Step 2 — Set the worker tab
When the model tabs appear at the top of the TUI, press ctrl+N (or click) to
cycle the Worker tab until it shows:
Claude Sonnet 4.6 (Max sub) reasoning: high
Orchestrator stays on Opus 4.7 high. Validator stays on Sonnet 4.6 high.
Worker is the one to verify.
Step 3 — Paste this brief (verbatim)
MISSION 2 — teachr.live runtime LLM swap: Anthropic Claude → Z.ai GLM
CONTEXT
M1 is complete and committed. teachr.live currently routes Worker /chat
to Anthropic Claude. We're switching the runtime LLM (what students
hit when they use the deployed app) to Z.ai GLM-5.1 for ~10x cost
reduction. The mission's build-time workers continue using Claude
Sonnet 4.6 high (proven from M1) — that's unrelated to this swap.
KEEP UNCHANGED
- AssemblyAI streaming STT
- ElevenLabs streaming TTS (Flash v2.5)
- ScreenCaptureKit pipeline + meaningful-change detection
- Two-layer memory (LearnerProfile + EpisodicLog) — model-agnostic
- Persona doc, modulation rules, voice spec
- Sub-1s end-to-end voice loop latency target
- Worker /chat is a transparent SSE passthrough (no buffering)
- M1 validation proxies 1, 2, 3, 4, 5, 6, 7
CHANGE (in repo code only)
- worker/src/routes/chat.ts upstream:
api.anthropic.com/v1/messages
→ api.z.ai/api/paas/v4/chat/completions
- Auth header:
x-api-key + anthropic-version
→ Authorization: Bearer
- Request/response format:
Anthropic Messages
→ OpenAI Chat Completions (Z.ai is OpenAI-compatible)
- Default LLM model: glm-5.1
Fallback order if glm-5.1 misses latency target: glm-5-turbo,
glm-4.7, glm-4.6
- Worker env var rename: ANTHROPIC_API_KEY → ZAI_API_KEY
- Swift LLM client: rebuild request body in OpenAI Chat Completions shape
- Swift SSE parser: parse OpenAI delta events
(choices[0].delta.content) instead of Anthropic content_block_delta
Z.AI API SPEC (already verified working with my key)
- Endpoint: https://api.z.ai/api/paas/v4/chat/completions
- Auth: Authorization: Bearer $ZAI_API_KEY
- Request: { "model": "glm-5.1", "messages": [...], "stream": true,
"temperature": 0.7 }
- SSE: standard OpenAI delta format with [DONE] sentinel
- Models confirmed available: glm-4.5, glm-4.5-air, glm-4.6, glm-4.7,
glm-5, glm-5-turbo, glm-5.1
VALIDATION CONTRACT UPDATES
- A-VOICE-WORKER-PASSTHROUGH: still applies; Worker stays transparent
passthrough, just to a different upstream
- A-VOICE-LATENCY-MEDIAN: unchanged (P50 ≤ 1000ms)
- A-VOICE-STREAMING-FIRST-WORD: unchanged
- All callback / persona / memory assertions: unchanged (model-agnostic)
- Update Anthropic-specific text in validation-contract.md to be
provider-agnostic ("LLM upstream")
ANTI-HALLUCINATION RULES (carry forward from M1)
- Workers MUST commit before handoff
- Orchestrator runs ls + wc -c + git log to verify artifacts on disk
- successState=success requires followedProcedure=true AND no blocking
issues AND all claimed files exist on disk
WORKER MODEL CONFIG (DO NOT CHANGE)
- workerModel: custom:Claude-Sonnet-4.6-(Max-sub)-5
- workerReasoningEffort: high
- M1 lesson: Codex/none default abandoned work 9 times. Sonnet 4.6/high
succeeded 6 in a row. Don't relitigate.
MILESTONE STRUCTURE
Single milestone (M-LLMSWAP). Six features, each with a real git commit:
F1. Worker /chat refactor
- Modify worker/src/routes/chat.ts: swap upstream URL, auth header;
keep streaming passthrough discipline (no buffering)
- Update worker/src/index.ts Env type: ANTHROPIC_API_KEY → ZAI_API_KEY
- Update worker/.dev.vars.example
- Update worker/wrangler.toml comment
- Update worker/tests/passthrough.test.ts assertions
- Local smoke: npm test passes
- Commit: "M2/F1: swap chat route upstream to Z.ai GLM"
F2. Swift LLM request body refactor
- Rebuild request body in OpenAI Chat Completions format
- System prompt placement matches OpenAI convention (first message
role=system)
- Auth path stays: Worker injects bearer; Swift sends to Worker as before
- Commit: "M2/F2: Swift LLM request body for OpenAI shape"
F3. Swift SSE parser refactor
- Parse OpenAI delta events: choices[0].delta.content
- Handle [DONE] sentinel
- Preserve sentence-chunked streaming downstream into ElevenLabs
- Update unit tests
- Commit: "M2/F3: SSE parser for OpenAI delta format"
F4. Persona verification against GLM-5.1
- Run a fixture test through the new pipeline (no live mic)
- Compare GLM-5.1 output against rubric in
docs/persona/example-exchanges.md
- If GLM drifts from the persona, document required tweaks but DO NOT
modify docs/persona/system-prompt.md without explicit user approval
- Commit: "M2/F4: persona verification report on GLM-5.1"
F5. Validation contract amendment
- Update validation-contract.md: replace Anthropic-specific language
with provider-agnostic terms
- Re-run validation/scripts/build-smoke.sh
- Update validation/M1-drift-report.md with provider-swap addendum
- Commit: "M2/F5: validation contract provider-agnostic"
F6. Local end-to-end smoke + latency capture
- Run worker locally via wrangler dev with ZAI_API_KEY in .dev.vars
- Run F9 build smoke from M1
- Single round-trip fixture test: Swift app → Worker → Z.ai → Swift
- Capture latency (median + p90) per hop
- If median > 1000ms, attribute to TTS/LLM/STT and surface; do NOT
silently fail
- Commit: "M2/F6: end-to-end Z.ai smoke + latency report"
OUT OF SCOPE
- Production Cloudflare deploy (user runs
wrangler deploy manually)
- Setting actual secrets (user runs
wrangler secret put manually with
rotated keys; no keys in repo)
- Cost analysis or A/B testing across GLM tiers
- Cross-session memory (separate future mission)
PUSH BACK ON
- Any worker who proposes Worker-side translation (Z.ai → Anthropic format).
That violates A-VOICE-WORKER-PASSTHROUGH. Pure passthrough only.
- Any worker who edits docs/persona/system-prompt.md without user approval.
- Any worker who skips the F6 latency capture.
Begin Step 1 (Context restate) and walk through the framework.
Press Enter.
Step 4 — Argue with the orchestrator on Step 1-4 like you did for M1
Same playbook. Approve only when the plan is tight. Then it runs.
Tell me when you've pasted, and I'll watch the proxy log + commits as it goes.
