: .md, .json, .yaml, .yml, .txt, .csv. Paths are resolved relative
to the .prs file. Formatters emit referenced files alongside SKILL.md in the output directory.
Parameterized Skills
Skills in .promptscript/skills/<name>/SKILL.md support template parameters via
YAML frontmatter. Define params in frontmatter and use {{variable}} in content:
---name:reviewdescription:'Review {{language}} code for {{standard}}'params:language:type:stringstandard:type:stringdefault:'best practices'references:-references/architecture.md---Reviewthecodeusing {{language}} conventionsfollowing {{standard}}.
The references field in SKILL.md frontmatter lists files to attach to the skill's context.
Paths are relative to the SKILL.md file.
Imported SKILL.md frontmatter is bounded: 256 KiB per document, 10,000 YAML nodes, 32 nesting
levels, 2,000 entries per mapping or sequence, and 64 KiB per string value. Documents over any
limit are rejected before their YAML values are converted.
The validator (PS016) checks that required skills exist, detects self-references,
and catches circular dependency chains.
Skill Contracts (Inputs/Outputs)
Skills can declare typed inputs and outputs in SKILL.md frontmatter:
---name:security-scandescription:'Scan for vulnerabilities'inputs:files:description:'Files to scan'type:stringseverity:description:'Minimum severity'type:enumoptions: [low, medium, high]
default:mediumoutputs:report:description:'Scan report'type:stringpassed:description:'Whether scan passed'type:boolean---
Field types: string, number, boolean, enum (with options list).
The validator (PS017) checks field types, ensures enum fields have options,
and warns if param names collide with input names.
Shared Resources
Skills in a folder can share common resources via .promptscript/shared/:
.promptscript/
shared/
templates.md # Shared across all skills
style-guide.md
skills/
review/
SKILL.md # Gets @shared/templates.md, @shared/style-guide.md
deploy/
SKILL.md # Also gets shared resources
Files in shared/ are automatically included in every skill with @shared/ prefix.
@agents
Custom subagent definitions. Compiles to .claude/agents/ for Claude Code,
.github/agents/ for GitHub Copilot, .factory/droids/ for Factory AI, etc.
Imported agent definitions are qualified by an aliased @use:
@use ./frontend-team as frontend
@use ./backend-team as backend
If both imports define reviewer, the resolved names are frontend.reviewer and
backend.reviewer. Unique unaliased imports keep their original names. Conflicting unaliased
definitions stop compilation with source and import diagnostics instead of silently overwriting
one another. Native targets map dots to hyphens, so frontend.reviewer becomes
frontend-reviewer.
Supports mixed models per agent: specModel sets a different model for
Specification/planning mode (GitHub, Factory), specReasoningEffort sets reasoning
effort for the spec model (Factory only, values: "low", "medium", "high").
Factory AI droids support additional properties: model (any model ID or "inherit"),
reasoningEffort ("low", "medium", "high"), and tools (category name like "read-only"
or array of tool IDs).
Optional args string array; each argument remains one argument.
cwd: "project" runs from project root. Other values are portable forward-slash
paths relative to project root. Hook config file location does not set command cwd.
Environment-root and Git-root wrappers exit before script or command execution when
the required root is unavailable. Native-cwd and workspace-cwd targets retain host
cwd fields and report PS4002 when PromptScript cannot verify that cwd.
timeoutMs range is 100-600000. matcher uses target-native tool names, so a
matcher valid for one target may match nothing on another.
pre-terminal-command supplies native defaults: Factory Execute, Claude and
Codex Bash, Windsurf pre_run_command, Cursor run_terminal_cmd, Gemini
run_shell_command, and VS Code run_in_terminal. Override a native tool name
with targets.<name>.matcher. Cursor, Gemini, and VS Code report best-effort
PS4002 warnings. GitHub and Grok omit the event with PS4002.
