canic-academic

name	canic-academic
description	Use when working with Canic fleets, local academic ICP CLI targets, sharding, canic install or upgrade flows, canister ID shell helpers, canic info list, canic medic, canic metrics, or raw icp commands inside a Canic-managed downstream project.

Canic Academic

Use this skill before changing a Canic-managed downstream fleet or debugging a local academic deployment.

First Checks

Run these before editing topology or endpoint code:

canic status
canic fleet config <fleet> --verbose
canic --network academic info list <deployment> --verbose
canic --network academic info env <deployment>
canic --network academic medic <deployment>

Use fleet config for configured intent and info list for deployed registry truth. If they disagree, diagnose the deploy/install state before changing app logic.

ICP CLI Rules

• For Canic commands, pass the target explicitly: canic --network academic ....
• For raw ICP CLI commands in this workflow, clear stale shell network state: env -u ICP_NETWORK icp ... -e academic.
• Do not rely on exported ambient network variables in scripts that also call Canic.
• Do not use $ROOT for a root canister principal. Use CANIC_ROOT and role-scoped names such as CANIC_USER_HUB and CANIC_USER_SHARD.
• Generate scripts/canister_ids.sh from deployed state with mkdir -p scripts and canic --network academic info env <deployment> > scripts/canister_ids.sh. Source it after install or reinstall and treat it as the canonical helper.

Sourced Shell Helpers

Do not put set -e in scripts intended to be sourced into an interactive shell. Use functions that return status instead. Executable scripts may still use strict shell options.

Install Or Upgrade

• canic install <fleet> is for fresh local creation or recreating local state after the ICP CLI replica lost canisters.
• If the canister exists and only Wasm changed, treat it as an upgrade flow. Check canic info list and canic medic before and after.
• If canic install is blocked on existing state, decide between reinstall, raw ICP upgrade, or deployment registration repair. Do not keep retrying the same install blindly.

Sharded Calls

For protected Canic internal endpoints, call through generated endpoint descriptors and pass the caller role explicitly:

use canic::__internal::core::api::ic::canic::CanicInternalClient;
use canic::ids::CanisterRole;

let response: MyResponse = CanicInternalClient::new(shard_pid)
    .call_update_result(
        &project_shard_endpoint(),
        CanisterRole::new("user_hub"),
        (tenant_id, request),
    )
    .await?;

Use tuple Candid arguments for multi-argument methods. Scripts and external tests should call public, non-internal endpoints instead of protected internal methods.

Metrics

When canic_metrics is missing or a tier is empty, check:

canic fleet config <fleet> --verbose
canic --network academic info list <deployment> --verbose
canic --network academic metrics <deployment> --kind core --nonzero

Likely causes are a profile that does not enable the tier, stale deployed Wasm, or a rebuild/reinstall that did not actually happen.