Intrinsic Core Debugging

What it does

When your agent uses it

• Tasks that involve Container orchestration
• Tasks that involve Debugging
• Tasks that involve gRPC and Protobuf

Example prompts

• “/intrinsic-core-debugging”

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

1. 1Ingress and transport layer (localhost:17080 / Envoy VirtualService): External and inter-pod gRPC traffic routes through Envoy. Named…
2. 2Kubernetes control plane and container readiness layer (app-intrinsic-base, app-resources, skills): Services initialize upstream channels…
3. 3Executive and Behavior Tree orchestration layer (ExecutiveService): Enforces two-phase operation lifecycle (CreateOperation to stage…
4. 4Digital twin and multi-world synchronization layer (ObjectWorldService): Separates static scene ("init_world"), runtime belief state…
5. 5Motion planning and real-time control layer (MotionPlannerService & rs-icon): Collision checking over BVH trees (CoalCollisionChecker)…

What it can do on your machine

Read from SKILL.md and the folder at commit 3fd7856. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  Shell commands in SKILL.md call:

  • kubectl
  • bazel

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md. Its commands use kubectl, which can reach the network depending on how they are called.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Intrinsic Core Debugging loads about 3.8k tokens when it runs, and up to ~44k if it reads all its reference files. Until then it costs about 182 tokens; SKILL.md has 1,186 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~182

When it runs · the whole SKILL.md, loaded when a task matches

~3.8k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~44k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

• NoteRuns commands with sudoSKILL.md:8

  encies: localhost:17080, inctl, kubectl, sudo coredumpctl, grpc.

• NoteRuns commands with sudoSKILL.md:19

  `kubectl describe pod` and crashes via `sudo coredumpctl`.

• NoteRuns commands with sudoSKILL.md:55

  | **Core dump analysis** | `sudo coredumpctl list` and `sudo coredumpctl info <pid>` | Inspects process terminations and

• NoteRuns commands with sudoSKILL.md:78

  n: Inspect crashed container logs or run sudo coredumpctl info for SIGSEGV tracebacks.

• NoteRuns commands with sudoSKILL.md:129

  e dumps with `kubectl describe pod` and `sudo coredumpctl`.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from intrinsic-ai/intrinsic-core at commit 3fd7856, republished under its Apache-2.0 licence (© intrinsic-ai). 1,186 words, ~3,752 tokens.

name

intrinsic-core-debugging

description

Meta-level debugging workflows, architectural layer isolation, and progressive disclosure routing across Envoy ingress, Kubernetes pods, Behavior Trees, ObjectWorld synchronization, ICON real-time control, and hardware modules. Triggers: debug Intrinsic Core failures, gRPC UNIMPLEMENTED, UNAVAILABLE no healthy upstream, Ports not open, Behavior Tree stalls, ICON overruns, multi-world desynchronization, container crashes. Subsystems: INGRESS, ENVOY, KUBERNETES, EXECUTIVE, WORLD, MOTION_PLANNING, ICON, PERCEPTION. Dependencies: localhost:17080, inctl, kubectl, sudo coredumpctl, grpc. Anti-keywords: inctl service create, inctl skill create, inctl solution list, legacy WORKSPACE, cloud deployments.

[Anomaly detected on workcell]
  │
  ├─► [Symptom: rpc error: code = Unimplemented desc = (empty description)]
  │     ├─► Precondition: Client issued gRPC request or inctl command across localhost:17080.
  │     ├─► Diagnostic check: Inspect x-resource-instance-name header against VirtualService routes.
  │     └─► Targeted action: Supply --instance_name=<name> for asset instances, or omit header for core platform services.
  │
  ├─► [Symptom: rpc error: code = Unavailable desc = no healthy upstream]
  │     ├─► Precondition: VirtualService route matched, but target pod or sidecar is not serving.
  │     ├─► Diagnostic check: Run kubectl describe pod <pod> -n <ns>; check container exit codes (128, 137, 139).
  │     └─► Targeted action: Inspect crashed container logs or run sudo coredumpctl info for SIGSEGV tracebacks.
  │
  ├─► [Symptom: Ports not open: <service>.<namespace>:8080]
  │     ├─► Precondition: Solution startup health gate timed out before target service bound port 8080.
  │     ├─► Diagnostic check: Trace synchronous upstream gRPC channels (e.g. executive -> simulation_service:8088 -> gzserver).
  │     └─► Targeted action: Resolve upstream dependency stall (e.g. Gazebo mesh loading) before checking target container.
  │
  ├─► [Symptom: Behavior Tree staged via inctl process set does not execute]
  │     ├─► Precondition: Tree staged in ExecutiveService via CreateOperation.
  │     ├─► Diagnostic check: Verify operation state via ExecutiveService/ListOperations.
  │     └─► Targeted action: Issue ExecutiveService/StartOperation or call executive.run() to trigger execution.
  │
  ├─► [Symptom: ICON control loop overrun: Long duration between read_status_calls]
  │     ├─► Precondition: Real-time loop (500 Hz / 1 kHz) missed deadline on isolated real-time core.
  │     ├─► Diagnostic check: Inspect rs-icon logs; compare futex sleep (exec) vs. computation (rs + proc + ac).
  │     └─► Targeted action: If exec > 95%, resolve host CPU preemption, E-core scheduling, or CFS quota throttling.
  │
  ├─► [Symptom: Gazebo simulation ignores world edits or live robot poses freeze]
  │     ├─► Precondition: Behavior Tree executed in simulation preview or multi-world desynchronized.
  │     ├─► Diagnostic check: Run inctl world reset --address=localhost:17080 or inspect world updater paused state.
  │     └─► Targeted action: Reset worlds via inctl world reset or supply start_from_world_state=worlds.EditWorldId.BELIEF.
  │
  ├─► [Symptom: failed to upload image: failed to dial "/run/containerd/containerd.sock": connect: connection refused]
  │     ├─► Precondition: inctl asset install or sideloading attempted across localhost:17127 / localhost:17080.
  │     ├─► Diagnostic check: Confirm failure is on backend k3s host daemon, NOT in local client environment.
  │     └─► Targeted action: Stop after at most 2 attempts. Never debug local /run or proxy sockets; verify hermetically via bazel build and bazel test.
  │
  └─► [Symptom: grpc: received message larger than max (X vs. 4194304)]
        ├─► Precondition: Mesh, point cloud, or complex BehaviorTree transferred across default channel.
        ├─► Diagnostic check: Check transfer payload size against default 4 MB ceiling.
        └─► Targeted action: Configure channel options with [("grpc.max_receive_message_length", -1)].

© intrinsic-ai, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 9 other files (references) in .agents/skills/intrinsic-core-debugging of intrinsic-ai/intrinsic-core.

• SKILL.md
• references/assets-and-solutions.md
• references/bazel.md
• references/executive-and-behavior-trees.md
• references/geometry-and-math.md
• references/longrunning-operations.md
• references/motion-planning-and-icon.md
• references/perception-and-vision.md
• references/platform-logging-and-status.md
• references/world-and-kinematics.md

Open the folder on GitHubat commit 3fd7856