This skill bridges the CATLX Universal AI Operating System and the JARVIS Add-On capability package. It ensures the two systems can cooperate without conflict, respecting each other's additive policies and authorization ladders.
When to activate
User requests involve capabilities from both CATLX and JARVIS domains
A CATLX request maps to a JARVIS-defined capability area (or vice versa)
The user asks "Can JARVIS do X?" or "Use CATLX for Y"
Ambiguous requests that could be handled by either system
Two-mode routing
MODE A — CATLX-first (default)
Classify the user's intent using CATLX's capability-index.json routing table
If the intent maps to a JARVIS-defined area (see bridging table below), delegate to JARVIS via skill({ name: "jarvis" }) with the JARVIS operating model
JARVIS executes the task using its 10-step cycle (goal → compose → plan → execute → verify → record → recover → evaluate → learn)
Return the JARVIS result to the user, optionally weaving in CATLX context
If JARVIS cannot complete, CATLX fallback runs its own workflow
MODE B — JARVIS-first
Classify the user's intent using JARVIS's capability routing (the 20 capability areas in skills/jarvis/SKILL.md)
If the intent maps to a CATLX-defined area (see bridging table below), delegate to CATLX via skill({ name: "catlx-orchestrator" })
CATLX executes using its skill ecosystem
Return the CATLX result, optionally weaving in JARVIS context
If CATLX cannot complete, JARVIS fallback runs its own cycle
Bridging table — intent mapping
CATLX intent area
JARVIS equivalent area
Bridge condition
capability-creation
capability-creation
Route to JARVIS capability-composition skill
decision-support
proactive-assistance
Route to JARVIS if decision needs evidence-based evaluation
digital-world-model
world-state-management
Route to JARVIS for state management
failure-recovery
recovery
JARVIS recovery cycle is default; CATLX fallback if needed
general-reasoning
self-evaluation
Route to JARVIS self-evaluation
high-reliability
high-reliability
JARVIS health-monitoring + prediction
knowledge-updating
memory-management
Route to JARVIS memory store
long-horizon-autonomy
goal-management
JARVIS goal-management is primary
multi-agent-coordination
agent-coordination
JARVIS agent-coordination
multimodal-fusion
screen-understanding
Route to JARVIS for screen UI analysis
persistent-goals
goal-management
Same area — JARVIS persistence wins for goal storage
predictive-assistance
prediction
Route to JARVIS prediction
project-management
project-management
Route to JARVIS project-management
real-time-conversation
multimodal-context
Route to JARVIS for conversation context
skill-creation
skill-creation
JARVIS skill-creation pipeline
tool-routing
tool-routing
JARVIS tool-routing
verification
verification
JARVIS verification-first contract
workflow-optimization
workflow-optimization
JARVIS workflow optimization
world-state-management
digital-world-model
Route to JARVIS world-model
Reverse bridging (JARVIS → CATLX) uses the same table in reverse — if a JARVIS request maps to a CATLX area, delegate to catlx-orchestrator.
Bridging procedure
Classify intent using the bridging table above
Check authorization: both systems have non_override: true additive policies. The most restrictive auth level wins. If CATLX requires EXPLICIT_APPROVAL and JARVIS requires AUTHORIZED_AUTONOMOUS, the more restrictive (EXPLICIT_APPROVAL) governs.
Delegate: invoke the target system's gateway skill:
CATLX → JARVIS: skill({ name: "jarvis" }) with intent description
JARVIS → CATLX: skill({ name: "catlx-orchestrator" }) with intent description
Receive and synthesize: the originating system receives the result and presents it to the user, noting which subsystem handled which part
Cycle protection: maintain an active-chain marker (catlx↔jarvis). If the same cross-system round is detected, skip redundant re-invocation and reuse the existing result
Policy compatibility
Both CATLX and JARVIS adhere to additive-only policies (non_override: true):
CATLX: policies/antigravity-delegation.md rule 6b (explicit user model requests override everything, highest priority)
Combined: When both systems are active, the user's explicit model request (rule 6b HIGHEST PRIORITY) governs model selection. Model fallback stays within Gemini only; Claude/GPT are permanently prohibited as automatic fallbacks for both systems.
Verification-first: Both require observed evidence for success (never fabricate)
Rollback: Both use additive-reversible schemes (default UNREGISTER)
Placeholders: audit.scan_placeholders must find 0 TODO/TBD/FIXME markers in any generated content for either system
Integration with AGENTS.md
The combined routing table appears in the JARVIS Operating Capability System section of AGENTS.md (already added additively). Cross-system routing references:
Never create competing duplicate skills. If a capability exists in both systems, reuse the existing component and register the mapping in the bridging table. Never create a second knowledge source — connect back to the canonical layers.
