mirror of https://github.com/SHOGGOTH-SECTOR/sica-fondt.git synced 2026-07-31 16:16:26 +00:00

Reorg step 1: consolidate repo under core/, drop the mafiabot label

Mechanical structure-only pass (no document/guide content edits):

- Move src/ichor, src/endocrine, and docs/ into core/; the old top-level
  mafiabot_core/ becomes core/. Everything now lives under a single core/ root.
- Drop the "mafiabot" label from the Ada/Alire crate: core.gpr (project Core),
  alire.toml name = "core"; the generated *_config.* regenerate as core_config.*.
- Repoint build-critical wiring only:
  - .claude/skills/run-sica-fondt/smoke.sh (ponyc + Ada + COBOL paths)
  - core/src/endocrine R source()/runner paths and the test glob
  - .github/workflows/ci.yml (working-directory + gpr name)

Verified green: smoke ALL GREEN (Pony Ichor, Ada core crate + tests, COBOL E1
vault); R endocrine 14/0; Octave ETR 26/0/1.

Deferred (reserved for a less-ephemeral doc/guide pass): docmap.yaml, the guide
files and nested AGENTS.md/README prose + now-stale relative links, SOUL.md, and
the deeper ring/storage restructure.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_015hmgREHNsxYCuim33yUF2c

2026-06-23 07:46:47 +00:00

2.3 KiB

Raw Blame History

F2 — SAE monitor (DESIGN-FIRST)

1. Component

The interpretability monitor — a sparse-autoencoder watcher pointed at the machinery (subagents), never the homonculus. The outward-facing half of the central cut ("trust the center, verify outward"). Also the detector that fires the G1 stress loop.

2. Status / certainty

DESIGN-FIRST · ABSENT. Topology C5 (watches subagents, never Brain); detection mechanism C1/C2.

3. Language & location

TBD · new location e.g. src/sae/. ML interpretability (SAE over subagent activations).

4. Does / does-not

Does: watch subagents (F3) + BERT/LoRA (they're AI systems); detect agent outputs in opposition to the top Eth-Int convictions (A7), via the convictions' semantic-isomorphy tags → fire G1.
Does-not: watch the Brain/scratchpad (forbidden by the central cut); correct cognition (it detects, it doesn't edit — elimination breeds obfuscation, per self-doc B2).

5. Interface contract (proposed)

watch(subagent_activations) -> findings (structure verification).
detect_opposition(outputs, top_convictions A7) -> opposition_signal → handed to G1 (stress-loop).
Hard boundary: no hook into the Brain (C4) — structurally impossible by construction.

6. Dependencies & stubs

A7 top convictions (with isomorphy tags) — stub: fixed conviction set. F3 subagents — stub: canned activations. G1 consumes its signal — stub: print the signal.

7. Invariants / laws

L1 (C5): SAE points at the machinery, never the homonculus (the Brain is the one thing nothing audits).
L2 (C4): detection only — no closed elimination loop on cognition (judge the fruits at conduct, not the mind).
L3 (C2): opposition = output semantically isomorphic to a top conviction's antithesis (A7 tags).

8. Build steps

Fix the no-Brain-hook boundary structurally. 2. Build subagent activation watching. 3. Build the conviction-opposition detector (needs A7 isomorphy tags). 4. Wire the G1 stress emission.

9. Tests

Structural: no Brain hook exists. Opposition-detection: an output matching a top conviction's antithesis fires; aligned output doesn't.

10. Open items

SAE training/target (C1). The isomorphy match (output ↔ conviction tag) mechanism (C2, shared A7/G1).

2.3 KiB Raw Blame History