Target overrides may change event, matcher, timeoutMs, statusMessage,
continueOnFailure, enabled, or cwd. Native hook files are emitted only in
target modes that support additional files:
Target
Hook output
Mode
Claude Code
.claude/settings.json
full
Factory AI
.factory/hooks.json
multifile, full
GitHub Copilot
.github/hooks/promptscript.json
multifile, full
Cursor
.cursor/hooks.json
full
Codex
.codex/hooks.json
multifile, full
Gemini CLI
.gemini/settings.json
multifile, full
Windsurf
.windsurf/hooks.json
multifile, full
Grok Build
.grok/hooks/promptscript.json
full
VS Code Agent
.github/hooks/promptscript-vscode.json when vscode override exists
target-specific
Simple mode and targets without native project hooks report PS4002 instead of
silently dropping hooks. Use prs compile --watch as fallback. Plugin-only and
agent-scoped integrations are not emitted as universal project hooks.
Each generated command carries a PromptScript ownership marker. CLI cleanup removes
only marked entries and preserves user hooks/settings. Removing @hooks removes a
fully owned generated hook file and prunes directories left empty. prs hooks install factory
migrates unambiguous legacy hooks from .factory/settings.json; ambiguous
entries remain for manual review.
Factory compilation performs the same migration when .factory/hooks.json is
absent. Use prs compile --dry-run to preview the changes or
--no-migrate-factory-hooks to keep warning-only behavior. Unknown events,
malformed entries, and mixed ownership abort without a partial migration.
@hooks compilation is separate from prs hooks install. The latter installs
auto-compilation and generated-output protection for supported AI tools. Copilot VS
Code Agent hooks use promptscript-vscode.json; GitHub Copilot repository hooks use
promptscript.json.
@mcpServers
Project-local Model Context Protocol servers. Requires syntax 1.4.0.
Content detection: PromptScript blocks in .md are parsed as a .prs fragment;
YAML frontmatter with name/description is loaded as a skill definition;
otherwise content is treated as free-form knowledge.
testing! replaces the inherited value. Fields without ! keep normal merge behavior.
Replacement works after @inherit and @use, including aliases and nested target paths.
A missing field is set. The modifier is rejected for @skills, which retain their dedicated
merge and sealing semantics.
Replacing complete targets with @override
Syntax 1.5.0 adds atomic replacement for an existing block or nested value:
@override requires the complete target path to exist, applies in declaration
order, and cannot bypass sealed skill properties. Later @extend declarations
merge into the replacement. Use @extend for additive changes, field! for
compatibility replacement of one direct regular field, and @override for
intentional complete replacement.
Skill-aware @extend semantics
When extending a skill definition via @extend, individual skill properties follow specific merge
strategies rather than the generic block merge rules:
Path matching is normalized ("!./foo.md" matches "foo.md"). Only works in @extend blocks
on references and requires. Validator PS028 warns about ! in base definitions.
Overlay consistency warnings
The resolver emits warnings during compile when an overlay drifts from its base. Always shown
(not gated by --verbose):
Orphaned extend — @extend target "X" not found — overlay will be ignored. Triggered when
the targeted block doesn't exist (base removed or renamed).
Stale skill target — @extend creates new skill "X" — base does not define it. Triggered
when an @extend inside @skills would create a new skill instead of extending an existing one.
Negation orphan — Negation "!path" did not match any base entry — it may be stale.
Triggered when a !entry in references/requires doesn't match anything in the base.
These come from the resolver, not the validator (PS0XX rules). They appear during prs compile,
not prs validate.
Sealed properties
Prevent @extend from overriding specified replace-strategy properties:
@skills {
deploy: {
content: (triple-quoted text with critical workflow)
sealed: ["content", "description"]
}
}
sealed: true seals all replace-strategy properties. Attempting to override a sealed
property is a hard compilation error. Only the base skill author can set sealed —
overlays cannot add or modify it. Append-strategy properties remain extendable.
Validator PS029 warns about invalid entries in sealed.
Skill composition (inline @use)
Import sub-skills within a @skills block to compose multi-phase workflows:
@skills {
ops: {
description: "Production triage"
content: (triple-quoted text with orchestrator instructions)
}
@use ./phases/health-scan
@use ./phases/triage
@use ./phases/code-fix as autofix
}
Each @use resolves the referenced .prs file, extracts its skill definition and context
blocks, and flattens them as numbered phase sections into the parent skill's content. The
as alias form controls the phase display name. Validator PS027 checks composition validity.