1---2name: catlx-jarvis-bridge3description: CATLX ⇄ JARVIS Bridge4---5# CATLX ⇄ JARVIS Bridge67This skill bridges the **CATLX Universal AI Operating System** and the **JARVIS Add-On capability package**. It ensures the two systems can cooperate without conflict, respecting each other's additive policies and authorization ladders.89## When to activate1011- User requests involve capabilities from **both** CATLX and JARVIS domains12- A CATLX request maps to a JARVIS-defined capability area (or vice versa)13- The user asks "Can JARVIS do X?" or "Use CATLX for Y"14- Ambiguous requests that could be handled by either system1516## Two-mode routing1718### MODE A — CATLX-first (default)191. Classify the user's intent using CATLX's capability-index.json routing table202. If the intent maps to a **JARVIS-defined area** (see bridging table below), delegate to JARVIS via `skill({ name: "jarvis" })` with the JARVIS operating model213. JARVIS executes the task using its 10-step cycle (goal → compose → plan → execute → verify → record → recover → evaluate → learn)224. Return the JARVIS result to the user, optionally weaving in CATLX context235. If JARVIS cannot complete, CATLX fallback runs its own workflow2425### MODE B — JARVIS-first261. Classify the user's intent using JARVIS's capability routing (the 20 capability areas in `skills/jarvis/SKILL.md`)272. If the intent maps to a **CATLX-defined area** (see bridging table below), delegate to CATLX via `skill({ name: "catlx-orchestrator" })`283. CATLX executes using its skill ecosystem294. Return the CATLX result, optionally weaving in JARVIS context305. If CATLX cannot complete, JARVIS fallback runs its own cycle3132## Bridging table — intent mapping3334| CATLX intent area | JARVIS equivalent area | Bridge condition |35|---|---|---|36| `capability-creation` | `capability-creation` | Route to JARVIS `capability-composition` skill |37| `decision-support` | `proactive-assistance` | Route to JARVIS if decision needs evidence-based evaluation |38| `digital-world-model` | `world-state-management` | Route to JARVIS for state management |39| `failure-recovery` | `recovery` | JARVIS recovery cycle is default; CATLX fallback if needed |40| `general-reasoning` | `self-evaluation` | Route to JARVIS self-evaluation |41| `high-reliability` | `high-reliability` | JARVIS health-monitoring + prediction |42| `knowledge-updating` | `memory-management` | Route to JARVIS memory store |43| `long-horizon-autonomy` | `goal-management` | JARVIS goal-management is primary |44| `multi-agent-coordination` | `agent-coordination` | JARVIS agent-coordination |45| `multimodal-fusion` | `screen-understanding` | Route to JARVIS for screen UI analysis |46| `persistent-goals` | `goal-management` | Same area — JARVIS persistence wins for goal storage |47| `predictive-assistance` | `prediction` | Route to JARVIS prediction |48| `project-management` | `project-management` | Route to JARVIS project-management |49| `real-time-conversation` | `multimodal-context` | Route to JARVIS for conversation context |50| `skill-creation` | `skill-creation` | JARVIS skill-creation pipeline |51| `tool-routing` | `tool-routing` | JARVIS tool-routing |52| `verification` | `verification` | JARVIS verification-first contract |53| `workflow-optimization` | `workflow-optimization` | JARVIS workflow optimization |54| `world-state-management` | `digital-world-model` | Route to JARVIS world-model |5556**Reverse bridging** (JARVIS → CATLX) uses the same table in reverse — if a JARVIS request maps to a CATLX area, delegate to `catlx-orchestrator`.5758## Bridging procedure59601. **Classify intent** using the bridging table above612. **Check authorization**: both systems have `non_override: true` additive policies. The most restrictive auth level wins. If CATLX requires `EXPLICIT_APPROVAL` and JARVIS requires `AUTHORIZED_AUTONOMOUS`, the more restrictive (`EXPLICIT_APPROVAL`) governs.623. **Delegate**: invoke the target system's gateway skill:63 - CATLX → JARVIS: `skill({ name: "jarvis" })` with intent description64 - JARVIS → CATLX: `skill({ name: "catlx-orchestrator" })` with intent description654. **Receive and synthesize**: the originating system receives the result and presents it to the user, noting which subsystem handled which part665. **Cycle protection**: maintain an active-chain marker (`catlx↔jarvis`). If the same cross-system round is detected, skip redundant re-invocation and reuse the existing result6768## Policy compatibility6970Both CATLX and JARVIS adhere to **additive-only** policies (`non_override: true`):7172- **CATLX**: `policies/antigravity-delegation.md` rule 6b (explicit user model requests override everything, highest priority)73- **JARVIS**: `policies/antigravity-delegation.md` rule 6b (same clause, synced)74- **Combined**: When both systems are active, the **user's explicit model request** (rule 6b HIGHEST PRIORITY) governs model selection. Model fallback stays within Gemini only; Claude/GPT are permanently prohibited as automatic fallbacks for both systems.75- **Verification-first**: Both require observed evidence for success (never fabricate)76- **Rollback**: Both use additive-reversible schemes (default `UNREGISTER`)77- **Placeholders**: `audit.scan_placeholders` must find 0 TODO/TBD/FIXME markers in any generated content for either system7879## Integration with AGENTS.md8081The combined routing table appears in the **JARVIS Operating Capability System** section of AGENTS.md (already added additively). Cross-system routing references:8283- `skill({ name: "catlx-orchestrator" })` — invoke CATLX orchestrator from JARVIS context84- `skill({ name: "jarvis" })` — invoke JARVIS operating model from CATLX context85- `skill({ name: "catlx-jarvis-bridge" })` — explicit bridge invocation (rare; usually auto-routed)8687**Never** create competing duplicate skills. If a capability exists in both systems, **reuse** the existing component and register the mapping in the bridging table. **Never** create a second knowledge source — connect back to the canonical layers.8889## Source / provenance9091- **CATLX side:** `C:\Users\HP\.config\opencode\catlx-skill-system\`, `C:\Users\HP\.config\opencode\skills\catlx\`92- **JARVIS side:** `C:\Users\HP\.config\opencode\jarvis-addon\`, `C:\Users\HP\.config\opencode\skills\jarvis\`93- **Bridge:** This skill, `C:\Users\HP\.config\opencode\catlx-skill-system\skills\catlx-jarvis-bridge\`94- **Cross-references:** AGENTS.md JARVIS section, CATLX orchestrator SKILL.md routing table
Run npx skillmds@latest add samraddhashrivastavatech/catlx-jarvis-bridge in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
CATLX ⇄ JARVIS Bridge It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
SAMRADDHASHRIVASTAVATECH (@samraddhashrivastavatech) published this skill. Their other Agent Skills are listed on their SkillMD profile.
SAMRADDHASHRIVASTAVATECH