​

Skill metadata

​

Reference: full SKILL.md

​

Actual Computer Setup Skill

​

When to Use

• User wants to add actual.inc as an inference provider (cloud relay or local).
• User has an ac_ key and wants Mibyan routed through their Actual cluster.
• User wants fully-local, on-device inference via the Actual daemon.
• Troubleshooting: Actual requests failing with cryptic 400s or empty streams.

​

Prerequisites

• Mibyan has first-class actual provider support (provider id actual, aliases actual-computer, actualcomputer, aci). Do NOT configure Actual as a custom_providers / providers.actual.* entry on current Mibyan — the built-in provider owns the name and handles base-url normalization, the Responses transport, and local no-auth automatically.
• Relay mode: an Actual account and an ac_ inference key from https://actual.inc/user/keys.
• Local mode: the user has installed the daemon (curl -fsSL "https://actual.inc/install" | bash) and completed device authorization by running actual once and opening the printed https://actual.inc/device?code=... URL in a browser. Relay that URL to the user and WAIT — never invent an email or authorize on their behalf. Codes expire in 5 minutes; re-run actual for a fresh one.

​

How to Run

​

Relay / API mode

1. Put the key in .env (secrets only — never config.yaml): append ACTUAL_API_KEY=ac_... to ~/.mibyan/.env.
2. Verify the key and discover models with terminal:
   curl -s https://api.actual.inc/v1/models -H "Authorization: Bearer $ACTUAL_API_KEY"
   
3. Select provider + model:
   mibyan config set model.provider actual
   mibyan config set model.default "MODEL_ID_FROM_DISCOVERY"
   
4. Verify end-to-end:
   mibyan chat -Q -q "Reply with exactly: ACTUAL_OK" --provider actual -m MODEL_ID
   

​

Local mode

1. Human has installed + authorized the daemon (see Prerequisites).
2. Download and load a model (scriptable once authorized):
   actual models search "qwen2.5 0.5b instruct gguf" --limit 8 --no-prompt
   # Downloads REQUIRE an explicit quantization (409 ambiguous_model_download otherwise):
   actual models download "Qwen/Qwen2.5-0.5B-Instruct-GGUF/Q4_K_M"
   actual models list        # note the INSTALLED name (differs from download id)
   actual models load "qwen2.5-0.5b-instruct-q4_k_m"   # load by installed name
   
3. Point Mibyan at the daemon. ACTUAL_BASE_URL with a loopback host flips the built-in provider into local no-auth mode automatically — no key needed: append ACTUAL_BASE_URL=http://127.0.0.1:8080 to ~/.mibyan/.env, then:
   mibyan config set model.provider actual
   mibyan config set model.default "INSTALLED_MODEL_NAME"
   
4. Verify (reduced toolset — see context-window pitfall below):
   mibyan chat -Q -q "Reply with exactly: LOCAL_OK" --provider actual -m INSTALLED_NAME -t file,web
   

​

Quick Reference

​

Pitfalls

1. reasoning_effort trap (handled by Mibyan since the first-class provider). Actual’s SGLang/vLLM backends accept only none/low/medium/high/max; xhigh/ultra used to fail with a cryptic Expecting value: line 1 column 1 (char 0) (a wrapped HTTP 400). The built-in provider clamps xhigh→high and ultra→max on the wire. If a request still 400s this way on an old Mibyan, set a per-model cap: agent.reasoning_overrides.<model>: high in config.yaml.
2. Context-window overflow on small local models. Mibyan’ default toolset is ~26k tokens of schemas plus a ~9k-token system prompt. A model loaded with a 32k context overflows before the first turn, and llama.cpp-family servers emit a bare data: [DONE] — Mibyan reports Provider returned an empty stream with no finish_reason. This is NOT an SSE bug. Fixes: restrict tools (-t file,web), load the model with a larger n_ctx, or pick a >=64k-context model for the full toolset. Upstream tracking: #51448 (do not file new issues; add evidence there). Related but distinct: #65631 (HTTP-200 SSE carrying a 400), #56516 (reasoning-only streams).
3. Download ids vs installed names. actual models download takes repo/QUANT and 409s without an explicit quantization; actual models load takes the INSTALLED name from actual models list.
4. Reasoning models returning empty content. GLM/Qwen reasoning variants emit thinking in a separate reasoning field and can burn a small max_tokens entirely on reasoning. Give generous max_tokens before assuming failure.
5. Do not create a custom provider named actual. Older setup guides (pre first-class support) wrote providers.actual.* config blocks. On current Mibyan the built-in provider wins the name; stale custom blocks are ignored or conflict. Remove them and use the env vars + model.provider flow above.

​

Verification

# Relay:
mibyan chat -Q -q "Reply with exactly: ACTUAL_OK" --provider actual -m MODEL
# Local (small model — reduced toolset):
mibyan chat -Q -q "Reply with exactly: LOCAL_OK" --provider actual -m MODEL -t file,web
# Provider status (local no-auth shows key_source=local-offline):
mibyan status