Parameterized Inheritance (Template Variables)
Use {{variable}} placeholders in a parent/template file, and pass values
from the child file via @inherit or @use with (key: value) syntax.
IMPORTANT: Variables are NOT set from promptscript.yaml or CLI. They are
passed from one .prs file to another through @inherit or @use.
Step 1: Create the template (parent file with params in @meta):
# base.prs - reusable template
@meta {
id: "service-template"
syntax: "1.0.0"
params: {
serviceName: string
port?: number = 3000
}
}
@identity {
"""
You are working on {{serviceName}} running on port {{port}}.
"""
}
Step 2: Inherit with values (child file passes params):
This skill is automatically included when compiling with prs compile. No manual copying needed.
To disable, set includePromptScriptSkill: false in your promptscript.yaml.
When remote imports are used, run prs lock to generate or update the lockfile
before compilation. It records the exact resolved commit for each dependency.
Integrity hashes (SHA-256) are included for registry references to detect
tampering or drift. This enables reproducible builds across machines and CI.
Commit promptscript.lock to version control.
Use --ignore-hashes on prs compile or prs validate to skip integrity
hash verification when needed.
Policy Engine
Define organizational policies in promptscript.yaml to validate skill extensions:
policies:-name:adjacent-layers-onlykind:layer-boundarydescription:'Only adjacent layers can extend each other'severity:errorlayers: ['@core', '@team', '@project']
maxDistance:1-name:protect-contentkind:property-protectiondescription:'Content override requires explicit approval'severity:warningproperties: ['content', 'description']
-name:approved-registrieskind:registry-allowlistdescription:'Extensions must come from approved registries'severity:errorallowed: ['@core', '@team']
Policy kinds: layer-boundary (controls layer distance), property-protection
(prevents overriding specific properties), registry-allowlist (restricts extension sources).
Severity: error (fails validation) or warning (reported only).
Skip with --skip-policies during development (never in CI).
Syntax Version Validation
The syntax field in @meta declares the PromptScript language version (semver).
Adds @agents and @workflows; reserves internal @prompts
1.2.0
Adds @examples (few-shot input/output pairs)
1.3.0
Adds explicit regular block field replacement in @extend
1.4.0
Adds @hooks, @mcpServers, and @plugins
1.5.0
Adds @header section titles, @override replacement, and unquoted ${VAR} values
Block Version Requirements
Block
Minimum Syntax Version
@agents
1.1.0
@workflows
1.1.0
@examples
1.2.0
@hooks
1.4.0
@mcpServers
1.4.0
@plugins
1.4.0
All other built-in blocks are available from 1.0.0.
Regular block field replacement with field!: value requires syntax 1.3.0.
Generated section title overrides with @header require syntax 1.5.0.
Atomic replacement with @override requires syntax 1.5.0.
Unquoted ${VAR} references as values require syntax 1.5.0.
Generated Section Headers
Use @header inside a registered owner block to rename human-readable output
sections without changing filenames, frontmatter, XML tags, or structured keys:
@header "Title" targets the block's primary section.
@header <section-key> "Title" targets an owned derived section.
Titles must be non-empty, single-line strings.
Source overrides take precedence over formatter configuration and target defaults.
Child inheritance, imported source, and the latest root extension take precedence.
An initial ## Heading in a registered text-only primary owner is a syntax
1.5.0 compatibility fallback. Explicit @header metadata wins.
Ordinary header and headers fields remain domain data.
Validation Rules
PS018 (syntax-version-compat): warns when resolved blocks or syntax features require a higher version than declared. Requirements from inheritance, imports, and skill composition are included. Suggestion: run prs validate --fix.
PS019 (unknown-block-name): warns when a block name is not a known PromptScript type, with fuzzy-match suggestions for typos.
PS037 (valid-section-headers): rejects invalid titles, unknown or unowned section keys, duplicate overrides, and nested extension overrides.
PS038 (valid-block-shape): rejects unsupported built-in block shapes and warns about formatter-sensitive legacy shapes or multiline shortcut scalars.
