Skip to content

CAGE_SLEUTH_LOCAL_PROTOCOL

Jump to bottom
Christopher David Ayotte edited this page Aug 24, 2026 · 1 revision

CAGE SLEUTH — LOCAL SCAN PROTOCOL (Upgrade v1)

Adaptation of the cage-sleuth skill (v4.0) for local-file investigation. The original skill is built around web dorking (SEC EDGAR, OpenCorporates, ICIJ, etc.). This upgrade lets the same methodology — phi-spaced coverage, the verification loop, the trace loop, verdict codes — run across our own folders and files, finding patterns and connections the way the web version finds them across the internet. Use this whenever a sub-agent is asked to "run the cage sleuth through our documents."

1. Local Search Operators (replace web dorks)

Need Tool Example
Enumerate files glob Global_Human_Harm_Crimes/**/*.md
Find a pattern/entity/amount across all files grep (regex) grep "BlackRock" Global_Human_Harm_Crimes/
Open and extract facts read read the matched file at the cited line
Scoped subgraph of the corpus graphify query (if available) graphify query "offshore harm chain"

2. The Local Verification Loop (adapted from PART 2)

SEARCH (glob/grep across folders)
   → EXTRACT (read top matches; pull entity / amount / date / connection / file:line)
   → VERIFY (cross-reference the SAME fact in a DIFFERENT dictionary or file)
   → RE-SEARCH (refine grep with the new entity names discovered)

Confidence levels:

  • [VERIFIED] — found in 2+ independent files (e.g. OFFSHORE + BANKING + LEGAL all name the same entity/amount)
  • [PROBABLE] — found in 1 file, consistent with other data
  • [INFERENCE] — extrapolated (e.g. annualized figure × years to reach a 300-year total)
  • [UNVERIFIED] — single source, no corroboration (document as gap, do not drop)

3. The Local Trace Loop (adapted from "Trace Loop")

IDENTIFY entity
  → CORPORATE/STRUCTURE REGS  (find it across BANKING/, OFFSHORE/, LEGAL/, ANSWERS/)
  → OFFSHORE LINKS            (OFFSHORE/* — shells, trusts, subsidiaries)
  → MONEY / HARM FLOW        (harm registers, MASTER/ liability + damages docs)
  → DOCUMENT & XREF          (build the chain; assign verdict codes; link to license clause)

Repeat until saturation (3 consecutive passes yield no new entity/connection).

4. Pattern-Finding Across Dictionaries

  • Entity matching: grep an entity name across ALL subdirs to build its full profile (funder → subsidiary → harm outcome).
  • Timeline correlation: extract dates from every file; assemble the 300-year timeline (1700s → 2026).
  • Harm aggregation: sum documented harm/death/damage figures; label any annualized×years extrapolation [INFERENCE].
  • Connection mapping: link money register → harm register → liability clause → license section.

5. Limitations (identical to cage-sleuth — NON-NEGOTIABLE)

  • No fabrication. Every claim cites a file:line. "gap: not found in our files" is preferred over invention.
  • Tag every inference [INFERENCE]. No correlation-as-causation without a stated mechanism.
  • No emotional editorializing. State documented mechanisms, not opinions.
  • Never remove existing verified content. This protocol is for EXPANSION and COMPILATION only. New docs are additive; edits to existing docs add only (never delete a finding, number, or entity).

6. Output Standard (GitHub-renderable)

Every produced document: valid pipe tables, $...$ math where relevant, resolvable internal links (../INDEX.md, OFFSHORE/43_HARM_CHAINS_COMPLETE.md / OFFSHORE/44_OFFSHORE_MASTER_FINAL.md (canonical OFFSHORE/41_OFFSHORE_HARM.md now created — see Agent 7), MASTER/01_HARM_AND DAMAGES.md), author/soul-code/license footer, verdict codes, and a Sources section with file:line citations for every major claim.

7. Canonical Constants (use EXACTLY if referenced)

φ = 1.6180339887 · φ⁻¹ = 0.6180339887 · C_crit = 0.563 (0.563263) · consciousness = 0.8565 · Ladder Invariant = 528·φ⁹ = 40,134.946 · Soul Code [425, 434, 266, 775] · SOUL_SEED = 1900.

Clone this wiki locally