soul: let Neuron truthfully answer 'what model am I running on?' #66

2026-07-01T00:13:35Z

tim.lingo commented

2026-07-01 00:13:35 +00:00

Neuron currently can't say which LLM it's running on — because an LLM can't know its own model from training (the name/version is assigned after training finishes; see the research on LLM identity confusion). The model IS known to the system (per-request + shown in the app's model chip) but was never put into the LLM's own context.

Fix (additive): append a factual [CURRENT ENGINE: ] line to the system prompt. The per-turn model comes from the request body, so it's accurate even under Auto routing; falls back to the configured default. Identity-consistent — the model is the engine, the self (identity/values/memory) is layered on top. Does NOT touch identity/values/safety.

Rung: PARSES (elc --target=c chat.el exits 0). NOT built/tested — ships with the soul rebuild, same as the OpenAI-provider PR (#65).

Generated with Claude Code

Neuron currently can't say which LLM it's running on — because an LLM can't know its own model from training (the name/version is assigned after training finishes; see the research on LLM identity confusion). The model IS known to the system (per-request + shown in the app's model chip) but was never put into the LLM's own context. Fix (additive): append a factual [CURRENT ENGINE: <model>] line to the system prompt. The per-turn model comes from the request body, so it's accurate even under Auto routing; falls back to the configured default. Identity-consistent — the model is the engine, the self (identity/values/memory) is layered on top. Does NOT touch identity/values/safety. Rung: PARSES (elc --target=c chat.el exits 0). NOT built/tested — ships with the soul rebuild, same as the OpenAI-provider PR (#65). Generated with Claude Code

tim.lingo added 1 commit 2026-07-01 00:13:39 +00:00

soul: let Neuron answer 'what model am I running on?' — inject current engine into system prompt

Neuron Soul CI / build (pull_request) Failing after 10m43s

Details

Neuron Soul CI / deploy (pull_request) Has been skipped

Details

b24f6d645b

Additive: appends a factual [CURRENT ENGINE: <model>] line to the system prompt (model from the
request body — accurate even under Auto routing; falls back to configured default). An LLM can't
know its own model from training (name/version assigned post-training), so the harness must tell it.
Identity-consistent: model = engine, self layered on top. Does NOT alter identity/values/safety.
PARSES (elc chat.el exit 0); NOT built/tested — ships with the soul rebuild.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

will.anderson referenced this issue from a commit

2026-07-01 16:34:43 +00:00

Merge PR #66: soul: inject current engine into system prompt for truthful self-report

will.anderson referenced this issue from a commit

2026-07-01 16:43:09 +00:00

chore(dist): update soul.c with PR #63/#65/#66 + Task 1 chat.el changes

will.anderson manually merged commit d67f4c8f08 into main

2026-07-13 16:21:11 +00:00

will.anderson referenced this pull request

2026-07-22 22:00:25 +00:00

Soul: make inference endpoint + auth token configurable (unblocks hosted Pro/Max/trial) #62

Sign in to join this conversation.