PS039 (agent-namespaces): validates qualified agent name segments and checks them against recorded import provenance.
PS021 (use-block-filter): errors when only and exclude are both specified in @use parameters.
PS025 (valid-skill-references): errors when a references entry points to a file with a disallowed extension or a path that cannot be resolved.
PS026 (safe-reference-content): warns when a referenced file contains potentially sensitive content (e.g., secrets, credentials).
PS027 (valid-skill-composition): warns about conflicting phase names or excessive phases in composed skills.
PS028 (valid-append-negation): warns when negation prefix ! appears in base skill definitions (only effective in @extend).
PS029 (valid-sealed-property): warns when sealed contains non-replace-strategy property names.
PS030 (policy-compliance): validates skill extensions against organizational policies defined in promptscript.yaml.
Target formatters report PS4002 when a hook event or field has no native equivalent,
when a target cannot guarantee project-root execution, or when output mode cannot emit
the additional hook file.
Fixing Syntax Versions
prs validate --fix # Auto-fix syntax versions in .prs files
prs upgrade # Upgrade all .prs files to the latest version
--fix rewrites the syntax: "..." line in each file's @meta block to match the minimum version required by resolved blocks and syntax features. It follows inheritance, imports, and skill composition. It only upgrades, never downgrades.
prs upgrade upgrades all files to the latest known syntax version regardless of what blocks they use.
CLI Commands
prs init # Initialize project (auto-detects existing files)
prs init --yes --targets claude factory
prs init --dry-run # Preview initialization
prs init --auto-import # Initialize + static import of existing files
prs migrate # Interactive migration flow
prs migrate --static # Non-interactive static import
prs migrate --llm # Generate AI-assisted migration prompt
prs migrate --static --dry-run
prs compile # Compile to all targets
prs compile --watch # Watch mode
prs compile --ignore-hashes # Skip integrity hash verification
prs build <name> # Compile a named build profile
prs validate --strict # Validate syntax
prs validate --fix # Auto-fix syntax version declarations
prs validate --skip-policies # Skip policy engine evaluation
prs upgrade # Upgrade all .prs files to latest syntax version
prs import CLAUDE.md # Import existing AI instructions
prs import CLAUDE.md --dry-run # Preview import conversion
prs inspect <skill> # Show skill composition provenance
prs inspect <skill> --layers # Show layer-level breakdown
prs explain <path> # Explain source and composition provenance
prs explain <path> --format json # Machine-readable provenance history
prs hooks install # Install auto-compilation hooks for AI tools
prs hooks install claude # Install hooks for a specific tool
prs hooks uninstall # Remove installed auto-compilation hooks
prs hooks uninstall claude # Remove hooks for a specific tool
prs skills add <source> # Add a remote skill (@use + lock update + SKILL.md validation)
prs skills add <source> --strict # Treat validation warnings as errors
prs skills add <source> --skip-validation # Bypass Agent Skills spec checks (not recommended)
prs skills remove <name> # Remove a skill (@use line + lock entry)
prs skills list # List all imported skills
prs skills update # Re-resolve markdown-imported skills (re-validates + re-hashes)
prs pull # Update registry
prs diff --target claude # Show compilation diff
prs diff --all --format json # Machine-readable diff report for CI
prs diff --format json --include-content # Include generated content in the report
prs lock # Generate/update promptscript.lock
prs lock --dry-run # Preview lockfile changes
prs update # Re-resolve all remote imports to latest
prs update <url> # Update a specific registry
prs vendor sync # Copy cached deps to .promptscript/vendor/
prs vendor check # Verify vendor matches lockfile
prs resolve @alias/path # Debug: show how an import resolves
prs registry list # Show configured registries and aliases
prs registry add <alias> <url> # Add a registry alias
prs init --yes requires explicit, detected, or user-configured targets. It does not invent
default tools. For existing projects, prs migrate preserves promptscript.yaml, isolates static
output under .promptscript/migrated/, leaves source instructions untouched, and performs no
writes when no candidates are detected.
Output Targets
49 supported targets. Key examples:
Target
Main File
Skills
GitHub
.github/copilot-instructions.md
.github/skills/*/SKILL.md
Claude
CLAUDE.md
.claude/skills/*/SKILL.md
Cursor
.cursor/rules/project.mdc
.agents/skills/*/SKILL.md
Antigravity
.agent/rules/project.md
-
Factory
AGENTS.md
.factory/skills/*/SKILL.md, .factory/droids/*.md
OpenCode
OPENCODE.md
.opencode/skills/*/SKILL.md
Gemini
GEMINI.md
.agents/skills/*/skill.md
Windsurf
.windsurf/rules/project.md
.windsurf/skills/*/SKILL.md
Cline
.clinerules
-
Roo Code
.roorules
-
Codex
AGENTS.md
.agents/skills/*/SKILL.md
Continue
.continue/rules/project.md
-
Hermes
AGENTS.md
-
+ 36 more
See full list in documentation
Targets that share an output path (for example Factory, Codex, and every AGENTS.md target) are
reconciled in one output plan before anything is written. Identical content merges silently;
differing content reports PS4001, where a formatter output replaces an earlier one and a
resource or the auto-injected PromptScript skill keeps the file already planned. Paths are compared
case-insensitively and NFC-normalized on every platform.
Formatter Documentation
For detailed information about each formatter's output paths, supported features, quirks, and example outputs:
Full formatter reference:docs/reference/formatters/ (7 dedicated pages + index of all 49)
llms-full.txt: Available at the docs site root - contains all documentation in a single file for LLM consumption
All 49 formatters indexed at:docs/reference/formatters/index.md with output paths, tier, and feature flags
Auto-Compilation Hooks
Instead of running prs compile --watch manually, install hooks so your AI tool
triggers compilation automatically when you edit .prs files:
prs hooks install # Auto-detect and install for all detected tools
prs hooks install claude # Install for a specific tool
Hooks also protect generated files from direct edits — when an AI agent tries
to edit a compiled output (e.g., CLAUDE.md), the write is blocked with a message
pointing to the source .prs file. Supported tools: Claude Code, Factory AI,
Cursor, Windsurf, Cline, GitHub Copilot, Gemini CLI.
Using {{var}} in the root file without @inherit - template variables only work
in a parent file that defines params in @meta, with values passed by the child
via @inherit ./parent(key: value) or @use ./fragment(key: value). They are NOT
set from promptscript.yaml or CLI flags
Using @examples with syntax: "1.0.0" or "1.1.0" - @examples requires
syntax version 1.2.0. Run prs validate --fix to auto-upgrade
Migrating Existing AI Instructions to PromptScript
Automated: prs import
The fastest way to convert existing AI instructions to PromptScript:
prs import CLAUDE.md # Convert a single file
prs import .github/copilot-instructions.md
prs import AGENTS.md --output ./imported
prs import --dry-run CLAUDE.md # Preview without writing
prs import automatically:
Detects the source format (Claude, GitHub Copilot, Cursor, Factory, etc.)
Maps content to appropriate PromptScript blocks (@identity, @standards, etc.)
Generates a valid .prs file with @meta block
Preserves the original intent and structure
Supported source formats:
CLAUDE.md (Claude Code)
.github/copilot-instructions.md (GitHub Copilot)
.cursorrules or .cursor/rules/*.mdc (Cursor)
AGENTS.md (Factory AI / Codex)
.clinerules (Cline), .roorules (Roo Code)
.windsurf/rules/*.md (Windsurf)
Any Markdown-based AI instruction file
Manual Migration
For complex migrations or when prs import needs refinement:
Source Pattern
PromptScript Block
"You are..." persona text
@identity
Project description, tech stack
@context
Coding conventions, style rules
@standards
"Never...", "Always...", hard rules
@restrictions
/command definitions
@shortcuts
Skill/tool definitions
@skills
Agent/subagent configs
@agents
Reference docs, API specs
@knowledge
After import, split into modular files (context.prs, standards.prs, etc.)
and compose with @use in project.prs. Run prs validate --strict then
prs compile to verify output matches the original.